09:00 - 11:00
Arino, Olivier - ESA
Sentinel-1 Mission Status
Potin, Pierre - ESA
Sentinel-2 Mission Status
Gascon, Ferran - ESA
Sentinel-3 Mission Status
Gascon, Ferran - ESA
Detection of water bodies with multi-temporal Sentinel-1 SAR observations: examples from West Africa and Greenland
Santoro, Maurizio (1); Cartus, Oliver (1); Wiesmann, Andreas (1); Wegmüller, Urs (1); Kellndorfer, Josef (2); Defourny, Pierre (3); Arino, Olivier (4) - 1: Gamma Remote Sensing, Switzerland; 2: Earth Big Data LLC, USA; 3: Université Catholique de Louvain, Belgium; 4: ESA ESRIN, Italy
Large-scale mapping and monitoring of water bodies is a relevant topic of investigation in the remote sensing community thanks to the increasing demand for water-related map products by the scientific community. The public availability of extensive datasets of remote sensing observations furthermore foster the development of algorithms to detect water bodies. Optical datasets are the major source of information in such approaches thanks to multi-decadal observations and well-established detection techniques (Pekel et al., 2016; Carroll and Loboda, 2017). Yet, water body detections based on optical data suffer from incorrect detections if the spectral signatures differ in time and are affected by missed water detections in case of persistent cloud cover (e.g., in the tropics). In comparison, the data pool of SAR observations is sparser both in space and time, except for coarse resolution observations, which, however, could be exploited to generate a few data layers with information on water extent (Santoro et al., 2015; Lamarche et al., 2017) and water occurrence (Santoro et al, unpublished). It was only recently with the routine operations of Sentinel-1A and -1B (2016) that repeated high-resolution observations at continental scale started being collected. Such datasets foster new research towards the development of water body detection approaches with SAR data and allow an assessment of the true potential of active microwave observations in a water mapping and monitoring scheme to be applied at large scales. The latter aspect is supported by the increased availability of grid and cloud computing facilities responding to the need of the larger throughput of data (e.g., Earth Big Data’s Cloud Processing system).
In this contribution we present recent findings in terms of SAR signatures of water bodies and land from multi-year Sentinel-1A and -1B acquisitions over arid and polar landscapes. In past studies with SAR data (Santoro et al., 2015), these were flagged as particularly critical because of similar backscatter values over water, bare soil and permanent snow/ice causing significant water commission errors. We then present our current approach to map water bodies with multi-temporal Sentinel-1 observations and discuss large-scale map products. Support Vector Machines have been applied to seasonal metrics of the Sentinel-1 backscatter over West Africa to generate a map of permanent water bodies. The overall accuracy was above 90% with omissions in correspondence of shorelines due to the predominant effect of land features in mixed water-land pixels. Yet, the major highlight of our water body map is the full characterization of the water system of West Africa, which indeed is not the case in publically available datasets based on optical data for the same time frame. For the conference, it is foreseen that the research is extended to map marginal lakes and supra-glacial lakes of Greenland so to provide a more comprehensive overview on Sentinel-1 capabilities to map and monitor water bodies.
Global water surface dynamics: toward a near real-time monitoring using Landsat and Sentinel data.
Pekel, Jean-Francois (1); Belward, Alan (1); Gorelick, Noel (2); De Felice, Luca (1) - 1: European Commission - JRC, Italy; 2: Google Earth Outreach
Global surface water dynamics and its long-term changes have been documented at 30m spatial resolution using the entire multi-temporal orthorectified Landsat 5, 7 and 8 archive for the years 1984 to 2015. This validated dataset recorded the months and years when water was present, where occurrence changed and what form changes took (in terms of seasonality), documents inter-annual variability, and multi-annual trends. This information is freely available from the global surface water explorer https://global-surface-water.appspot.com. Here we extend this work by combining post-2015 Landsat 7 and 8 data with imagery from the Copernicus program's Sentinel 1 and 2 satellites. Using these data in combination improves the spatial and temporal resolution. The improved geographic and temporal completeness of the combined Landsat / Sentinel dataset also offers new opportunities for the identification and characterization of seasonally occurring waterbodies. These improvements are also being examined in the light of reporting progress against Agenda 2030's Sustainable Development Goal 6, especially the indicator used to measure 'change in the extent of water-related ecosystems over time'.
Advancing the use of Earth Observations for the monitoring of water-relating ecosystems in the context of the Sustainable Development Goals
Tchadie, Alain Michel
Crane, Stuart; Campbell, Jillian; Bernhardt Elisabeth, Harlin, Joakim; Tchadie, Alain Michel; Midha, Nisha - UN Environment, Nairobi, Kenya
The UN-Water Integrated Monitoring Initiative for Sustainable Development Goal (SDG) 6, otherwise known as GEMI, aims to support countries in monitoring water and sanitation. UN Environment, is responsible for SDG Indicator 6.6.1 (change in the extent of water-related ecosystems over time) which requires assisting national policy- and decision makers on measuring and collecting data on their water-related ecosystems. The first round of data collection conducted in 2017 revealed that it is extremely difficult for countries to collect and report data required for Indicator 6.6.1. Country capacity is often lacking and Indicator 6.6.1 itself is complex and multi-faceted. The complexity of the indicator is also indicative of its true value, which is to inform countries how a waterbody is changing over time, so that countries can then act to protect and/or restore that waterbody and freshwater ecosystem. Therefore, UN Environment has initiated collaboration with some members of the Global Earth Observation community such as the National Aeronautics and Space Administration (NASA), the European Space Agency (ESA), the Joint Research Centre (JRC) and Google Earth Engine, to compliment this lack of data using satellite imagery. Satellite imagery has been demonstrated to accurately estimate the spatial extent of freshwater ecosystems as well as certain water quality parameters of freshwater ecosystems. There is also increasing evidence that the spatial extent of vegetated wetlands can be generated by earth observations. While this data provision is promising, one of the principal challenges faced is national ownership and validation of geospatially-generated data. This collaborative approach to monitor and report on Indicator 6.6.1 and Target 6.6 is a worthy opportunity to enhance national progress on protecting and restoring water-related ecosystems over the next 12 years until 2030.
Key words: Freshwater ecosystems, SDG, global monitoring, Satellite Imagery.
11:20 - 13:00
A Long-Term, Dynamical, High-Spatial Resolution Inundation Extent Dataset at Global Scale, from the Combination of Multiple Satellite Datasets
Aires, Filipe (1,2); Prigent, Catherine (1,2); Lehner, Bernhard (3); Yamazaki, Dai (4); Fluet-Chouinard, Etienne (5); Pekel, Jean-Francois (6); Bousquet, Philippe (7); Jimenez, Carlos (2) - 1: LERMA, France; 2: Estellus, France; 3: McGill University; 4: Univ. of Tokyo, Japan; 5: University of Wisconsin-Madison, USA; 6: Europe Commission – Joint Research Centre; 7: LSCE/CEA, France
Several satellite observations can monitor surface water inundation over the globe: Visible/infrared observations (e.g. MODIS/Landsat), active (e.g. SAR) and passive microwave. Each one of these observations has its own advantages and drawbacks: for instance, passive microwave has a low spatial resolution while visible observations cannot detect water below clouds or vegetation. We will review several of the techniques currently used.
The Global Inundation Extent from Multi-Satellites (GIEMS) database is derived from multiple satellite observations (visible, active and passive microwave). It provides multi-year monthly variations of the global surface water extent at about 25kmx25km resolution, from 1993 to 2007. Work is currently being conducted to obtain measurements at 10-day intervals and extend the time-series to ~30 years until present. GIEMS spatial resolution is usually compatible with climate and global land surface model outputs (for instance for methane emission) but is clearly not adequate for local applications that require the characterization of small individual water bodies. A downscaling approach for GIEMS was developed by Aires et al. (2017) relying on a floodability index generated from topographic data. The resulting downscaled GIEMS-D3 database possesses the same attributes as GIEMS, but with a high spatial resolution (90m) allowing to spatially delineate waterbodies and inundated areas globally.
GIEMS-D3 is assessed by analysing its spatial and temporal variability over the Niger, Amazon, Ganges-Brahmaputra and Mekong basins. GIEMS-D3 is also compared with other independent satellite databases such as GSWO (Landsat), G3WBM (Landsat) and GLWD (Lehner & Döll, 2004). In contrast to visible-based datasets, GIEMS-D3 is capable of representing inundations beneath dense cloud and vegetation cover, but it also suffers from topography limitations (Aires et al. 2018). As a result, large differences can be observed between GIEMS-D3 and the other datasets, particularly over the tropics and high latitudes.
An incipient international project aims to synthesize various global inundation datasets to provide the scientific community with a single harmonized and comprehensive surface water cover data product. The effort focuses on combining GIEMS-D3, the Landsat-derived GSWO (Pekel et al. 2017), the HydroLAKES (Messager et al. 2016), and the MERIT DEM topography (Yamazaki et al. 2017) datasets.
Service cases and service lines of SWOS (Satellite based Wetland Observation Service) for wetland monitoring from local to global level
Weise, Kathrin (1); Hoefer, Rene (1); Schwarz, Michael (1); van Valkengoed, Eric (2); Franke, Jonas (3); Thulin, Susanne (4); Eberle, Jonas (5); Truckenbrodt, John (5); Zander, Franziska (5); Abdul Malak, Dania (6); Sanchez, Antonia (6); Schroeder, Christoph (6); Strauch, Adrian (7); Muro, Javier (7); Firoka, Eleni (8); Guelmami, Anis (9); Mino, Eric (10); Flink, Stephan (11); Hilarides, Lammert (11); Plasmeijer, Anouska (12); Ling, Matthew (13); o'Conner, Brian (13) - 1: Jena-Optronik GmbH, Germany; 2: TerraSphere BV, Netherlands; 3: RSS - Remote Sensing Solutions GmbH, Germany; 4: Brockmann Geomatics Sweden AB, Sweden; 5: Friedrich Schiller University of Jena , Germany; 6: University of Malaga, Spain; 7: University of Bonn, Germany; 8: Greek Biotope Wetland Centre (EKBY), Greece; 9: Mediterranean Wetlands Observatory Tour du Valat , France; 10: SEMIDE / EMWIS, France; 11: Wetlands International, Netherlands; 12: European Regional Office IUCN (International Union for Conservation of Nature), Belgium; 13: UN Environment World Conservation Monitoring Centre UNEP-WCMC, Great Britain
Wetlands are the most fragile and threatened ecosystems and they are one of the fastest declining ecosystem types worldwide. About 64% of wetlands are lost since 1900 and 76% of freshwater plants and animals disappeared between 1970 and 2010 (source: Ramsar factsheet 3.2).
At the same time wetlands are hot spots of biodiversity and provide diverse and valuable ecosystem services; such as Water supply & purification, disaster reduction or Climate change mitigation and Carbon sequestration.
The pressure on wetlands is likely to intensify in the coming decades due to increased global demand for land and water and due to climate change. Decision-makers can help slow, stop and reverse that process and stakeholders in all levels of governance have to be involved to make a change. But information on wetlands extent, their ecological character and their services is often scattered, underestimated and difficult to find and access.
In this respect, SWOS (Horizon-2020 project funded by the European Union) provides monitoring tools and information on wetland ecosystems, derived from Earth Observation data.
SWOS is assisting reporting and monitoring obligations for environmental policies at different scales (from local to global) and contributes to an improved integration of wetlands in policy. SWOS is working in very close cooperation with the Ramsar Convention on Wetlands, the Group on Earth Observations (GEO) and other international partners and it complements the MAES process under the EU Biodiversity Strategy to 2020 by supporting the maintenance and restoration of wetland ecosystems and their services. SWOS allows an information based creation of conservation and restoring measures which maintain biodiversity and essential ecosystem services.
Wetlands are very dynamic ecosystems, therefore the free available Sentinel satellite data of the Copernicus mission are an excellent basis to map wetland areas, to derive information on the ecological status and on trends in wetlands and mapping long and short term changes.
The presentation will inform about the results of SWOS, about the defined and implemented service cases and the available service lines.
Service line 1, the Map and indicator production, has been demonstrated for about 50 wetland locations in Europe, Africa and Middle East. SWOS developed standards for the map production, for nomenclatures and developed 9 wetland indicators and many sub-indicators as it can be applied e.g. for Ramsar or SDG 6.6.1 reporting.
Service line 2, the Software development, delivers the free available SWOS software toolbox GEOclassifier, which provides all tools for map production and indicator calculation.
In Service line 3, Training/Capacity Building, the SWOS team developed a training program to teach users from different working levels how to produce new maps and indicators.
Service line 4, the SWOS / Wetland community portal, is making available all maps produced in the frame of SWOS. In addition the portal connects wetland information with free available European and global layers that are useful for wetland monitoring.
Wetlands give us ecosystem services for free! SWOS delivers infrastructure for monitoring and provides a basis for decision making and actions!
A surface water body dataset with daily temporal resolution – Selected examples and application potential of the Global WaterPack
Klein, Igor; Gessner, Ursula; Hirner, Andreas; Dietz, Andreas; Dech, Stefan; Kuenzer, Claudia - DLR-DFD, Germany
Information on the distribution of inland waters bodies and their seasonal variability is important for supporting water and land management and informed decision making. Furthermore, such data can be crucial basis for scientific analyses in the field of regional and global environmental research. Key functions and services of wetlands for example are closely related to the temporal variations of surface water availability and inundation cycles. Over the past years the mapping of temporal water dynamics from earth observation data has received increasing attention. Open data policies as well as the progress in computing power and processing techniques have played a major role for this development. In the past decades there has been several optical instruments available collecting data on global scale with very high temporal resolution, such as MODIS or SPOT-VGT/Proba-V. With the launch of Sentinel-3 and Suomi-NPP such observations at global scale and high temporal resolution has been secured for the near future. Recently, many studies already underlined the relevance of mapping water with high temporal resolution presenting essential results and interesting findings. In our research we are presenting DLR’s Global WaterPack, a MODIS-based 250m time series dataset of surface water dynamics with daily temporal resolution. Using examples of lakes and reservoirs from around the Earth, the potential of the Global WaterPack to capture relevant parameters such as inundation frequency and duration, timing of flooding and water retreat, as well as freezing and thawing cycles, is presented. Results are compared with high resolution spatial reference data and in-situ measurements. The application potential of the Global WaterPack time series for monitoring and assessing changes is discussed. Furthermore, we discuss this potential based on selected examples which underline the added value of water surface detection at high temporal resolution in the context of climate and environmental change. Short and long-term dynamics of lakes and reservoirs and the underlying climatic or anthropogenic processes often cannot be determined in detail in regards to time by assessing interrupted time series or multi-temporal watermask snapshot observations. The entire MODIS time series from 06/2002-today is currently being processed and the dataset will be tested as input for global hydrological models. Additionally, quantification of uncertainty is being planned and will be part of future development.
Taking advantage of ESA’s Grid Processing On Demand to generate a European flood record based on 10 years of ENVISAT ASAR imagery
Matgen, Patrick (1); Chini, Marco (1); Hostache, Renaud (1); Pelich, Ramona (1); Zhao, Jie (1); Delgado, José Manuel (2,3); Sabatino, Giovanni (2,3) - 1: Luxembourg Institute of Science and Technology; 2: Progressive Systems Srl; 3: ESA Research and Service Support
Hydrological extremes such as floods and droughts have enormous environmental, social and economic consequences and it is expected that climate change effects combined with a growing global population will increase their impacts in the future. Central to mitigating global inundation risk are high resolution data on how floodplains and wetlands inundate in response to river discharge dynamics. However, in spite of decades of tracking floods on a global basis using a variety of spaceborne sensors, as of today, users have no access to a comprehensive, historical record of events. The ESA-funded HASARD (2012-2014) project led to the generation of different innovative Synthetic Aperture Radar-based flood hazard mapping functions implemented in software. The most notable algorithms can be used to systematically, rapidly and automatically detect, record and disseminate all observable and relevant changes of water bodies and to exploit large SAR data collections to map flood hazard. Many of these algorithms have been implemented on ESA’s grid processing on demand platform (GPOD) which opens up the possibility to process even high-resolution data sets at a large scale in a fully automatic fashion.
The objective of this study is to make use of the algorithms implemented on the GPOD infrastructure to generate a first European-scale database of flood inundation maps at 75 m pixel spacing based on 10 years of ENVISAT ASAR imagery. Within a hierarchical splitting framework, the processing chain’s underlying flood mapping algorithm uses a histogram thresholding operation and region growing process to delineate the extent of flooding. First, the probability density function (pdf) of the open water backscatter values in the SAR data is estimated. The difficulty in parameterizing such a distribution function originates from the fact that flooded areas often represent only a small fraction of a SAR scene. Hence, the distribution of a SAR scene’s backscatter values is often not clearly bimodal and it becomes difficult, if not impossible, to accurately parameterize the distribution function of backscatter values associated with surface water. We therefore developed a hierarchical split-based tiling approach (HSBA) that searches for tiles of variable size that allow parameterizing each SAR scene’s statistical distribution function. The tiling and the parameterization processes are fully integrated. The theoretical pdf of water backscatter is then used to define an optimal set of parameters for the subsequent thresholding and region growing processes that complete the flood mapping process.
The processing chain is used to automatically and efficiently generate an archive of flood extent maps derived from the collection of ENVISAT ASAR imagery over Europe. We describe the success of this effort as well as the current limitations and reflect on the refinements that are currently being implemented before (re-)processing the collection of ENVISAT ASAR and Sentinel-1 imagery in their entirety.
Industrial Mapping Of Water Bodies And Wetlands Using Algorithms Of Big Data Analytics In A Cloud-Based DIAS Environment
Lorenzo, Alberto (1); Vaitkus, Gediminas (2) - 1: Indra Sistemas SA; 2: GEOMATRIX UAB
Land Analytics EO Platform is a processing platform with analytics capabilities developed during 2017 by Indra with the support of ESA’s GSTP programme. It is designed for multi-cloud deployment and thought as a DIAS-ready tool. It is capable of producing ARD (Analysis Ready Data) and downstream products and services efficiently due to its scalability in a cloud environment. Additionally, the platform allows the ingestion of EO intermediate products in a Big Data environment, where analytics algorithms are applied in order to perform time-series analysis and extract hidden information using NO-SQL databases.
The development of our technological solution for automated processing of high resolution optical satellite images and extraction of water and wetlands layers started back in 2009-2012 when the consortium led by Indra with scientific support of GEOMATRIX UAB produced and updated EU-Hydro and EU-DEM in two projects under GMES RDA Preparatory Action and GMES Initial Operation (GIO-Land) Production of a photo-interpreted river network and water bodies and a modelled drainage network with catchments and drainage lines was derived from EU-DEM product. A continuous RTD effort took place throughout the entire period of EU-Hydro and GIO-Land, which laid a solid foundation for the second phase of EEA Core Mapping Services (CMS), where mapping of wetlands was replaced with a concept of mapping physical wetness of the environment.
After the launch of Copernicus Sentinel-1A and Sentinel-2A satellites, GEOMATRIX UAB developed an automated pre-processing solution for extraction of Water and Wetness layers from a combined time-series of pre-processed Sentinel-1 polSAR and Sentinel-2 MSI products. In 2017 our CMS Water and Wetness mapping solution evolved into a series of binary processors specifically designed for Sentinel-1/2 products. These DIAS-type processors were designed for cloud-based production environment with the following functions: (1) image pre-processing into a production-ready format (atmospheric correction, calibration, terrain-flattening, ortho-rectification, reprojection, aggregation of satellite passes, etc.) and (2) computing of biophysical parameters and extraction of individual water and wetness masks from each image.
The algorithms and binary processors for water and wetness extraction were successfully tested in Indra’s Land Analytics Platform. Production can be done on a global scale due to parallel image processing on virtual machines and capability to operate within a grid-based production framework. The use of analytical functions allows for evolving the concept of EEA CMS Water and Wetness into a taxonomy that can be used to identify wetlands in an ecological concept and to detect water bodies disregarding their cyclical changes over time. Those applications are essential in many regions of southern Europe, which are constantly suffering from surface water depletion and annual development of drought conditions.
14:20 - 16:00
Using the Sentinels for Maintaining and Updating Topographic Map Features: The Case of Lakes in the Danish Natural Environment Portal
Tottrup, Christian; Rasmussen, Mikkel; Nyborg, Lotte - DHI GRAS, Denmark
In Denmark, the Danish Natural Environment Portal provide users access to environmental data concerning for instance natural resources, soil contamination and water quality. Such data is a prerequisite for being able to monitor, manage and safeguard our natural and environmental resources in the best possible way.
Many natural habitats are protected by Section 3 of the Nature Conservation Act (NCA) including but not limited to lakes, meadows, marshes and streams. In order to enforce the NCA these habitats need to be monitored to ensure their preservation and/or rehabilitation. Habitat monitoring and assessment has been conducted through a variety of methods, including ground-based and remotely sensed data collection methods. With the advent of the Sentinels, and their enhanced capacities and a free and open data policy, there has been an increased interest in testing the applicability of developing a continuous cost and time effective monitoring tool.
The overall purpose of this presentation is to demonstrate how data from Sentinel-1 and -2 can be used to update and quality assure Lakes already present in the Danish Natural Environment Portal. The study is based on time series analyzes of both Sentinel-1 and -2 data. The presentation will emphasize the importance of implementing adequate pre-processing to ensure accurate geolocation and reduce cloud contamination. In addition, it reviews the advantages of time-series analysis and machine learning to provide statistical information on frequency of water coverage and as the basis for classifying all lakes into permanent, temporary or non-existing.
The methods have been developed and tested in a representative study area (North Jutland), and based hereof the outlook for implementing this on a national scale is provided, and with further perspectives on how this may have applicability beyond the present case e.g. for regions with poor area information (e.g. Greenland) and as support for the SDG monitoring requirements, which calls for national inventories of open water bodies under Goal 6 on ensuring access to water and sanitation for all.
Potentials of the Copernicus Program for Detection and Monitoring of Tropical Wetlands. Examples from Rwanda.
Hentze, Konrad (1); Strauch, Adrian (1); Franke, Jonas (2); Thonfeld, Frank (1); Muro, Javier (1); Steinbach, Stefanie (1) - 1: University of Bonn, Germany; 2: Remote Sensing Solutions GmbH, Germany
Challenges and progress within the scope of the DeMo-Wetlands project, that aims to develop a framework to map status of tropical wetlands in Rwanda, are discussed. In DeMo-Wetlands, straightforward and reproducible methods are developed and will be shared with the scientific community, as well as governments, NGO’s and international conventions such as the Ramsar Convention on Wetlands.
Project results are delivered on a three-leveled concept: First, potential wetlands are identified by using a morphometric analysis. Secondly, Sentinel-2 imagery is used to map existent wetlands within these areas. Finally, wetlands are characterized through a temporal inundation dataset, spatio-temporal LULC information and a dataset on land use intensity dynamics; all based on Sentinel data. These results will be validated and automatized in the future to achieve a suitable solution of wetland mapping in tropical countries.
First results of the different developed products like wetland inundation and use intensity are presented forming the basis for further development and optimization of products. In addition, the cooperation with national stakeholders will be presented, as they will be equipped with tools aiming to address their needs for the cooperation with and the reporting to international conventions.
The high resolution and repetition rate of Sentinel imagery make it perfect to provide low cost and consistent solutions for tropical countries to map changes of land use, even of highly dynamic systems such as wetlands. Rwanda serves as a national demonstrator to highlight these opportunities but also to address existing challenges such as cloud cover and a lack of adequate ground truthing methods in the context of mapping wetlands in the global south.
Development of a 16-year Surface Water Fraction Dataset from MODIS Data for the Mediterranean
Li, Linlin (1); Vrieling, Anton (1); Skidmore, Andrew (1,2); Wang, Tiejun (1) - 1: ITC, University of Twente, The Netherlands; 2: School of Environmental Sciences, Macquarie University, Australia
Detailed knowledge on surface water distribution, and its seasonal, inter-annual and long-term variability is of high importance for water management, ecosystem assessment and biodiversity conservation. Recently, new multi-temporal Landsat-based assessments of surface water, such as the Global Surface Water Dataset (GSWD) developed by the Joint Research Centre (JRC), have become available for the globe. Nonetheless, the temporal resolution of Landsat alone may not capture rapid changes in surface water due to extreme events and human activities. In addition, for areas such as the Mediterranean, Landsat may have large temporal gaps due to cloud cover, which affect the accuracy of surface water seasonality and other derived products.
To more accurately represent the short hydroperiods and short-duration flooding of water bodies, it is critical to have imagery coincident with the short hydroperiods flooding. The Moderate Resolution Imaging Spectrometer (MODIS) provides global measurements of the land surface with high frequency, and is therefore well suited for long-term mapping and monitoring surface water at a finer temporal resolution. To account for small water bodies, a MODIS-based analysis would need to move beyond a binary land/water classification method.
We developed an approach for the accurate mapping of surface water fraction for the Mediterranean at fine temporal resolution from MODIS data, in order to assess long-term surface water dynamics, inclusive of small water bodies. Specifically, we developed a global rule-based regression-tree method using spectral metrics derived from MODIS data and topographic parameters generated from a 90 m spatial resolution digital elevation model. The regression-tree model was trained and evaluated with the JRC monthly water history datasets, which are multi-temporal binary water maps derived from Landsat imagery and made available through Google Earth Engine. We achieved a high accuracy of surface water fraction estimates (R2 = 0.91, RMSE = 11.41%, MAE = 6.45%) when comparing actual and predicted water fraction using an independent validation dataset. We then applied the algorithm to 16 years of MODIS data (from 2000 to 2016) to generate a time series of surface water fraction maps at 500-m spatial resolution and 8-day temporal resolution for the Mediterranean. From these maps we also derived secondary products such as water occurrence (i.e. the frequency with which water was present on the surface over 16 years) and seasonality (i.e. the number of month water was present for each year). Our product complements surface water products at fine spatial resolution by adding more temporal detail, which benefits the effective monitoring and assessment of the seasonal, inter-annual and long-term variability of water resources, inclusive of small water bodies. Although our study focused on the Mediterranean area, it has potential to be applied globally.
Water storage variations in densely impounded catchments in NE Brazil from 2009 -2016 using TanDEM-X and RapidEye satellite data
Zhang, Shuping (1,2); Foerster, Saskia (1); Delgado, José Miguel (3); Schuettig, Martin (3); Medeiros, Pedro (4); de Araújo, José Carlos (5); Waske, Bjoern (2) - 1: GFZ Potsdam, Germany; 2: Free University of Berlin (FU Berlin), Berlin, Germany; 3: University of Potsdam; 4: Federal Institute of Education, Science and Technology of Ceará (IFCE), Maracanaú, Brazil; 5: Federal University of Ceará (UFC), Fortaleza,Brazil
Water supplies in northeastern Brazil strongly depend on the numerous surface water reservoirs of various sizes. Among them, small reservoirs constitute the largest proportion. Due to the remoteness of their locations and their large number, the long-term water surface dynamics of these reservoirs remain poorly known. Previous studies with remote sensing imagery of coarse resolution often failed to map the small water bodies and are mostly constrained to the mapping of the water surface variations. Whereas, knowledge of the water storage capacities and actual volumes of the reservoirs is essential for understanding, managing, and modelling the local and regional water resources.
In this study, we focus on four catchments of different reservoir densities in NE Brazil. DEMs of high spatial resolution and accuracy were generated from bistatic TanDEM-X with single-pass interferometry and validated with the global TanDEM-X DEM. As the TanDEM-X data were acquired at the end of the dry season in 2015 when most reservoirs were dried up,the DEMs generated from these TanDEM-X data reveal the bathymetry of the reservoirs clutters. Time series of RapidEye satellite imagery were acquired from 2009 to 2016 for the same study area to map the water area changes. From the DEMs of high accuracy and resolution, the parameters regarding the water storages capacities of all the reservoirs within the four studied catchments were derived and analyzed for regional characteristics. Furthermore, the dynamic water storages of the reservoirs in the region was revealed for the period of 2009 to 2016. The contribution of reservoirs of different sizes along the time series was also presented. This study act as a pilot for the further water resources study in the region. The results of our study provide insights for the effective water management and will facilitate potential improvement of the local and regional hydrological modelling.
Large Scale Water and Wetness Detection using a Multi-Sensor and Multi-Temporal Approach
Riffler, Michael; Moran, Andrew; Dullek, Björn; Schleicher, Christian; Walli, Andreas; Weichselbaum, Jürgen - GeoVille Information Systems and Data Processing GmbH, Austria
The detection and consistent mapping of open water bodies is a baseline requirement for scientific research, political and economic decision-making and all sorts of applications such as flood hazard management and development aid, among others. When time series are considered, significant natural and anthropogenic changes and trends in the state of the earth’s environment can be identified and regularly monitored. In order to provide a homogeneous, high resolution (HR) inventory of water cover over large areas (e.g. continental scale), Earth Observation (EO) satellite data has proven to be an ideal source of remote detection. However, hitherto a global and homogeneous HR water inventory is still not available.
As a starting point for a possible future global water cover datasets we present results of three different projects:
The methods developed by GeoVille are highly automated and were applied on a pan-European scale for the production of the “Water and Wetness” 20m HR layer on a European scale for the European Environment Agency (EEA). The initial “Water and Wetness” status layer, which reflect the permanent and temporary water- and wet areas over Europe, was delivered and accepted by EEA in October 2017. The method was also successfully applied for several large regions in Africa within the GlobWetland Africa project and for further large test sites in Africa and South America within the framework of the EO4SD – Water Resource Management project, where continual production occurs in near real time down to a bi-weekly time scale. Both projects are financed by the European Space Agency (ESA).
The applied approach is based on large amounts of observations over continuous time periods of multi-sensor EO satellite data of HR Sentinel-1 and -2 data and supported by historical input from preceding Landsat mission to fully cover the dynamics of open surface water bodies.
The classifications are deduced (i) from a selection of appropriate optical water and wet sensitive bio-physical indices and image composites on varying time scales from seasonal to monthly (Ludwig et al., 2017) and (ii) from radar-derived backscatter data (Sentinel-1 complemented by historical Envisat ASAR) (Naemi et al., 2009; Wagner et al., 1999). Using a rule-base, both the optical and radar data are fused into joint water and wetness layers, each adapted to the specific user specifications.
All results undergo a scientifically sound validation process and quality checks demonstrating the very high reliability and quality of the outputs.
Ludwig, C., Walli, A., Schleicher, C., Weichselbaum, J., and Riffler, M. (2017), A highly automated algorithm for wetland detection using multi-temporal optical satellite data, Manuscript submitted for publication.
Naeimi, V., Scipal, K., Bartalis, Z., Hasenauer, S., Wagner, W (2009). An Improved soil moisture retrieval algorithm for ERS and METOP scatterometer observations. IEEE Trans. Geosci. Remote Sens., 47, 1999–2013.
Wagner, W., Lemoine, G., Rott, H. (1999). A Method for Estimating Soil Moisture from ERS Scatterometer and Soil Data. Remote Sensing of Environment, 70, 191-207.
16:00 - 18:00
From Simulations to Reality – Sentinel-2A for retrieving Water Constituents and Benthos in Lakes
Dörnhöfer, Katja; Oppelt, Natascha - Kiel University, Germany
Water constituents, such as phytoplankton, total suspended matter or coloured dissolved organic matter (CDOM) and, in shallow water, benthic substrate composition are important indicators of lake water quality. Recent monitoring techniques still focus on in-situ sampling at specific measurement points or mapping of benthic substrate composition along diving transects leading to temporal intervals ranging from months to several years. Rapidly changing lake conditions such as bloom events, however, require alternative monitoring approaches, which are able to capture the spatial and temporal dimensions of these changes. Remote sensing is an ideal solution for this purpose. Until recently, the absence of high spatial resolution sensors suitable for inland water body analyses narrowed water remote sensing to large water bodies. The new Sentinel-2 satellites offer a spectral, radiometric and temporal resolution, which is potentially suitable for retrieving water constituents; the spatial resolution further allows a derivation at high spatial detail. Additionally, it offers the possibility to analyse narrow shallow water areas and to map benthic substrate composition.
To improve the knowledge about Sentinel-2’s capability for mapping water constituents and benthic substrate composition we conducted a two-step accuracy analysis. In a first step, we used a state-of-the art, physically based radiative transfer model, i.e. Hydrolight, to simulate Sentinel-2A-like remote sensing reflectances of water bodies with varying optical characteristics. For simulated spectra, water constituent concentrations differed between clear, CDOM-enriched and phytoplankton-rich waters with different phytoplankton types (i.e. diatoms or cyanobacteria). For shallow water spectra, we varied bottom substrate composition and water depth. Using an independent, physically based inversion model, i.e. WASI, we inversely modelled the simulated Sentinel-2A-like spectra and compared the retrieved water constituent concentrations, substrate composition and water depths with the respective input values of Hydrolight simulations. Thus, we provide a theoretic accuracy measure for Sentinel-2A analyses with WASI. In a second step, we applied WASI to atmospherically corrected (using the Modular Inversion and Processing System MIP) Sentinel-2A data acquired over two lakes with optical characteristics similar to the Hydrolight simulations, i.e. a eutrophic CDOM-enriched lake and an oligotrophic lake. In optically deep water, retrieved concentrations from Sentinel-2A imagery (chlorophyll-a, total suspended matter, CDOM) were compared to in-situ measurements. In optically shallow water, we assessed the accuracy of retrieved water depths with echo-sounding data. Due to a lack of up-to-date mappings, we evaluated benthic substrate composition qualitatively based on previous mappings.
Mapping Biomass Of Macroalgae In The Mar Piccolo Of Taranto (Southern Italy, Mediterranean Sea) By Means Of High Resolution Satellite Remote Sensing
Micheli, Carla (1); Cecere, Ester (2); Cibic, Tamara (3); De Cecco, Luigi (1); Petrocelli, Antonella (2); Pignatelli, Vito (1); Portacci, Giuseppe (2); Rubino, Fernando (2); Borfecchia, Flavio (1) - 1: ENEA, Italian National Agency for New technologies, Energies and Sustainable Economic Environment. Research Centre Casaccia 2400/00123 Roma, Italy.; 2: Consiglio Nazionale delle Ricerche (CNR), Istituto per l’Ambiente Marino Costiero IAMC, 74123 Taranto, Italy; 3: OGS (Istituto Nazionale di Oceanografia e di Geofisica Sperimentale), Sezione Oceanografia, Via A. Piccard 54, 34151 Trieste, Italy.
An integrated method based on the most recent satellite remote sensing techniques was implemented for mapping the biomass of the aquatic vegetation growing in the Mar Piccolo of Taranto (Southern Italy, Mediterranean sea). This lagoon is constituted by two water bodies delimited by a narrow land strip located near a strongly urbanized coastal territory of Ionian sea that is affected by local anthropogenic impact factors (urban settlement coupled with industrial and navigation activities). Here we have assessed the distribution of two photosynthetic populations of the macro-algae Caulerpa prolifera and Hypnea cornuta (Kützing) J. Agardh growing in the two inlets of the lagoon.
The Landsat 8 OLI coupled and the Sentinel 2 MSI (of the Copernicus EU program) belonging to the new HR (High Resolution) operative sensor family providing improved spectral and radiometric resolution respects to the previous ones, were exploited for the methodology implementation. The satellite multispectral data, including the entire coastal area of interest, were acquired and atmospherically preprocessed with the objective to test their improved mapping capabilities on basis of the previously developed operative methods and near synchronous sea truth data. The processed point samples measurements were exploited for multispectral data calibration with the support of the statistic and bio-optical modelling approaches to obtain improved thematic maps of the distributions of the biophysical parameters of interest, at suitable ground resolution. The obtained results were then customized in the perspective not only to support the sustainable management of this fragile coastal areas but also to provide the knowledge basis for the exploitation of the algae biomass in the framework of biofuels and bioenergy production from these renewable sources.
Testing A Flood Mask Correction Method Of Optical Satellite Imagery Over Irrigated Agricultural Areas
Michail, Emmanouil (1); Moumtzidou, Anastasia (1); Gialampoukidis, Ilias (1); Avgerinakis, Konstantinos (1); Scarpino, Maria Gabriella (2); Vrochidis, Stefanos (1); Vingione, Guido (2); Kompatsiaris, Ioannis (1); Labbassi, Kamal (3); Menenti, Massimo (4); Elghandour, Fatima-ezzahra (3) - 1: Centre for Research and Technology Hellas - Information Technologies Institute, Greece; 2: Serco SpA, Italy; 3: Chouaib Doukkali University, Morocco; 4: Technical University Delft, Netherlands
The identification of flooded areas over Earth Observation (EO) satellite images has paved the way to monitor damaged areas and take effective actions. Classifying all pixels of a satellite image as a flooded area or not allows for creating maps which are then used by civil protection agencies and first responders. In this work, a method, firstly implemented for Emergency Management Service (e.g. Copernicus), will be applied for the first time to the detection of changes of surface water bodies, based on water volumes data available from the Moroccan Demonstration Area of the H2020 MOSES (Managing crOp water Saving with Enterprise Services) Project.
The novel method is based on the combination of Mahalanobis Distance-based classification for flood mask creation and morphological post-processing for flood mask correction. The classification is performed by using four-dimensional classification features derived directly from the image pixels (namely R, G, B and Near-Infrared channel). The mask correction consists of three steps.
The first step relies on the assumption that if the percentage of pixels classified as flooded in an image is very small, and then probably these pixels are misclassified samples. So if the percentage of flooded pixels in an image is less than a preset threshold, they are set to non-flooded (the threshold has been empirically set to 5% after examining classification results of a training set).
The second post-processing step aims at eliminating small flooded areas (groups of a few pixels, usually 1 to 10, classified as flooded in non-flooded areas), which are potentially false positives, by applying connected-component analysis: the algorithm counts the number of pixels in such areas and in case it is less than a threshold (10 pixels) marks them as non-flooded.
Finally the third post-processing step, aims at eliminating small non-flooded areas inside flooded area, which are probably false negatives, by applying image dilation and erosion.
Experiments have been performed on satellite images, collected from PlanetLabs  and have been provided in the context of MediaEval2017 - Multimedia Satellite Task (http://www.multimediaeval.org/mediaeval2018). In this work we also examine the performance of the method in the context of water body detection over irrigated agricultural areas.
The current technique is going to be extended in the future by considering synthetic aperture radar (SAR) data fusion and additional features in the context of H2020-EOPEN (opEn interOperable Platform for unified access and analysis of Earth observatioN data).
This work is supported by the projects MOSES (H2020-642258), beAWARE (H2020-700475) and EOPEN (H2020-776019), funded by the European Commission.
 Planet team, 2017. Planet Application Program Interface: In Space for Life on Earth, San Fransisco, CA
New perspectives for Sentinel-2 to support Arctic research
König, Marcel; Oppelt, Natascha - Kiel University, Germany
The Arctic Ocean covers an area of about 14.000.000 km² around the Earth’s northern pole. It is the most isolated ocean having only few connections to the Pacific and Atlantic Ocean. In contrast to other oceans, Arctic surface water appears in solid phase as sea ice most of the year, making it an extraordinary type of waterbody.
As sea ice reflects most of the incoming solar radiation, it is of great importance for the Earth’s energy budget being both – an element as well as an indicator of climate change. The ice surface, however, is not a homogenous layer but a mosaic of different surface types, which change significantly during the seasons and thus surface albedo changes accordingly. During the melt period, surface albedo decreases and more energy is being absorbed, amplifying the melting process – a positive feedback mechanism called the ice-albedo-feedback.
In recent years, the Arctic Ocean has undergone profound changes, highlighted by extreme sea ice minimum extents in 2007, 2012 and 2016. As the dominating element of change, sea ice became a hot topic in various disciplines such as climatology, ecology and even logistics. Hence, detailed spatio-temporal observations of the Arctic Ocean are a key to improve our understanding of processes and interactions in this region, which also is essential to improve e.g. climate and ecosystem models.
The remoteness and inaccessibility of the Arctic Ocean makes satellite remote sensing the primary tool for regular observations of this region. At present, active and passive microwave sensors are the main monitoring instruments. Despite the limited availability of sunlight during winter and frequent cloud cover during summer, multiple bands in the VIS-NIR/SWIR wavelength regions enable different applications of optical satellite remote sensing, e.g. albedo retrieval and water colour analysis. Some sensors with low spatial resolution such as MODIS and AVHRR have been applied for Arctic sea ice monitoring; many sea ice features such as ponds, ridges or leads, however, have a size in the order of tens of meters and therefore cannot be mapped appropriately using these sensors.
Nevertheless, the European Space Agency’s Sentinel-2 sensors reach up to 84° N and may cover vast areas of the Arctic Ocean. Sentinel-2A and B offer spatial, radiometric and spectral resolutions that enable observations of sea ice features such as melt ponds, ridges and open water areas with unprecedented precision. Simultaneously, the increased temporal resolution near the poles increases the chance to match clear-sky events.
To demonstrate the potential of Sentinel-2 for Arctic research, we use a Sentinel-2A L1C product acquired during an RV POLARSTERN cruise in summer 2017 to distinguish different surface types (open water, shallow snow, deep snow, melt pond, ridges). Applying a maximum-likelihood classification approach, we highlight the capabilities of Sentinel-2 to map sea ice features and different surface types.
Mapping Phytoplankton Abundance and Diatom Fraction in the Chesapeake Bay
Zheng, Guangming (1,2); DiGiacomo, Paul M (1) - 1: NOAA, United States of America; 2: GST, Inc.
Satellite remote sensing is a useful tool for mapping coastal water quality synoptically and frequently. One of the focal points of water quality remote sensing is phytoplankton. It drives aquatic ecosystems and is often used as an indicator of eutrophication status. However, in coastal waters it is difficult to derive chlorophyll-a concentration ([Chl-a]) from the spectral remote-sensing reflectance, Rrs(λ), because of the challenge to separate phytoplankton and nonalgal suspended and dissolved materials. In addition, it is even more difficult to distinguish different types of phytoplankton from remote sensing. Improving our capabilities to estimate these parameters are useful from the water quality stand point because information on phytoplankton composition in addition to [Chl-a] can potentially help build a more robust surrogate for nutrients than using [Chl-a] alone, thus aiding in the use of satellite data for nutrient pollution assessments. We recently developed new approaches to derive [Chl-a] (Zheng and DiGiacomo, Remote Sens Environ, 2017) and phytoplankton diatom fraction (Zheng and DiGiacomo, Limnol and Oceanogr, 2017) from satellite data. In this study, we used these two algorithms to estimate the phytoplankton parameters in the Chesapeake Bay which is the largest estuary in North America. Phytoplankton abundance ([Chl-a]) were derived based on its light absorption coefficient at ~670 nm; whereas information on phytoplankton diatom fraction were extracted based on the blue-to-red spectral band ratio of its light absorption coefficient. These approaches allowed us to obtain more robust information on phytoplankton in turbid waters. Using these new methods we analyzed the phytoplankton phenology in the Chesapeake Bay, and show that phytoplankton abundance and composition exhibit seasonal varaiations that have been previously reported by field studies, so does the composition of nonalgal suspended particles. Our results suggest that the new appoaches can provide more detailed and specific information about phytoplankton. This initial success suggests that the new approaches are promising and deserve further efforts in this direction in the future.
The Use of Research and User Support for Sentinel Core Products (RUS) for Artic Lake Ice Monitoring
Šmejkalová, Tereza (1); Castro Goméz, Miguel (1); Palazzo, Francesco (1); Remondiere, Sylvie (1); Guzzonato, Eric (2); Mora, Brice (2); Jeansou, Eric (3); Soleilhavoup, Isabelle (3); Fabry, Pierre (4) - 1: Serco SPA, Italy; 2: CS, France; 3: Noveltis, France; 4: Aʟᴏɴɢ-Tʀᴀᴄᴋ, France
Since the launch of the first satellite (Sentinel-1A) in April 2014, the free and open data of the Sentinel satellites provide essential information to monitor our environment. With this massive amount of data (expected 10 petabytes of data each year with all Sentinel satellites operational), the challenge is now moving towards storage and processing capabilities. The RUS service (Research and User Support for Sentinel Core Products) provides a scalable platform in a cloud environment to facilitate the uptake of Copernicus data. In this contribution, we demonstrate its use through a case study to evaluate the ice cover dynamics of thermokarst lakes in Northern Russia over the operational period of Sentinel-1 constellation (2014 - present).
Thermokarst lakes are a major component of permafrost landscapes and have been shown to be sensitive indicators of climate change. It has been well documented that temperatures at high latitudes rise with rate at least double the global average. While the time series available from Sentinel-1 alone is too short to draw any conclusions about the changes in the lake ice regime, we aim to present the methodology implemented within the RUS environment and the advantages of RUS when processing large datasets
Mapping the Seagrass and coastal Habitats of Mediterranean Islands using the new HR satellite Multispectral sensors
Borfecchia, Flavio (1); Micheli, Carla (1); De Cecco, Luigi (1); Sannino, Gianmaria (1); Struglia, Maria Vittoria (1); Di Sarra, Alcide Giorgio (1); Gomez, Carlo (3); Mattiazzo, Giuliana (2) - 1: ENEA,Italian National Agency for New Technology, Energy and Sustainable Economic Development, Italy; 2: Polytechnic University of Turin, Mechanical Engineering Dep. Turin Italy; 3: Cantieri Navali Esposito S.n.c, Pantelleria-Italy
The Italian islands of the southern Mediterranean, like Pantelleria and Lampedusa, are characterized by high transparency of coastal waters and Posidonia oceanica (PO) meadows, with sea beds habitats that still present significant levels of biodiversity and specific adaptation to accentuated hydrodynamics of their shores. Despite their shallow waters wealth and natural heritage, often safeguarded by protected areas, the increasing anthropogenic activities, linked to tourism and fishing, negatively impact on these natural ecosystems with consequent potential damages and needs for their more effective monitoring for supporting their sustainable management. Additional impact factors derive from the difficulties of supplying electrical power in many islands, not connected to the national electricity grid, with consequent ship traffic inrease for local electricity production through fossil fuels. Since various years Lampedusa is interested by increase of vessel traffic linked to illegal immigration from north African countries and the related patrol and rescue European missions, with potential additional impacts on the PO meadows growing there. In order to provide eco-compliant electric power from renewable resource in 2013 the first Italian Inertial Sea Wave Energy Converter (ISWEC) prototype developed by Polytechnic of Turin has been installed in the coastal area offshore of Pantelleria. The prototype installation and operation involves an interaction with local PO and seagrass meadows and possible water transparency decreasing. Thus monitoring of local PO ecosystem is mandatory in order to allow the detection of potential stress and damages due to ISWEC related activities and/or other factors. During last 2015 and 2016 years a sea truth campaigns over the areas of interest has been performed and point measurements of several biophysical parameters (biomass, shoot density, cover) related to PO phenology has been acquired by means of original sampling method on stations distributed along a bathymetry gradient, starting from the ISWEC location, at 31 m. of depth.
Once suitably corrected for atmospheric noises the satellite EO techniques are recognized as effective multiscale tools for monitoring shallow water and marine ecosystems. In this context our goal was to develop a methodology, integrating satellite remote sensing HR (High Resolution) techniques with in situ observations and biophysical parameters analysis, for suitably monitoring and mapping PO meadows growing in coastal shallow waters of Pantelleria and Lampedusa islands.
The Landsat 8 OLI with the Sentinel 2 MSI (of Copernicus EU program) synchronous satellite multispectral HR data (with 10-30 m of above ground resolution), including the entire coastal areas of interest, were acquired and preprocessed with the objective to test their improved mapping capabilities of PO distribution and related biophysical parameters. Different approach were applied for the indispensable atmospheric preprocessing focusing on the AOD (Aerosol Optical Depth) and adjacency effects noise contributions removal and using auxiliaries atmospheric information about the aerosol load made available by the ENEA station for climate observation Roberto Sarao of Lampedusa island. The data previously obtained using side-scan sonar were exploited for calibration/validation of remote sensing derived maps, using advanced supervised classification methods.
Pan-sharpening Methods Applied on Sentinel-2 Imagery for Mapping Inland Water Bodies
Ronchetti, Giulia; Sona, Giovanna - DICA (Department of Civil and Environmental Engineering), Politecnico di Milano, Italy
In the last two years great importance was given to Sentinel-2 mission because it provides high resolution and high frequency multispectral images, which can be employed in different applications.
In this context, this work investigates the use of Sentinel-2 imagery with the aim of detecting inland water bodies in natural and man-made environment. According to literature, the Modified Normalized Difference Water Index (MNDWI) gives reliable results in terms of mapping inland water bodies. This index is derived from the green and the short-wave infrared (SWIR) bands, that have different spatial resolution in Sentinel-2 imagery, 10 m and 20 m respectively. MNDWI map with resolution of 20 m can be obtained by upscaling the green band, instead maps with higher spatial details can be generated by downscaling the SWIR band with the application of pan-sharpening on Sentinel-2 data. Since there is no panchromatic (PAN) band in Sentinel-2 images, the 10 m resolution channel with the highest correlation with the SWIR one is chosen as PAN-like band. Considering the high number of pan-sharpening methods, this work evaluates four common algorithms, usually available in satellite image processing software packages: the Intensity, Hue, Saturation (IHS) transformation, the Principal Component Analysis (PCA), the Brovey method and the Gram-Schmidt method. All the algorithms are implemented in Matlab®, starting from codes available online and specifically adapted to fit to Sentinel-2 data.
Two different study cases are reported in this work. Preliminary studies are conducted on the Po river delta area, mainly natural environment, in order to test and compare the methods. According to correlation values, the near infrared (NIR) band is selected as PAN-like band. MNDWI maps produced by the application of the Brovey method and the IHS transformation give the most reliable results. The first algorithm can find better no water areas, while the latter detects more accurately water bodies. Other studies are carried out on a more urban site: the eastern area of the city of Milan, both rural and man-made environment, full of canals and quarries.
Orbital Grain Size Mapping From Sentinel 2 Images
Marchetti, Giulia (1); Bizzi, Simone (1); Belletti, Barbara (1); Carbonneau, Patrice (2); Castelletti, Andrea (1) - 1: Politecnico di Milano, Italy; 2: Durham University, UK
In the last two decades, remote sensing (RS) technology has opened up new possibilities for river science, providing continuous data over entire catchments, allowing for a comprehensive river management and monitoring.
A key variable to delineate and characterise river geomorphic units and to better understand fluvial processes is bed material grain size distribution and its pattern of change along rivers and over time. Field based approaches are still dominating in river surveys and their prerequisites limit their application at the network scale and may fail in guaranteeing an objective and repeatable monitoring assessment (Bizzi et al., 2015). Depending on surface characteristics, sunlight reﬂected from the soil changes as function of many parameters such as surface roughness, linked to geometry and shape of single grains. Previous studies have shown that there is a correlation between particles and spectral signature of a sediment (Black et al., 2014, Carbonneau et al., 2004). Building on this evidence, this paper investigates the potential of Sentinel 2 multispectral data in discriminating classes of sediment (from fine gravel to coarse cobbles) of exposed river sediment bars. The methodology uses near ground sUAS imagery in order to correlate local grain sizes to Sentinel 2 radiance values.
Results show that the relation between reflectance values registered by Sentinel 2 sensors and surface grain size percentiles is significant. The most sensitive band capable alone to explain around 70% of the variance is in the SWIR region. This methodology has then been applied on about 500 km of the major Italian river, the Po river. The resulting ﬁning pattern is comparable to others reported in literature and is coherently linked with main tributaries and river infrastructures existing along its course.
This method represents potentially a major advance in our current ability to characterize ﬂuvial habitats and processes along major river systems, thanks to the characteristics of Sentinel 2 data, free available worldwide with a return time frequency of about 5 days. Moreover, the combination of satellite and sUAS remote sensed technologies introduces a new, objective, efficient and repeatable methodology for grain size mapping in large rivers, opening up broad perspectives for future fluvial survey practices. Further studies are required to better understand physical explanations behind observed phenomena by testing the methodology on other large rivers different in geology, lithology and geographic position (i.e. latitude).
Environmental Monitoring of the Wetlands in the network RAMSAR using techniques of Observation of the Earth with Sentinel 2
García Fernández, Miguel Angel
García Fernández, Miguel Angel (1,2); Perez Gonzalez, Maria Eugenia (1) - 1: Complutense University of Madrid, Spain; 2: Carlos III University of Madrid, Spain
This work is the initial stage of the doctoral thesis "Environmental monitoring of the wetlands in the network RAMSAR using techniques of Observation of the Earth with RADAR and optical technology" developed by Miguel Angel García Fernández in the Geography Department of the Universidad Complutense de Madrid (Spain).
The first stage of the doctoral thesis is the environmental monitoring of several interior wetlands belonging to the network of RaMSAR in Spain, through the use of optical images of Sentinel 2.
In a second stage will be performed the analysis of these same wetlands with the Sentinel 1 images will be to see the potential of environmental monitoring of these small wetlands RAMSAR combining optical and RADAR images.
The study area covers a total of 10 wetlands RAMSAR in Andalucia:
Reserva Natural Laguna del ChincheReserva Natural Laguna HondaReserva Natural Laguna del Conde o El SalobralReserva Natural Lagunas de ArchidonaReserva Natural Lagunas de CampillosLaguna de Fuente de PiedraEmbalses de Cordobilla y MalpasilloReserva Natural Laguna de TíscarReserva Natural Laguna de los JaralesLagunas del Sur de Córdoba (Zóñar, Rincón y Amarga)
In this first phase of environmental monitoring with optical images, is done the control of different quality indexes of the wetland in order to verify the good ecological status of these wetlands in the hydrological year (2016-2017), some index in this stage are:
1 Secchi disc depth
2 Chlorophyll a.
4 Concentration of solids in suspension.
5 Dissolved organic matter.
8 Thermal contamination.
9 Water layers and depth of the wetlands
The ultimate goal of this phase is to generate a reliable and simple methodology that allows to check the status of these wetlands in various controllable parameters with optical images and be able to generate a reproducible and applicable methodology in wetlands of similar features in other places.
Implementing Different Methodologies for the Delineation of Mediterranean Wetlands
Morant, Daniel; Doña, Carolina; Picazo, Antonio; Ferriol, Carmen; Santamans, Anna C.; Rochera, Carlos; Camacho, Antonio - Cavanilles Institute of Biodiversity and Evolutionary Biology, University of Valencia, Spain
Wetlands are considered as biodiversity hotspots and active ecosystems with an outstanding importance due, among other important reasons, to the ecosystem services they provide. The Ramsar Convention, the Habitats and Water Framework European Directives, and other international, European or national commitments, aim to assess the wetland conservation and ecological status to achieve a favourable condition of these systems. Environmental reporting and other management and monitoring reports require the delineation and area estimation of the water-covered and wetland areas for their specific purposes. Remote sensing techniques allowed to carry out the water delineation of wetlands in an automatic, simple, fast and, sometimes, accurate way. Other methods developed in GIS and geoportals have enabled users to delimit manually these surfaces. In our work, a compilation study of the different water delineation methods was applied to saline temporal inland and coastal Spanish wetlands in order to identify and analyse the free options users have nowadays. Different image analysis processes were applied to a sequence of temporal free available Landsat images. Supervised and unsupervised classification, single band threshold study, NDWI index estimation, as well as a specific water identification algorithm, were implemented. Free downloaded historic aerial images and orthophotos were also used, thus delineating the water bodies by the available tool of polygon creation on a GIS. In parallel, the water-covered surface was estimated using two geoportals, both of them with available high-quality base orthophotos. The first one was the free version of the Google Earth application, from which water delineation could be done manually by its area calculation tool. The other was a portal developed by the SWOS project, where many indices and products were grouped, some of them allowing to estimate the water coverage. Surface water coverage for the different wetlands, when possible, was estimated by all methods, and results in every studied method did not strongly differed. Therefore, many of the assessment criteria set by the legal requirements, plus wetland mapping, management, monitoring and policy development, could be estimated from the results obtained using these methodologies by the interested administrations, organisations and entities.
Estimation of water coverage in lenitic systems by combining remote sensing and genetic programing. Application to lakes in the Mediterranean basin of the Iberian Peninsula.
Doña, Carolina (1); Morant, Daniel (1); García Picazo, Ana (1); Sánchez, Juan M. (2); Camacho, Antonio (1) - 1: Cavanilles Institute of Biodiversity and Evolutionary Biology, University of Valencia, E-46980 Paterna, Valencia, Spain.; 2: Department of Applied Physics, Regional Development Institute, University of Castilla-La Mancha, Campus Universitario S/N, 02017 Albacete, Spain.
The water balance and hydrological variations are intimately tied to potential ecological changes in a lenitic ecosystems. Understanding the dynamics of water in lakes helps the conservation and recovery of these ecosystems. The long-term monitoring of the lakes to study their hydrological patterns in detail helps to develop adequate management plans for lake protection, conservation and recovery, as required by the European Water Framework and the Habitats Directives. In this work, we validate a recent remote sensing methodology to estimate the changes in water coverage of lenitic systems from satellite images. This approach is based on an algorithm developed using Genetic Programing (GP) techniques that allows to discern between water and non-water pixels using the near infrared band of Landsat-7 and different thresholds, for playa lakes and non-playa lakes. This algorithm was first developed to be applied to shallow lakes of the Biosphere reserve of La Mancha Húmeda, Spain. In this work, we assess the performance of the model focusing on several typologies of lakes located in a semiarid region, such as the Mediterranean basin of the Iberian Peninsula. A total of 25 lakes and ponds of 9 typologies were used to carry out the study. These typologies include marshes; salt ponds; as well as saline, freshwater, seasonal and permanent lakes, among others. Results were compared with estimations from high resolution images, as reference dataset. The classification errors were evaluated by means of the kappa coefficient. The GP algorithm shows acceptable results, with accuracy values over 85% in most of the study cases, and acceptable kappa values in the estimation of water pixels (κ ≈ 0.8). As it might be expected, some discrepancies are found in the estimation of water pixels when the satellite and reference image differ temporary in seasonal lakes. Furthermore, based on the spectral matching between Landsat and Sentinel-2 sensors, it is intended to apply this methodology to Sentinel-2 imagery, improving the spatial resolution from 30 m to 10 m. To sum up, this approach is a useful tool to monitor the hydrological patterns of seasonal and permanent lakes in semiarid areas. This might be useful for management strategy-linked lake conservation, and help to accomplish the goals of both the European Water Framework Directive and the Habitats Directive.
SAR Based Mapping of Flooded Areas for the Validation of Short-term Flood Forecasting
Ponomarenko, Maria; Pimanov, Ilya - St. Petersburg Institute for Informatics and Automation of the Russian Academy of Sciences (SPIIRAS), Russian Federation
Effective operation of flood forecasting systems requires reliable real-time data on inundation areas for timely calibration and verification of the used hydrodynamic models. In case of river floods caused by heavy rainfall the opportunity to obtain data from optical sensors is limited because of dense cloud cover. Synthetic aperture (SAR) techniques are increasingly used today due to ability of SAR to operate independently of natural light and cloud cover receiving high spatial resolution data in near real-time mode. Another important advantage of SAR is free and open access to radar data - thanks to Sentinel missions. For instance, for the territory of Russia Sentinel-1 performs SAR imaging with 2-4 days coverage frequency providing moderate geometric resolution data.
The study area is the city of Veliky Ustyug (Russia) located at the confluence of rivers Suhona and Ug. To identify flooded areas the automatic processing of RADARSAT-2 and Sentinel-1 data processing was carried out in open-source software. Thematic processing consisted of texture analysis and automatic image classification based on thresholding. The visualization of created layers was performed on the basis of information analytical system RegionView developed in SPIIRAS[2,3]. The results of SAR data processing were compared with contours obtained from Sentinel-2 and Russian satellite Resurs-P optical images. According to spatial resolution of data and selected processing technology, it is possible to achieve high accuracy of flood mapping in open areas with low urbanization. Areas of at least 24 m2 are required for a reliable detection.
Due to the constant improvement of data quality, revisit and coverage frequency, automation of processes and reduction of the requirements for end-user systems, integrated interdisciplinary solutions such as described above allows to achieve new results of flood forecasting. The described method was applied as one of the approaches to verify the results of flood forecasting. The output data of independent processing of various data sets (optical and SAR data, field measurements and crowdsourcing, simulation results) demonstrated high convergence. The mismatch of the number of flooded infrastructure objects on the test areas between the described approaches was within the range of 2-17 % (the assessment of the difference is the subject of further research).
The result confirms that SAR data can be successfully applied for operational flood forecasting. In particular, the use of Sentinel data provides an opportunity to regularly obtain and integrate SAR images into the operational flood monitoring and forecasting systems. Further development of proposed methodic aims at joint processing of SAR and optical data and including InSAR techniques.
The research described in this paper was supported by the Russian Science Foundation (project 17-11-01254).
Medium Sized Rivers Wet Channel Probabilistic Mapping from Short Time Sentinel-1 Data Stack
Asaro, Francesco (1); Prati, Claudio Maria (1); Belletti, Barbara (1); Bizzi, Simone (1); Carbonneau, Patrice (2) - 1: Department of Electronics, Information and Bioengineering, Politecnico di Milano, Italy; 2: Department of Geography, Durham University
Several terabytes of data are acquired every day from the Sentinel constellations worldwide. For the first time, this extensive availability of data, both in space and time, offers the possibility to build a methodology suitable for systematic and consistent remote sensed hydromorphological river monitoring, which still represents an open issue. In detail, Sentinel-1 SAR IW-TOPS weather independent observations, thanks to a 20-by-5 m spatial resolution and a six days revisit time over Europe, are in principle suitable to extract the wet channel of large and medium sized rivers (i.e. wider than 25 m). This work presents a methodology to generate fixed-confidence river wet channel masks while preserving spatial resolution. This methodology is a development of an already existing approach that uses probabilistic thresholds on the backscatter intensity (Matgen et al., 2011) computed through Probability Density Function (PDF) modelling. Additionally, the method is applied on micro stacks of data relative to steady windows where the river hydrological discharge is on average constant. The usage of stacks allowed a spatial resolution saving multitemporal despeckling, which produces a strong reduction of water and land PDFs overlapping area (biased pixels region) in the image histogram. Moreover, the multitemporal setting offered the possibility to employ coherence information in addition to intensity one. In so doing, we improved classification performance of biased pixels. A further reduction of the biased area was obtained applying to the image an “ad-hoc” spatial filter accounting for river patterns. This methodology has been implemented in the MATLAB environment and successfully tested on a stretch of Sesia River in Piedmont (Italy) during a low discharge window from June to September 2017. A distributed error of 0.3 pixel/m has been assessed in comparison to near-high resolution (3.7 m) optical images.
Keywords: Copernicus programme; Sentinel-1; river monitoring; water mapping; Synthetic Aperture Radar
Spatial and Temporal Variability of Optical Water Types in Largest Latvian and Estonian Lakes in 2017
Soomets, Tuuli (1); Jakovels, Dainis (1); Uudeberg, Kristi (2) - 1: Institute for Environmental Solutions, Latvia; 2: Tartu Observatory, Estonia
The remote sensing of inland waters is starting to get more attention. But inlands waters are known to be optically complex and more diverse than marine or ocean waters. So, the remote sensing of these waters is more challenging. First step in remote sensing of lakes could be the classification of the water, because different optical water types need different approach to obtain correct water quality products. The optical water type classification used is based on the differences in the reflectance spectra of the lake water. This classification clusters lake waters into 5 different optical classes: Clear, Moderate, Turbid, Very turbid, and Brown. Lake optical type can vary in time and space, depending on the changes in weather, biological composition and physical attributes. We studied the optical water types in three different Latvian lakes: Burtnieks, Lubans and Razna, and in two Estonian lakes: Peipsi and Võrtsjärv. Both, Sentinel-2 MSI and Sentinel-3 OLCI full resolution data was used. The comparison between two satellites was carried out to understand if both satellites are suitable for classifying the water types. Although Sentinel-3 is designed for water monitoring, its moderately low spatial resolution limits the research to only several largest lakes in Latvia and Estonia. Sentinel-2 can provide nearly the same data acquisition frequency and its much higher spatial resolution could allow assessment of practically all lakes. The primary goal of this study was to discover the spatial and temporal differences of lake optical water types in Latvia and Estonia during 2017. This could give us not only better overview of the changes of the lake water but also possibly better retrieval of level 2 products. Additionally, a comparison of Sentinel-3 and 2 data for optical classification was performed.
An Image Classification and Geoprocessing Workflow to Facilitate Network Analysis of Multichannel Rivers
Connor-Streich, Gabriel; Henshaw, Alex; Harvey, Gemma - Queen Mary College, University of London, United Kingdom
Braided rivers are some of the most dynamic geomorphic systems on Earth. Such dynamism means that these rivers pose notable problems for the people that live by and interact with them, however we still have large gaps in our understanding of braided river dynamics that stem, in part, from the spatio-temporal limitations of field-based braided river research. The increased availability of archival satellite remote sensing data, for example Landsat, provides a potential means to circumvent these issues, yet we require new analytical techniques to deal with such large-scale, multi-temporal data. Graph theory, the branch of mathematics tailored to network analysis, offers exciting possibilities for the characterisation of multichannel river network structure and behaviour and linkage to causal processes but methodological developments are required to facilitate this. In this paper we present a workflow to derive graph representations of multichannel rivers from remotely sensed imagery. The workflow is adaptable, allowing for different types of imagery (air photos, multispectral satellite data, hybrid, etc.) and classification techniques, but is exemplified here using Object-Based Image Analysis (OBIA) techniques for fast, accurate classification of Landsat or Sentinel imagery that extracts the three land covers of greatest morphological interest in braided river studies: water (braided channels), vegetation and gravel. Simple geoprocessing techniques (applied in ArcGIS) are used to extract channel centrelines from the classified channel network. These centrelines also capture the connectivity between channels via the bifurcations and confluences that give braided rivers their distinctive planform. The connectivity between channels is derived and recorded within an attribute table as a “from” and “to” field for each pair of connected channels. These connectivity lists are used to generate graph models of braided river networks that can subsequently be analysed using graph theory. A wide range of graph theory metrics and indices have been developed in diverse fields of research. We have identified a suite of these measures that are likely to have a morphological interpretation and, finally, present some results highlighting the utility of indices derived using graph analysis for understanding change in a gravel bed braided river (the River Tagliamento, Italy).
Comparison of Sentinel Water Mask and other indices for water mapping on Sentinel-2 images
Robak, Anna; Milczarek, Marta; Gadawska, Alicja - Space Research Centre Polish Academy of Science, Poland
Flooding is one of the most common and widespread climatic hazard. It carries multiple risks, both to human health and to the economic growth on a micro and macro scale. For that reason, it is crucial to monitor inland water surface using multiple data sources. As flooding is likely to become more common, more intense and larger in scale in many areas, it is important to improve techniques and methods for detecting water. This improvement should be aimed especially at remote sensing techniques, which occur to be the most cost- and time-effective methods.
The use of datasets from new sensors implies adaptation of existing methods or development of new ones to obtain useful information for various purposes. The overall aim of the research was to test existing, commonly used water indices on datasets from sensor MSI onboard Sentinel-2 satellite and compare their results with a new index for water mapping – Sentinel Water Mask (SWM). The other two tested indices were: AWEInsh and MNDWI. Four different test sites were chosen in regards to the location, land cover and season. SWM was developed for Sentinel-2 imagery analysis. It is calculated on the image without atmospheric correction and provides quick and effective detection of water, which is especially important in flood rapid assessment for crisis management. The formula of SWM index has been constructed to increase the contrast between water and all other non-water areas, to be able to extract water more easily.
Visual interpretation shows that all of the tested water indices on Sentinel-2 data achieve better accuracy on the images before atmospheric correction. The SWM index has a greater ability to detect narrow streams than the other indices. Moreover, it works better on wetlands and built-up areas. Another advantage is that water visibility does not mix too much with clouds and shadows. Nevertheless, very deep shadows of clouds or tall buildings can still cause some problems. Statistic accuracy confirms visual interpretation. SWM index achieves the best results for both parameters Overall Accuracy and Kappa Coefficient. The overall accuracy of SWM is between 95.83% (AWEInsh – 89.83%, MNDWI – 88.17%) and 99.17% (AWEInsh – 98.18%, MNDWI – 95.5%) for images before atmospheric correction.
Mapping suspended particulate matter in West African water bodies with Sentinel2
Grippa, Manuela (1); Robert, Elodie (1); Martinez, Jean-Michel (1); Gosset, Cindy (1); Pinet, Sylvain (1); Soumaguel, Nogmana (2); Touré, Amadou Abdourahamane (3); Kergoat, Laurent (1) - 1: Géosciences Environnement Toulouse (GET), France; 2: Institut de recherche pour le développement (IRD), Bamako, Mali; 3: Université Abdou Moumouni (UAM) de Niamey , Niger
Suspended particulate matter (SPM) in surface waters plays a fundamental role on water resources, reservoir management and inland water ecology. In addition, SPM can carry and foster the development of viruses and bacteria pathogenic to humans, concerning particularly diarrheal diseases. In the Sahelian area, surface water is a critical water resource and is often employed for domestic uses. Mapping SPM in Sahelian water bodies is therefore a key issue. However, SPM spatio-temporal dynamics in this region is poorly known given, on the one hand, the scarce in-situ data available and, on the other hand , the high spatial and temporal resolution necessary to follow Sahelian water bodies from space, that only recent satellites can provide.
This presentation addresses the retrieval of SPM with high resolution optical sensors, particularly Sentinel2, in West African water bodies. Optical remote sensing in this region can be quite challenging given the extremely high and variables values of SPM found in tropical Africa as well as the high aerosol loading (mineral dust and biomass burning) that makes atmospheric corrections more difficult and may be detrimental to SPM retrieval.
Water sample have been routinely collected to determine SPM concentration over different kinds of water bodies: the Agoufou lake in northern Mali, the Bagré reservoir in Burkina Faso, two ponds in Southwest Niger and the Niger river at Niamey. Some additional measurements were also performed on few ponds in the Ferlo region, Senegal, and the Senegal river.
In-situ data were used to evaluate different indexes to derive SPM from the reflectance in the visible and infrared bands of LANDSAT and Sentinel2. We have found that the NIR reflectance from both sensors is well suited to map SPM up to high values and that the same retrieval algorithm can be applied to estimate SPM for the different kinds of water bodies (ponds, lakes and rivers) investigated. In addition, the atmospheric correction scheme employed to retrieve surface reflectance for Sentinel2 products provided by the Theia pole is found to be accurate enough for monitoring the SPM seasonal and interannual variability in this region.
Finally some application of SPM retrieved form Sentinel2 are reported showing the potential of this sensor for monitoring the Niger river “red flood” hydrological phenomenon, estimating water quality in the the Guiers reservoir, that provides water to Dakar, and identifying the drivers responsible for the SPM seasonal variability in Sahelian ponds and lakes.
Sahelian ponds and lakes seen by SWOT
Grippa, Manuela (1); Rouzies, Cyprien (1); Biancamaria, Sylvain (2); Blumstein, Denis (2); Gal, Laetitia (3); Gosset, Marielle (1); Kergoat, Laurent (1) - 1: Géosciences Environnement Toulouse (GET), France; 2: Laboratoire d'études en géophysique et océanographie spatiales (LEGOS); 3: Laboratoire d’étude des Interactions entre Sol-Agrosystème-Hydrosystème (LISAH)
Small water bodies provide a critical water resource for livestock and people living in the Sahel, they play an important role in eco-hydrosystems, in terms of biodiversity and emission of green house gases, and they impact the spread of water-borne diseases. Moreover they provide essential information to understand hydrology of these areas, that can be quite complex as testified by the paradoxical increase in surface water and river runoff during and after the big Sahelian drought.
Monitoring, modelling and better understanding the dynamics of this small water bodies is therefore a key issue. However, they have not been much investigated up to now given, on the one hand, the scarcity of in-situ data in these poorly instrumented regions and, on the other hand, the high spatial resolution necessary to quantify water amounts from spaceborne sensors. The SWOT mission, based on radar interferometry in the Ka-band at low incident angles, will allow monitoring water levels in small water bodies (down to 1km2) wordwide.
Here we assess the potential of SWOT for monitoring water volumes of ponds and lakes in the Sahel. We focus on northern Mali, which has witnessed an important increase in surface water and runoff despite the precipitation decrease. This site is instrumented by the AMMA-CATCH observatory that provides in-situ hydro-meterological variables including water level, turbidity and suspended sediments in the Agoufou lake.
Synthetic SWOT data were produced using the SWOT HR simulator developed at JPL-NASA that accounts for geometrical (layover) errors and instrumental noise. Input data were DEM and observed water levels. Water level were then retrieved from simulated SWOT data with an accuracy better than about 4 cm, well below the specifications. This allows an accurate monitoring of the seasonal cycle. Slightly worst results were obtained for another pond with a more elongated shape and steeper surroundings.
Data from the Global Precipitation Mission have been analyzed to investigate the spatio-temporal variability in the backscattering coefficient at the same frequency and incident angles as SWOT. The difference in sigma0 between water and soil can be small in some cases and therefore deriving water masks by SWOT in this region may not be always straightforward. Coupling SWOT derived water levels with water areas derived by optical remote sensing sensors, such as Sentinel2, will be very helpful to derive accurate estimations of water volumes.
Finally, water volumes derived by remote sensing were coupled to a water balance equation that accounts for precipitation, evaporation and surface/ subsurface exchanges, to quantify the water inflow into the lake. Ponds and lakes can be used as gauges to estimate runoff, an essential variable to constrain hydrological models. This highlights the unique opportunity provided by SWOT to monitor critical water resources.
Quantifying Distributed Water Availability in Small Dams in the State of Ceará, Brazil
Delgado, José Miguel (1); Zhang, Shuping (2); Schuettig, Martin (1); Foerster, Saskia (2) - 1: University of Potsdam, Germany; 2: GFZ German Research Centre for Geosciences
The state of Ceará in Brazil is part of the drought-prone northeast Brazil. Precipitation over inland regions is limited to a short three month season. Still, millions of people are able to live in this region due to thousands of dams that were build in the region. In Ceará, there are 8000 unmanaged small dams, as well as over 200 strategic, state-managed reservoirs.
To date there has not been any monitoring of the small dams, although they are vital for irrigation, cattle and as a source of drinking water after treatment for some rural communities. Knowing how much water is available in these reservoirs would allow for a better allocation of emergency measures.
By employing a combination of TanDEM-X and Sentinel-1 missions, we were able to (1) derive a 10 m resolution DEM of the bathymetry of the reservoirs during a very dry period in 2015 and (2) operationalize the weekly extraction of watermasks from Sentinel-1 SAR data. By combining the two products we are able to estimate the water availability across a large area. A spatial database is now available and growing containing a time-series of water extent in each of the over 8000 small reservoirs.
The methodology for the DEM generation from TanDEM-X consists of (a) filtering, geocoding and conversion, (b) classification of suitable areas for altimetry correction with ICEsat data, (c) offset calculation, (d) differential phase simulation and unwrapping, (e) phase to height conversion and finally (f) geocoding and postprocessing. For generating watermasks from Sentinel-1 the methodology and code is available on a public repository and includes subsetting, calibration, filtering and band arithmetics for threshold determination.
Here we present the validation of the first months of operation in three catchments, where field information is available concerning the height-area-volume relationship and water stage monitoring is available in real time. Further work to improve the results is still necessary and is being added to the operational workflow regularly.
A 10-day Mean Surface Water Extent at Global Scale at 0.25°x0.25° Spatial Resolution, from 1993 to Present: the Global Inundation Extent from Multi-Satellites 2.0 (GIEMS 2.0)
Jimenez, Carlos (1,2); Prigent, Catherine (2,1,3); Aires, Filipe (2,1,3) - 1: Estellus, France; 2: LERMA, Observatoire de Paris, France; 3: Department of Earth & Environment Engineering, Columbia University, USA
An initial methodology was developed to estimate the extent and dynamics of surface waters, at global scale, from multiple satellite observations (passive and active microwave observations and visible and near infrared). A monthly estimate of surface waters was produced for 1993-2007, with a 0.25°x0.25° spatial resolution: the Global Inundation Extent from Multi- Satellites (GIEMS) (Prigent et al., 2007, 2012; Papa et al., 2010). This dataset has been extensively used by the community, for methane emission estimation, for hydrological modeling for example. The data set has also been downscaled up to 90m, for local applications (Aires et al., 2017).
This initial GIEMS methodology required key ancillary data, and to obtain a climate quality dataset, these data must be of similar provenance and of constant quality over time. However, some data used in this initial method are not available after 2007. A new methodology has been developed that requires less ancillary data, to overcome the problems of discontinuity of data sources over long time record.
We will describe the new GIEMS methodology, along with the necessary data for the processing. Preliminary results will be presented. The new methodology is very satisfactorily implemented up to 2009, but satellite inter-calibration problems from the SSM/I to the SSMIS instruments affect the time series starting in 2010. This issue will be documented and the work underway to solve the problem will be described.
Aires, F., L. Miolane, C. Prigent, B. Pham, E. Fluet-Chouinard, B. Lehner, and F. Papa (2017), A Global Dynamic Long-Term Inundation Extent Dataset at High Spatial Resolution Derived through Downscaling of Satellite Observations. J. Hydrometeor., 18, 1305–1325, https://doi.org/10.1175/JHM-D-16-0155.1.
Papa, F., C. Prigent, F. Aires, C. Jimenez, W. B. Rossow, and E. Matthews (2010), Interannual variability of surface water extent at the global scale, 1993–2004, J. Geophys. Res., 115, D12111, doi:10.1029/2009JD012674.
Prigent, C., F. Papa, F. Aires, W. B. Rossow, and E. Matthews (2007), Global inundation dynamics inferred from multiple satellite observations, 1993 – 2000,J. Geophys. Res., 112, D12107, doi:10.1029/2006JD007847.
Prigent, C., F. Papa, F. Aires, C. Jimenez, W. B. Rossow, and E. Matthews (2012), Changes in land surface water dynamics since the 1990s and relation to population pressure, Geophys. Res. Lett., 39, L08403, doi:10.1029/2012GL051276.
SAR based change detection for mapping changes in water table levels
Muro, Javier (1); Strauch, Adrian (1); Thonfeld, Frank (1,2) - 1: Center for Remote Sensing of and Surfaces, Germany; 2: Remote sensing Research Group
Wetlands are often coupled with human systems. That makes them highly dynamic cover types and thus difficult to map and monitor from space. The Sentinel-1 constellation allows us to monitor land changes regardless of cloud conditions and with a high frequency of pass. We used a time series of 33 Sentinel-1 dual pol images for 2016 to map changes in an artificial wetland in Kerkini, Greece. The images are accessed and processed in Google Earth Engine via a Flask web application (https://github.com/mortcanty/earthengine). This allows the processing of a large amount of images in a few minutes. Points of change within the time series are determined according to Conradsen et al (2016). Results show a clear gradient of areas in which water fluctuates more often, and during which periods. Areas where agriculture is active show as well high rates of change. This methodology can be used by wetland managers to monitor wetland dynamics as well as cover changes caused by human activities in an operational way.
Rheticus® Marine: Sentinel and Copernicus data for operative and continuous monitoring of coastal waters and resources
Ceriola, Giulio; Drimaco, Daniela - Planetek Italia s.r.l., Italy
Rheticus® Marine is an automatic cloud-based geo-information service designed by Planetek Italia to deliver fresh and accurate satellite-based data and information for the monitoring of coastal seawater quality and marine resources. It is based on satellite open data, such as the ones from AQUA/TERRA, Sentinel-2 and Sentinel-3 missions and from CMEMS (E.U. Copernicus Marine Environment Monitoring Service).
At European level the Marine Strategy Framework Directive (MSFD) requires Member States to reach Good Environmental Status (GES) through the evaluation and improvement of 11 qualitative Descriptors among which Eutrophication. Rheticus® Marine uses CMEMS derived historical series of water quality parameters (e.g. chlorophyll and transparency) to identify sea areas that are homogeneous in terms of eutrophic behaviour and so are eligible for the determination of the MSFD zones where to perform the assessment of the GES. The designed service is tailored according to the needs expressed by the Italian authorities responsible for the MSFD implementation and it has obtained a high successful feedback from them.
Another relevant sector for which Rheticus® Marine provides operational services is Aquaculture. By real time monitoring and forecasting of relevant water quality parameters (obtained from MODIS and Sentinel-3/OLCI sensors, integrating and improving CMEMS products) Rheticus® Marine supports the daily decision activities of Aquaculture farms. A pilot project is currently running in Greece.
A further application is the support to Desalination Plants: by combining Sentinel-2 and Sentinel-3 data,Rheticus® Marine can supply real time alerts to plants’ operators about the occurrence of algae blooms (AB) – together with other water quality parameters – in the coastal areas and in the proximity of the plant’s water intake. This allows users to take opportune timely decisions to avoid damages and/or interruptions of the plant operation. A Pilot project was successfully run in United Arab Emirates.
Within the EUGENIUS network (European Group of Enterprises for a Network of Information Using Space), a H2020 project that provides viable market based Earth Observation services in different European regions, Planetek Italia is responsible of the marine service portfolio consisting of the mentioned Rheticus® Marine services, entirely based on Sentinel/Copernicus data.
Estimation of Wind Speed over Lakes with Sentinel-1 – Limitations and Application Potential
Katona, Timea (1); Bartsch, Annett (2,3) - 1: TU Wien, Austria; 2: b.geos, Korneuburg, Austria; 3: Zentralanstalt für Meteorologie und Geodynamik, Austria
The Sentinel-1 satellite mission provides high-resolution, all-weather, day and night Synthetic Aperture Radar images. Studies about wind parameters above the ocean are frequent, inland lakes were seldom investigated.
The objective of this study was to evaluate the potential of C-band SAR for wind retrieval over lakes. A number of lakes from Austria and Hungary (lakes in Salzkammergut, Lake Neusiedl and Balaton), where wind measurements from automatic weather stations are available, has been selected for evaluation. Functions for speed retrieval are developed from empirical relationships. Meteorological station data are used for calibration and validation. The lakes were divided into groups representing three distinct regions, Mondsee, Traunsee and Attersee constitute the group of lakes of the mountainous Salzkammergut. They are comparably deep and have been formed during former glaciations. They are surrounded by mountains up to 2100 m high. The mountains and the opened water surface determine the possible routes for wind. They are expected to vary considerably between the lakes.
To obtain the wind maps, the functions were developed from the relationship of the main parameters and the backscatter. Functions for different incidence angles according to the wind speed were determined. The wind direction has an impact on the RMSE of the of the determined wind speed, but is limited for correlations and thus the variations reflected in the time series. This can be attributed to the prevailing wind directions at the analyzed lakes. Nevertheless, there is a good correlation (0.65 – 0.88 for stations located at the lakes) between the measured and the calculated wind speed. The relationship is impacted by the representativeness of the meteorological stations, not only due to the distance from the lake but also bending of the coastline. Spatial patterns of wind speed suggest that the currently available stations at the Austrian lakes are only representative for a proportion of the lakes, especially in the Salzkammergut. The retrievals from Sentinel-1 at VV polarization can help to identify these patterns, although multi-incidence angle capability is lacking. The availability of morning and evening acquisition in irregular intervals limit the application potential of SAR derived wind speed information. General patterns can be however derived.
GlobWetland Africa: Implementing Sustainable Earth Observation Based Wetland Monitoring Capacity in Africa and Beyond
Tottrup, Christian (1); Riffler, Michael (2); Sun, Yiwen (3); Stelzer, Kerstin (4); Kittel, Cecile (5); Wang, Tiejun (3); Grogan, Kenneth (1); Ludwig, Christina (2); Bauer-Gottwein, Peter (5); Blüthgen Sølvsteen, Jonas (1); Wevers, Jan (4); Odermatt, Daniel (6); Skidmore, Andrew (3); Verkedy, Zoltan (3); Walli, Andreas (2); Ouedraogo, Paul (7); Paganini, Marc (8) - 1: DHI GRAS, Denmark; 2: GeoVille, Austria; 3: ITC, University of Twente, The Netherlands; 4: Brockmann Consult GmbH, Germany; 5: DTU Environment, Denmark; 6: Odermatt & Brockmann GmbH, Switzerland; 7: Ramsar, Switzerland; 8: European Space Agency, ESRIN, Italy
Lack of data, appropriate information and challenges in human and institutional capacity put a serious constraint on effective monitoring and management of wetlands in Africa. Conventional data are often lacking in time or space, of poor quality or available at locations that are not necessarily representative for wetlands. Therefore, the Ramsar secretariat, the global coordinating body of the Ramsar Convention on Wetlands, has long recommended making more use of new and innovative technologies, such as those offered by remote sensing. Yet, access to suitable remote sensing data for monitoring wetlands in Africa has also traditionally been constrained either because of high costs or, especially in Equatorial Africa, owing to frequent cloud cover. To meet these challenges the European Space Agency has launched GlobWetland Africa as a major initiative to facilitate the exploitation of satellite observations for the conservation, wise use and effective management of wetlands in Africa and to provide African stakeholders with the necessary EO methods and tools to better fulfil their commitments and obligations towards the Ramsar Convention on Wetlands.
The main objective of GlobWetland Africa (GW-A) is to provide the major actors involved in the implementation of the Ramsar Convention of Wetlands in Africa with an open source software toolbox for mapping, modelling and monitoring the wetlands across Africa. The GlobWetland Africa toolbox unifies proven and stable open source software into a single graphical user interface that will enable the users to access and exploit the increasing capabilities of new and freely available satellite observations primarily from the Sentinel missions of the European Copernicus initiative. The toolbox has full end-to-end image processing capabilities for wetland delineation, wetland habitat mapping, monitoring of inundation regimes and water quality as well as for mangrove mapping and river basin hydrology assessments. This poster introduces the GlobWetland Africa toolbox and reviews the contribution of Sentinel data to support the generation of the GlobWetland Africa products portfolio.
Monitoring of water bodies areas and soil moisture content in risky areas of Bulgaria
Nikolov, Hristo Stoianov (1); Atanasova, Mila Stoyanova (2); Shishkov, Toma (3) - 1: Space research and technology institute - BAS, Bulgaria; 2: National Institute of Geophysics, Geodesy and Geography - BAS, Bulgaria; 3: NIKOLA PUSHKAROV” INSTITUTE OF SOIL SCIENCE AND AGROECOLOGY
Along with other techniques data from satellite based SAR instruments data have been used for extraction of soil moisture content (SMC) in agriculture and emergency cases to mention few. The relationship between the returned SAR signal and the amount of water in the soils at the small depths is well known to be in linear in form, but suffers from the fact that is highly dependent of the composition of soils investigated. This is the reason such investigations have to rely on additional laboratory and in-situ pedological data as well. The specific objectives of this research was to establish reliable correlation between SAR signal and the SMC for one specific region and to prove that it can provide information that could be used operationally by the local authorities as precursor for water bodies areas increase and their better mapping. This information obtained is mainly based on SAR data from Sentinel-1 satellite, but auxiliary data being satellite borne (e.g. high and mid resolution optical images, passive microwave measurements) or databases for soils properties from field investigations were used as well. This scenario implemented has several advantages since it makes use of the higher temporal and spatial resolution of the S-1 data compared to other SMC missions and weather conditions independency. Some limitations due to unknown soil roughness and uncertain vegetation cover at time of the radar signal acquisition had to be overcome during this research.
The region that was studied throughout of this research is located north west part of Bulgaria where in summer of 2014 heavy flooding occurred that was triggered in Copernicus emergency mapping service. This event was mainly attributed to fast water volume increase in one dams located in that flat agricultural area. For this incident some 60 maps for soil loss, flood, erosion and landslide assessment, evaluation of damages and reconstruction/recovery have been produced. Results from this research are in form of maps elaborated for late spring and mid-autumn seasons. As stated before the aim of the authors was to provide one more source of reliable information for prediction of imminent flooding events caused by increased volume of the water bodies present in the area considered.
Operational Monitoring of Water Bodies Areas as Precursor for Landslides Activities
Atanasova, Mila Stoyanova
Nikolov, Hristo Stoianov (1); Atanasova, Mila Stoyanova (2) - 1: Space research and technology institute - BAS, Bulgaria; 2: National Institute of Geophysics, Geodesy and Geography - BAS, Bulgaria
The impact of the water level of Danube river and its tributaries is recognized by the scientists and practitioners to be one of the key factors for landslides development in the north west part of Bulgaria. In this area widely distributed are Pliocene and sandy clays as well as clayey loess. This is the reason why along the Danube river significant amount of step-shaped landslides being wide up to 3 km are observed. The specific geological setting of clays changes its strength and physical indicators under the waterlogging conditions caused mainly by the increased amount of water. It is to be mentioned that in the area investigated from the town of Vidin to town of Nikopol there are number of landslides both active and stabilized forming one almost uninterruptible strip. Most of the landslides present there are at fragile equilibrium very often lost under the increase of the amount of surface and underground water which are consider being the major cause for this. It was confirmed that in the last 50 years in the mentioned region about 21 new landslides have occurred which increased the risk in it.
In the framework of this study we have investigated the possibility of using SAR data from Sentinel-1 mission of ESA to monitor the shape of water bodies and the level of soil moisture in order to use this information as indicator for possible landslides activities. Some additional information from geological and hydrological maps was used as well in order to complement the research. For some of the landslides present in the region interferometric maps were created in order to estimate the surface deformation and for the same dates water areas were delineated and soil moisture estimated. Presented are the results of their combined use and discussed is the possible impact on the population, infrastructure and agricultural areas.
Database for Hydrological Time Series of Inland Waters (DAHITI)
Schwatke, Christian; Dettmering, Denise - DGFI-TUM, Germany
Satellite altimetry was designed for ocean applications. However, since some years, satellite altimetry is also used over inland water to estimate water level time series of lakes, reservoirs, rivers and wetlands. The resulting water level time series can help to understand the water cycle of system Earth and makes altimetry to a very useful instrument for hydrological applications.
In this poster, we introduce the "Database for Hydrological Time Series of Inland Waters" (DAHITI). DAHITI provides time series of water level heights of inland water bodies and their formal errors. These time series are available for the period 1992-2017 and have varying temporal resolutions depending on the data coverage of the investigated water body. A significant number of stations is available in near real-time with a latency of a few days. Currently, the database contains about 650 water level time series of lakes, reservoirs, rivers, and wetlands which are freely available after a short registration process via http://dahiti.dgfi.tum.de. The accuracies of the water level time series depend mainly on the extent of the investigated water body and the quality of the altimeter measurements. Hereby, an external validation with in-situ data reveals RMS differences between 5 cm and 40 cm for lakes and 10 cm and 140 cm for rivers, respectively.
In this poster, we introduce the functionality of the DAHITI web service as well as the available DAHITI products. Furthermore, selected examples of inland water targets are presented in detail.
SURF-WATER : A High-Resolution and Near Real Time Monitoring of Surface Water Extent using a Multi-Sensor & Multi-Temporal Approach
Peña Luque, Santiago (1); Pedinotti, Vanessa (2); Hagolle, Olivier (3); Andral, Alice (1) - 1: CNES, 18 avenue Edouard Belin , SI/2A Team, 31401 Toulouse cedex 9, France; 2: Magellium, 24 Rue Hermès, 31520 Ramonville Saint-Agne, Toulouse; 3: CESBIO, 18 avenue Edouard Belin ,31401 Toulouse cedex 9, France
Water detection is a complex task and various methods have been developed to extract water bodies from different remote sensing images. Many hydrological applications such as climate impact studies or flood forecasting require near real time information about where water is and how surface water is evolving.
In that context and in the frame of the SWOT (Surface Water and Ocean Topography) mission and the SWOT preparatory program, CNES with CESBIO proposes a new tool of water detection based on the combination of Sentinel 1 and 2 images.
The water detection is computed from the Sentinel 2A images with the contribution of Sentinel 1 images to enhance temporal resolution and to avoid potential limitations due to cloud conditions. It will lead to a near real time and periodical (biweekly, monthly) water bodies product and a water variations product.
The optical algorithm is inspired by the multi-temporal water detection method included in MAJA (MACCS-ATCOR joint algorithms) software used to obtain the Sentinel 2 Level 2A product at 20 m resolution. The multi temporal approach enables to separate water bodies from cloud shadows. Different spectral indexes are used to extract water pixels: NDVI, MNDWI, SWIR, red band to avoid snow and cloud/shadow image from the MAJA software.
The processing of the GRD Sentinel 1 images is done using the pre-processing CNES-CESBIO chain based on the open source OTB (Orfeo Tool Box https://www.orfeo-toolbox.org ). It gives a binary water/ no water image that is fully superimposed with a S2-A image. Both masks are then merged in a single one using a very simple approach.
The validation of the algorithm over different areas (Camargue, Mekong river, small rivers in France) shows a good ability in catching temporal and spatial evolution of surface water extent and the significant contribution of radar imagery in very cloudy regions. Further investigations are needed to improve the method especially over turbid water and a validation plan is planned on several other regions around the world.
Mapping of water permanence and fluctuations for updating the Ramsar Information Sheets using optical and radar data: A case study for two Greek Ramsar Sites and their catchments.
Fitoka, Eleni (1); Apostolakis, Antonis (1); Truckenbrodt, John (2); Tompoulidou, Maria (1) - 1: Greek Biotope Wetland Centre, Greece; 2: Friedrich-Schiller-University Jena Institute of Geography Department for Earth Observation
Provision of a suitable map or maps is a requirement under Article 2.1 of the Ramsar Convention – it is fundamental to the process of designating a Wetland of International Importance (Ramsar Site), and is an essential part of the information supplied in the Information Sheet on Ramsar Wetlands (RIS). Mapping of the key hydrological features and their seasonal variations, and in particular water permanence and fluctuations, is needed for describing physical and ecological features of wetland sites, including assessment of ecosystem services, as well as for improving the delineation of their boundaries inside and outside of the protected areas. Such maps are often absent or not updated for most of the Ramsar designated sites worldwide. In the current study, which is developed within the frame of Satellite-based Wetland Observation Service (SWOS) project, the proposed Surface Water Dynamic products, based on radar or optical satellite data, were tested at an area of 665.000 ha. Itrepresents the sub catchment areas of the two neighbouring Ramsar sites “Nestos and adjoining lagoons GR56RIS” and “Vistonida and Ismarida lakes and adjoining lagoons GR55RIS” located in Eastern Makedonia and Thrace region of Greece. Water detection techniques and indices (NDWI, MNDWI) were applied on multi temporal optical data (Landsat, Sentinel2) applying pixel based and object-based classification approaches. SAR time series analysis used the K-means algorithm to cluster the backscatter series of individual pixels in order to separate land and water occurrences purely in the temporal domain. Results from optical and radar multi temporal analysis were reclassified based on Temporal Frequency (TF) of water surface occurrences into four classes: permanent open water (TF>85%), seasonally open water (50%
Development of a Concept for Correcting Artifacts in Sunglint Originating from the Sentinel 2 Acquisition Mode.
Riedel, Sebastian (1,2); Gege, Peter (2); Oppelt, Natascha (1) - 1: Kiel University; 2: German Aerospace Center (DLR)
The correction of sunglint is a prerequisite for a successful quantitative evaluation of water remote sensing data. Insufficient correction of sunglint may be a limiting factor in the accuracy of derived products. Sunglint can exceed the water leaving radiance multiple times, and even in remote sensing images with no obvious sunglint patterns, small amounts of sunglint may be present. The pixel size of Sentinel 2 is much larger than the wave facets causing specular reflection of the sun. Since the effective sunglint of a pixel depends on the number of wave facets that reflect direct sunlight towards the sensor, the sunglint pattern visible in remote sensing images is therefore dependent on the spatial resolution of the sensor. Due to the inter-sensor and inter-band along-track displacement of the MultiSpectral Instrument, all Sentinel 2 bands in the calibrated image have a temporal offset which ranges from 234 ms to 6.9 s. Since the sunglint flash duration is mostly in the range of 1 to 100 ms, the sunglint contribution to different bands of one pixel is highly randomized. This randomization results in a highly variable sunglint pattern for each band and a sunglint offset which depends on the density of sunglint producing wave facets and their large-scale patterns. Established methods, which detect sunglint at a SWIR band and extrapolate to the entire spectrum, are able to correct the sunglint offset in Sentinel 2 scenes. However, sunglint patterns larger than the Sentinel 2 resolution cannot be corrected with established methods. We are currently developing a new method to identify and correct effects from spectrally randomized sunglint on a pixel by pixel basis. This new method may be applied after atmospheric correction either to the sunglint affected image or to remove residual sunglint after correction with an established method.
Monitoring of Climate Change Effects on Water Surface Temperature of the Bracciano Lake
De Santis, Davide; Del Frate, Fabio - University of Rome "Tor Vergata", Italy
The last three years were the hottest on global scale since historical measurements are available. In particular, in 2016 in Italy a temperature anomaly of +1.35 °C above the mean over the period 1961-1990 has been registered. On a regional scale, climate change is causing the temperature increase of coastal and inland waters. In the case of lakes, in particular, the water warming is causing rise in algal blooms frequency and in percentage of water lost through evaporation.
A case study to monitor lake warming over the last decade using satellites data has been the analysis of the water surface temperature trend of the Bracciano Lake. This water body is one of the most important natural drinking water reservoirs of the Rome Province area and it does not have a suitable network to monitor lake's parameters. As every deep lake, the Bracciano Lake is characterized by a greater thermal stratification stability in case of a water temperature increase. This phenomenon causes the depth decrease of the mixing layer and the lake is subject to a higher risk of eutrophication and a worsening in water quality.
In order to achieve the aim of the work, more than 200 satellite imagery have been elaborated: the data were acquired from Landsat 5, Landsat 7 and Landsat 8 from 2000 to 2016. The temperature of the water surface has been obtained through a radiative transfer equation-based algorithm (1). Temperature values from satellite data in comparison with in situ recorded data had shown a very good accuracy of the algorithm. Analysing the result has been possible to estimate that in 2016 the average temperature of the Bracciano Lake surface was 18.40 °C (+0.72 °C over the mean of the period 2000-2016). In particular, in 2016 the maximum water surface temperature was 26.86 °C in July and the minimum was 10.58 °C in February (respectively +1.32 °C and +0.51 °C over the mean of the period 2000-2016). Almost all the average values over the temporal ranges analysed (consisting of two or three months) have shown a general rising trend of in water surface temperature.
The data fusion with recent Sentinel 3 imagery, taking into account the different spatial resolutions, is currently considered to improve the temporal resolution of the analysis.
(1) Xiaolei Yu, Xulin Guo and Zhaocong Wu; Land Surface Temperature Retrieval from Landsat 8 TIRS—Comparison between Radiative Transfer Equation-Based Method, Split Window Algorithm and Single Channel Method
The influence of the incidence angle over the backscattering values for SAR-based flood mapping
Zhao, Jie (1,2); Chini, Marco (1); Pelich, Ramona (1); Hostache, Renaud (1); Matgen, Patrick (1); Wagner, Wolfgang (2) - 1: Luxembourg Institute of Science and Technology, Luxembourg; 2: Technische Universität Wien
Synthetic Aperture Radar (SAR) is the desired system for water bodies mapping from space since SAR sensors can acquire data of equal quality day-and-night and almost regardless of weather conditions. Moreover, SAR imagery has been often successfully used in mapping flood areas based on the fact that the backscatter from water bodies has a rather unique signature allowing them to be separated from other land cover classes relatively easily.
The Hierarchical Split-based (HSBA) algorithm introduced by Chini et al (2017) makes use of the hierarchical image tiling, histogram thresholding, region growing and change detection to delineate the extent of floodwater. It assumes that the distributions of water and non-water in a SAR image are two different Gaussian distributions which are mixed into a bimodal distribution. However, the flood areas in the SAR image usually represent a small portion of the whole scene and hierarchical image tiling is necessary in order to parameterize the two distinctive Gaussian distributions at a local scale. In HSBA, a bimodality test using the Ashman D coefficient, the Bhattacharyya coefficient and surface ration is first carried out in order to identify images subtiles where the histogram is bimodal. Then thresholding region growing and (optionally) change detection are used to separate the water from the background in order to identify the flooded areas.
This algorithm has been implemented on the Grid Processing On Demand environment (G-POD) of the European Space Agency (ESA) and can be applied Sentinel-1 data and Envisat ASAR data. Due to the difficulty to automatically define a relevant reference image for any SAR dataset, the single image-based HSBA without change detection has been applied to the complete data archive of the Envisat ASAR mission over Europe. This allowed us to generate a historical flood record. The evaluation of the results reveals that the generated dataset tends to over-estimate the extent of flooding, especially in densely vegetated areas that have potentially backscattering values close to those of open water. Two options are envisaged in this study to reduce this type of error: i) by integrating change detection into the processing chain and ii) by taking into account the influence on the backscatter of variations of the incidence angle within each SAR scene.
In theory, in any given SAR image the backscatter decreases with the increase of the local incidence angle. Consequently, it is not advisable to define a unique bimodal distribution function for the entire scene considering the differences in the water backscattering values between near- and far-range angles. Therefore, we assume that the HSBA should be applied to subsets of the original image based on the pre-defined ranges of incidence angles. The relationship between incidence angle and backscatter of water has been investigated and different sampling strategies were adopted to apply the algorithm on sub-regions where the variations of incidence angles are limited constant.
Test have been done on several images acquired over the Severn River in the UK and Danube Delta in Romania and more tests will be done in the future.
Comparing Existing Lake Databases in the SWOT Context
Cazals, Cécile (1); Pottier, Claire (2); Yésou, Hervé (3) - 1: C-S, 5 rue Brindejonc des Moulinais, 31500 Toulouse; 2: CNES, 18 avenue Edouard Belin, 31400 Toulouse; 3: ICube - SERTIT, 300 Bd Sébastien Brant, 67412 Illkirch-Graffenstaden, France
The SWOT mission (Surface Water and Ocean Topography) is an ambitious and innovative project aiming at mapping ocean and inland water levels from space. Over land, SWOT is designed to observe nearly all surface water bodies including small lakes and reservoirs (surface > 250x250 m2 with a goal of 100x100 m2), and rivers (width > 100m with a goal of 50 m). Products specific to rivers and lakes will link SWOT observations to known water bodies; therefore, global a priori river and lake databases are needed. This study is focused on lake mapping.
In order to build a dedicated lake database from existing ones, comparison is essential to select the most appropriate solution. In the SWOT context, the main constraints are:
• Global cover, ideally from 78° South to 78° North. It discards every national or continental database like Corine Land Cover (European Environment Agency) or National Land Cover Database 2011 (U.S. Geological Survey).
• The ability to separate lakes from rivers to select lakes only. Therefore, raster water masks without semantic information are discarded. This constraint excludes most of water bodies databases such as G3WBM [Yamazaki et al., 2015], Global Inland Water [Feng et al., 2000] or GLOWABO [Verpoorter et al., 2014] and also global land covers like Globeland30 [Cheng et al., 2015] of FROM-GLC (Tsinghua University).
• Completeness: one of the SWOT lake products is generated only for lakes in the a priori database, i.e., if an existing lake is missing in this database, the associated SWOT product will not be generated. Therefore, the reference database needs to be as exhaustive as possible,
• The minimum element size of the database is approximately 1 ha (i.e. SWOT goal).
Among the large number of available global water databases, only a few meet these criteria and have a free data distribution policy. In this study, we focus on three of them: the inland water layer of Natural Earth Data (NED, Free vector and raster map data @ naturalearthdata.com.), the Global Lakes and Wetlands Database (GLWD) from [Lehner and Döll, 2004], and water and reservoirs of the OpenStreetMap database (OSM).
As SWOT will provide global mapping (up to 78° latitude), a global statistical analysis is performed. Because of their particular characteristics [Downing et al., 2006], the comparison is carried out by continent. In order to compute accurate surface areas over lake entities, databases are cropped and projected in the appropriate UTM zone.
Two main statistical analyses are conducted: completeness and assessment of accuracy. The completeness of databases cannot be studied directly because no reference data exists at global scale, so a statistical comparison of databases is performed. In order to assess their accuracy, candidates are compared to the Pekel database [Pekel et al., 2016]. It has been chosen because of its relatively high assessed accuracy, and its well suited resolution to our study.
Mapping Submerged Aquatic Vegetation by Using a Sentinel-2A Time Series at Lake Starnberg (Germany)
Fritz, Christine (1); Schneider, Thomas (1); Dörnhöfer, Katja (2); Oppelt, Natascha (2) - 1: TU München, Germany; 2: CAU Kiel
Submerged aquatic vegetation (SAV) provides detailed information about the ecology of fresh water ecosystems (Melzer, 1999). SAV is a suitable and often used indicator for trophic state assessment as it is sensitive to varying nutrient conditions, water temperature, water level and water transparency. Changes in the trophic state induces variations in plant species composition, expansion, vegetation begin and senescence. To monitor the trophic state as prescribed by the European Water Framework Directive (WFD), SAV should be mapped every three years (European Commission, 2000). To detect changes in the trophic state and the water quality at an early stage, more frequent observations are required. Remote sensing methods offer a time- and cost-effective potential to observe seasonal and annual changes in water quality and SAV coverage. Remote sensing systems of our days are well suited to complement regular in-situ samplings to detect changes in sediment and SAV covered areas. The remote sensing approach is expected to close observation gaps between snapshots of single in-situ mappings. The high revisiting frequency and broad coverage of remote sensing data may compensate the reduced information on species compared to mappings. This study analyses and compares the results of littoral bottom coverage determination by two different methodological approaches: a semi-empirical method using depth-invariant indices after Lyzenga (1978, 1981) and a physically based, bio-optical method as implemented in the Water Simulator (WASI-2D, Gege 2013). As basis for this comparison, we analyzed the very first four appropriated Sentinel-2A scenes from 2015 of Lake Starnberg supported by in situ measurements based spectral signatures of lake bottom sediments and submerged aquatic vegetation high and low growing species. Within the growing season between August and September 2015, both methods allowed to distinguish vegetated and non-vegetated patches in shallow water areas. Furthermore, both methods allowed further differentiating vegetated areas in tall- and meadow-growing SAV species dominated.
Sentinel-2 and MODIS Land Surface Temperature Based Evapotranspiration for Computing Irrigation Efficiency
Kyalo, Daniel Kiluu (1); Zoltan, Vekerdy (2); Velde, Rogier van der (2); Odongo, Vincent Omondi (3,4) - 1: Machakos county, Kenya; 2: University of Twente, ITC; 3: Egerton University, Kenya; 4: Wageningen University
Sentinel-2 data provides reasonably high spatial-temporal resolution, making it ideal for use in estimating irrigation efficiency in highly heterogeneous landscapes. In this study, we employ the use of Sentinel-2 data to downscale MODIS Land Surface Temperature (LST) to a resolution of 10 m for estimation of irrigation efficiencies in open irrigation systems in Lake Naivasha basin. The intensification of irrigation activities, land use, and land cover changes have been postulated as some of the possible causes of fluctuations of the lake levels. The combined use of Sentinel-2 NDVI, thermal sharpened Sentinel-2 based MODIS LST data and other meteorological data were inputs in SEBS to derive high-resolution daily evapotranspiration maps, which were used to estimate the irrigation efficiencies of open field systems. Thermal sharpening is based on the assumption that land surface temperature is proportional to NDVI, and was considered suitable for downscaling LST because it depicts the presence/health of vegetation and consequently reflects the evapotranspiration potential of an area. Results show that the downscaled LST maps depict the landscape heterogeneity better than the original MODIS LST. A comparison of the aridity index to the irrigation efficiency showed that the efficiencies were generally high when the aridity index was low and seemed to decrease with an increase in aridity index. This disparity was indicative of lack of accounting for soil moisture in irrigation scheduling, and thus the tendency to over-irrigate during the wet season.
Global Lake Water Products within the Copernicus Global Land Service
Stelzer, Kerstin (1); Simis, Stefan (2); Brockmann, Carsten (1); Carrera, Laura (3); Steinmetz, Francois (4) - 1: Brockmann Consult GmbH, Germany; 2: PML, UK; 3: University of Reading; 4: HYGEOS
The Lake Water products within the Copernicus Global Land Service provide an optical and thermal characterization of nominally 1000 inland water bodies that belong to the world’s largest (according to the Global Lakes and Wetlands Database, GLWD) or are otherwise of specific environmental monitoring interest. The products contain four (sets) of parameters: lake water surface temperature, lake water reflectance (all wavebands that are available after atmospheric correction), turbidity (derived from suspended solids concentration estimates) and a trophic state index (derived from phytoplankton biomass by proxy of chlorophyll-a). Production and delivery of the parameters are over set 10-day intervals and mapped to a common global spatial grid at either nominally 1000m (~0.008°) or 300m (~0.0022°) resolution. The development of a 100m product for the water quality parameters is currently ongoing. LSWT is provided in 1km resolution only. The algorithms used to derive the input for the optical lake water products are implemented in the Calimnos processing chain and were tuned and validated against 13 predefined optical water types in the NERC (UK) GloboLakes project. The LSWT retrieval is based on developments from ESA SST-CCI and Globlakes. The products cover the timespan 2002 – 2012 (MERIS and AATSR lifetime) and are available as an NRT service derived from Sentinel-3 OLCI and SLSTR and later on from Sentinel-2 MSI. The products will be publicly available via the Copernicus Global Land Service Website and Viewing Services.
Glacial Lake Dynamics of Eastern and Western Himalaya: possible signature of climatic variability over the region
Goswami, Ajanta - IIT ROORKEE, India
On June 16, 2013 the whole southeast Asia was jolted by one of the largest glacial disater when the entire township of Kedarnath and Rambara was washed away by the outburst of Chorabari lake. The damage caused downstream was even more severe. This jolted the scientific community of the region and a necessity was realised to understand the dynamics of the glacial lakes of the region and its possible behaviour with climate change. This could help to possibly avert the risk.
Hence, to fill this scientific gap glacial lakes of Eastern and Western Himalaya were mapped and monitored on decadel scale using Corona, Survey of India Toposheets, Landsat and Sentinel data from 1970's to recent.
The dynamism involved with the glacial lakes were studied in detail using temporal data. A comparison of the glacial lakes of eastern and western Himalaya was done and it was observed that there is a striking difference in its dynamism and response to climate change. Hence to find the possible signature of climate change on glacial lakes, it was investigated further. The results indicates that the glacial lakes in eastern Himalaya are growing faster compared to Western Himalaya. The find the possible reason behind this different growth rate, the climatic pattern over the two region was studied. The result showed clear indication of striking climate variability in the two region. The western Himalaya is influenced by Karakorum anomaly because of which glaciers in this region are advancing in many parts and lakes are receding.
The present study has brought some significant results which shows clear indication of variable climate pattern in this Himalayan region and its impact on glacial lakes which can act as a proxy to identify climate change.
Copernicus Downstream Services for the Mapping of Water Bodies in Europe: the Eugenius Approach
Tholey, Nadine; Studer, Mathias; Maxant, Jérôme; Caspard, Mathilde; Yésou, Hervé; de Fraipont, Paul - ICube / SERTIT, Université de Strasbourg, 300 bd Sébastien Brant, CS 10413, 67412 Illkirch Cedex, France
Within the framework of the EUGENIUS (European Group of Enterprises for a Network of Information Using Space) H2020 project , an European network of geo-information service providers is delivering Earth Observation derived services to the regional and local European market; based on the combined exploitation of Copernicus data (especially Sentinel-1 and Sentinel-2) and local data, a common products catalog offer has been set up.
In a nutshell, EUGENIUS objectives are to develop viable market based Earth Observation services to be supplied in different European regions, accessible through a network of GIS platforms installed in each region, to industrialize applicative tools, to demonstrate the economic sustainability of the concept and to prepare the network expansion while also involving end-users in the whole life cycle of the project.
EUGENIUS applicative tools are operationally used earth observation services covering several thematic domains: urbanization growth, agriculture, forest, water quality (littoral zones) and natural risks (especially geo-physical and hydrological risks) monitoring. Within this service offer, the Flood Mapping and Monitoring Service, is a regional service dedicated to flood, water / wetlands , river basin, environmental, and land planning management actors as well as local authorities, states services or the insurance sector. Based on Earth Observation data exploitation, this service allows to collect geo-information and maps related to historical hydrological events or on-going regional (plain) floods which are not mapped by the Copernicus Emergency Mapping Service (EMS, a Copernicus core service). During the event, the (successive) instantaneous extent of the flood observed at the time of satellite acquisition, associated to the flood impact on the landscape can be delivered shortly after EO data reception After the event, flood synthesis geo-information products and statistics (eg: maximum extent and impact, flood duration) may be generated and delivered. Geo-information related to past events may also be requested through the exploitation of available archive satellite data.
In a first ramp up phase, the Flood Service is progressively put in place over the French Grand Est Region (ie formerly Alsace, Lorraine, Champagne-Ardennes regions). The 2016 January-February flood of the Ill River in the Alsace Plain was mapped and based on successive S1 data acquired between the 26th of January and 27th of February 2016. At this latitude, combing descending and ascending tracks, the acquisition can be daily over a 4 days period, with an average of 1.5 days over one month. Obtained results confirmed the capability of Sentinel-1 to monitor flood event of local/regional importance, thanks to its high resolution and high temporal revisit, which opens the door to an operational service. Moreover, flood related geo products, such as the occurrence map, highlight also the most sensitive areas. More recently, in January 2018 the Service was extented at the French Grand Est region scale, being triggered to follow the large inundations involved by the intense rain falls following the Eleanor storm. The flooded areas along the Moselle, Meuse, Nied, Sarre, Ill, Sauer have been mapped from Sentinel-1 imagery and results provided to the Users community through the Eugenius regional platform.
The Sentinels in Support of Water Transfer Mapping on Wetlands and Meadows
Mleczko, Magdalena (1); Mróz, Marek (1); Fitrzyk, Magdalena (2) - 1: Institute of Geodesy; Faculty of Geodesy, Geospatial and Civil Engineering; University of Warmia and Mazury in Olsztyn; Poland; 2: RSAC c/o ESA-ESRIN; Frascati, Italy
Nowadays data from Sentinel-1A/B became an important tool for flood monitoring, water bodies and wetlands mapping in local and regional scale. The biggest advantages of Sentinel-1 data are: clouds independence, large imaging swath, high frequency of revisits and data availability. Mapping permanent water bodies or temporary flooded areas with SAR sensors like Sentinel-1A/B is based on the detection of smooth surfaces (calm water surface) exhibiting very low backscatter due to the predominant specular reflection of microwaves. Great effort has been recently made by researchers developing new algorithms for the automatic thresholding of backscatter values, allowing the separation of water / non-water areas with high reliability. The main difficulty in open water mapping with SAR stems from the fact that wind induced roughness of water surface can strongly increase the backscattered signal. This could be misleading in thresholding and in the classification processes. Horizontal (HH) and cross-polarized (HV or VH) backscattered signals are more resistant to surface roughness than the vertical (VV) one. However, VH and HV components are characterized by low signal-to-noise ratio (SNR). The polarimetric configuration VV/VH of Sentinel-1A/B sensors is therefore not optimal for water surface detection and mapping in windy conditions. Despite this weakness Sentinel-1 turns out to be efficient in mapping shallow open waters on wetlands during seasonal floods. Very frequent observations with a short revisit time comparing to the low velocity of water mass transfer permit to eliminate from the time series images disturbed by windy conditions. Moreover, Sentinel-1 proves well in distinguishing partially flooded short vegetation (mainly grasslands) basing on the double bounce effect which results in low values of the Shannon Entropy. In this study we present results from the Biebrza Wetlands located in the North-East of Poland where backscatter thresholding together with polarimetric decomposition of Sentinel-1 data were successfully used to map stable open waters and seasonally flooded grasslands and meadows. Information on the water extent in the river valley during spring and summer seasons is crucial not only for eco-hydrological investigations, but also for more efficient management of the areas where limited agricultural activity is allowed. Results obtained in this study from Sentinel-1 data were validated by the measurements of water level using piezometers mounted within the area of research. Some cloud-free Sentinel-2 images were also used to validate the results with spectral indices like NDVI and NDWI.
Mapping of small water bodies in the Cape Winelands region of South Africa using Sentinel 1 and 2
Bangira, Tsitsi (1,2); van Niekerk, Adriaan (2); Iannini, Lorenzo (1); Menenti, Massimo (1,3); Vekerdy, Zoltán (4,5) - 1: Delft University of Technology, Department of Geoscience and Remote Sensing, P.O. Box 5048, 2600 GA Delft, The Netherlands; 2: Stellenbosch University, Department of Geography and Environmental Studies, Private Bag X1, Matieland, 7602, South Africa; 3: State Key Laboratory of Remote Sensing Science, Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing 100101, China; 4: University of Twente, Faculty of Geo-Information Science and Earth Observation (ITC), PO Box 217, 7500 AE, Enschede, The Netherlands; 5: Szent István University, Department of Water Management, Páter Károly u. 1., 2100 Gödöllő, Hungary
Small water bodies play an important role in agricultural and industrial production and are vulnerable to climate variability. Accurate surface water body mapping can provide crucial information for water security, flood mapping and water resources management. The recently launched Sentinel 1 and Sentinel 2 satellites can provide fine spatial resolution radar and multispectral images, respectively. This new dataset is potentially of significance for regional water body mapping and monitoring due to its free accessibility, spatial resolution and frequent revisit capabilities. The use of remote sensing techniques for mapping of clear open water bodies is easy and straight forward, however surface water does not only consist of open and clear water. Vegetation, sediments and dissolved substances can have significant impact on the observed signal. Either SAR or multispectral remote sensing is associated with certain weaknesses when delineating surface water bodies in complex environments. In order to take advantages of both data, a combination of SAR and multispectral data was used to delineate optically complex water bodies in the Cape Winelands district of South Africa. Machine learning and thresholding techniques were evaluated in the study. Machine learning algorithms namely support vector machines, k-nearest neighbour, decision trees and random forest were applied for classification of turbid, shallow, sedimented and eutrophicated water bodies. The proposed method is fully automated and represents a feasible way to show a change detection in heterogeneous water bodies. The method was validated using GPS surveys. This study, showed that different waters (e.g. turbid, sedimented, vegetated and eutrophicated) have different spectral properties which affect classification accuracies. SVM delineated different water bodies with better accuracy as compared to thresholding and other machine learning techniques. The study was done in the framework of the European Space Agency TIGER and ALCANTARA programs.
Keywords: water body, machine learning, thresholding, water security, Sentinel-1, Sentinel -2
09:00 - 10:40
Mapping of River Bodies and Ice Cover with Sentinel-1
Weintrit, Beata; Kubicki, Michał - Astri Polska Sp. z o.o., Poland
River icing is typical phenomenon in Poland during the winter season and can last from 1 to 3 months. River ice can cause floods through the blockage of water flow in the riverbed. In addition, icing affects the transportation function of the river. Observation of river ice is part of water management and presently is conducted through network of in-situ ice observation on limited number of rivers. Space-based observation using remote sensing techniques can provide a viable alternative and support institutions with continuous data for entire river sections. The proposed ice monitoring service is aimed to provide regular information on ice events with spatial resolution high enough to allow detection of local ice cover.
The detection of ice is based on Sentinel-1A and Sentinel-1B microwave satellites. Application of Sentinel radar sensors allows to collect image, independently from weather conditions (excluding intensive storms and snow falls), every 2-7 days, and process it in near real time. Microwave reflection of snow (including ice covered by snow), ice and water is significantly different, as long as the ice is not covered with thick layer of stable water. Research was conducted on the selected test sections of the Odra, Vistula and San Rivers in Poland. The spatial resolution of Sentinel-1 satellites (5x20 m in Interferometric Wide Swath mode) capabilities for effective ice detection on rivers was determined over 60 meters wide. As a result of the conducted research works, the fully automatic ice and water coverage classification of riverbed is developed.
River Ice Monitoring Service is being developed to provide regular, spatially continuous information about ice events on rivers during winter season. Data is presented as spatial layer representing several type of ice/water cover and as service allows also to generate statistical report relating to 1 km length sections of the river through web service. The research was conducted as part of the EO4EP project ‑ “Earth Observation for Eastern Partnership”, financed and supported by ESA.
Mapping Small Reservoirs in Semiarid Environment Using Multitemporal Synthetic Aperture Radar Data
Amitrano, Donato (1); Di Martino, Gerardo (1); Iodice, Antonio (1); Mitidieri, Francesco (2); Papa, Maria Nicolina (2); Riccio, Daniele (1); Ruello, Giuseppe (1) - 1: University of Napoli Federico II, Napoli, Italy; 2: University of Salerno, Fisciano, Italy
Information about the extension of water bodies is fundamental for the correct management of water resources. This is true especially in semiarid environment, where small reservoirs constitute a key aspect for the wellness of local population, in particular in rural areas.
Extraction of small reservoirs from remote sensing images is usually implemented using multispectral/optical sensors. In fact, they allow for the exploitation of simple information extraction techniques based on radiometric indices, very popular in the end-user community, whose amplitude is related to some physical characteristics of the feature they refer to. Dealing with water surfaces, the normalized difference water index (NDWI)  is probably the most common solution. However, in semi-arid environment, the main limitation of this approach is the cloud coverage, which is very probable during the wet season.
The use synthetic aperture radar (SAR) systems, thanks to all-weather and all-time imaging capabilities, allows for overcoming this problem, provided that user-friendly techniques for data analysis are made available to the end-user community, in which the expertise with radar data could be limited.
We propose an integrated framework for water resources management in semi-arid environment based on the joint exploitation of SAR data and hydrological models. In particular, SAR images are used to delineate (with high temporal resolution) the extension of small reservoirs in rural areas using innovative end-user-oriented time-series processing . Starting from reservoirs, several applications can be implemented. As an example, coupling reservoirs’ surface mask with a digital elevation model it is possible to estimate the retained volume, and to make forecasts about future water availability through calibrated hydrological models .
The purpose of this work is to provide end-users with a new tool for continuous and cost-effective water resources management and monitoring in semiarid regions.
 S. K. McFeeters, “The use of the Normalized Difference Water Index (NDWI) in the delineation of open water features,” Int. J. Remote Sens., vol. 17, no. 7, pp. 1425–1432, May 1996.
 D. Amitrano, G. Di Martino, A. Iodice, D. Riccio, and G. Ruello, “Small Reservoirs Extraction in Semiarid Regions Using Multitemporal Synthetic Aperture Radar Images,” IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., vol. 10, no. 8, pp. 3482–3492, 2017.
 D. Amitrano, F. Ciervo, G. Di Martino, M. N. Papa, A. Iodice, Y. Koussoube, F. Mitidieri, D. Riccio, and G. Ruello, “Modeling Watershed Response in Semiarid Regions With High-Resolution Synthetic Aperture Radars,” IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., vol. 7, no. 7, pp. 2732–2745, 2014.
Remote Sensing Estimation and Analysis of Lake Water Storage in Tibetan Plateau
Lu, Shanlong - Institute of Remote Sensing and Digital Earth, CAS, People's Republic of China
Due to the limitation of the use of data sources and technical methods, past research lacks awareness of lake water storages and their changes throughout the Tibetan Plateau, and cannot quantitatively describe the role of lake water bodies in the plateau area and their changes in the whole terrestrial water system and its changes. By using multi-spectral remote sensing images, laser and radar satellite altimeter data and digital elevation model (DEM) as data sources, this study analyze the spatial distribution pattern of lake water storage in the plateau area in the past 20 years and clarifies the change rule, through multi-source data synergy, multi-model synthesis and temporal and spatial characteristics analysis. The goal of the study is to answer the following two scientific questions: (1) how much water is stored in the plateau lakes and the proportion of lake water reserves in different areas to total water reserves? (2) what are the change trends of the lake water storages in different regions and different sizes, how much are the amount of changes, and what are the contributions to changes in total land water reserves in the plateau? The research results will help us to understand the temporal and spatial variation of lake water and its role in regional water system under the background of global warming, so as to provide scientific support for the development of water resources application and protection strategy in plateau area.
Mapping Sea Surface Dynamics In The Context Of Displaced Persons - The Case Of Rohingya Refugees In Bangladesh
Braun, Andreas; Hochschild, Volker - University of Tübingen, Germany
The influx of approximately 650.000 refugees from Myanmar into Bangladesh since August 2017 resulted in massive overcrowding of the reception centres of Kutupalong and Nayapara located approximately 30 kilometers south of Cox's Bazar in the division of Chittagong in South-Eastern Bangladesh.
In response to these developments, the government considered to relocate 100.000 of these refugees on the island of Thengar Char which has recently developed approximately 25 kilometers from the coast in the Bay of Bengal. Humanitarian organizations however raised highest concerns that this island provides habitable and safe conditions to these people.
This study investigates the two islands, Thengar Char and Bashan Char, based on Sentinel-1 radar data in order to give information on their spatial extent over a period of three years in order to demonstrate how large the land area is subject to changes. A total number of 62 images at intervals between 6 and 72 days are used according to data availability provided by Copernicus.
All images were radiometrically calibrated and geocoded. Due to the lack of topographic data over the sea, an average height Range Doppler approach was applied, leading to small shifts in geolocation. Coregistration between the images was found sufficient however. A Lee Sigmal filter was applied to reduce speckle.
Because water bodies appear dark on the images and the sandy islands are bright, the processed images give a first insight on the size of the islands for the given times. For visual comparison, mean, minimum, maximum and standard deviation images were calculated on the stack. These highight the range of variation in size of the islands, as well as the areas which are subject to change. RGB composites of different dates also serve for visual interpretation.
In a second step, an adaptive thresholding (based on Otsu's method) was applied to the images in order to get binary masks of land and water areas for each image. They show that the extent of the islands range between 4.000 and 7.600 hectares (Bashan Char) and between 1.650 and 2.500 hectares (Thengar Char) respectively.
On the methodological side, we compare the processing of the data with ESA SNAP, the RSS Cloud Toolbox and the Google Earth Engine to demonstrate pros and cons of each platform.
Our findings are currently being used by humanitarian organizations as evidence against the inhumane plans of the government of Bangladesh.
Surface Water Detection from Long Term Time Series of Earth Observation Data
Roberts, Dale (2); Mueller, Norman (1); Siquera, Andreia (1) - 1: Geoscience Australia, Australia; 2: Australian National University
Over the last few years, Geoscience Australia has developed a storage and analysis facility for Earth observation data and related datasets. Digital Earth Australia (DEA) contains Landsat data from 1986 to present and Sentinel-2 data from 2016 to present. The purpose of DEA is to provide researchers with the ability to do very large bulk analyses of earth observation data over continental areas and over time series. The first continental-scale water body analysis conducted on DEA was the `Water Observations from Space’ product, that classified water in every Landsat satellite observation from 1986 to 2017 using a decision tree methodology followed by a summary statistics model to show water extents and their frequency of occurrence. However there was very little use of time series analysis in the product.
Recently a new method of analysing earth observation time series has been proposed, where time series are analysed using high-dimensional statistics to provide characteristic pixel composite mosaics of the data over given time frames, and subsequent measures of variance associated with the data. Recently, we have been using these new products and developing new algorithms that harness the high-dimensional nature of the data in new ways for land cover identification, characterisation, and change detection.
In this paper we describe some new methods that use this high-dimensional approach to analyse time series of satellite observations for detecting surface water and statistics about its extent across the continent of Australia. We compare the results against our previous `Water Observations from Space’ method and discuss the advantages and disadvantages of the statistical time series methodology for water detection, land cover classification, and change detection more generally. Further, we explore how our new Sentinel-2 surface reflectance using the NBAR algorithm helps with classification and characterisation of these water bodies.
11:10 - 12:50
Inundation mapping with the L-band radiometer for carbon studies and flood mapping
Kim, Seungbum - NASA Jet Propulsion Lab, United States of America
Inundation and consequent anoxic condition induce methane release, which is one of the most potent greenhouse gases. Course-resolution radiometers are effective in mapping inundation over the continental scale of the methane emission. In comparison, the maps derived from high-resolution optical and radar data have restricted use due to clouds and limited coverage.
Flood monitoring is another application. High-resolution synthetic aperture radar (SAR) data are used frequently: uncertainties in the algorithm maybe alleviated by referencing to the radiometer-derived water fraction over a larger domain.
Towards these goals, we have derived the inundation extent at ~25 km resolution using the brightness temperature observations of the L-band Soil Moisture Active Passive (SMAP) mission by exploring the L-band capabilities to penetrate clouds and vegetation at 3-day revisit. Enhanced capability to eliminate the radiofrequency interference through the SMAP’s dedicated instrument is expected to further favor the outcome.
The assessments of the SMAP’s extent produce the following results. First, evaluations using the high-resolution maps made with Radarsat SAR (~5 m) and Landsat optical (~7 m) data showed the satisfactory difference of ~5 % (mean) and ~ 15 % (standard deviation) over the Canadian Prairies and California Delta at 8 temporal instances. The second target is West Siberia, one of the areas of significant methane exchance but the current estimates of the exchange differ by several times depending on the methodology. The SMAP water extent shows the consistent seasonality compared with the climatology (Global Inundation Extent from Multi-Satellites, GIEMS). SMAP’s water extent appears more realistic in winter than shown by GIEMS: even in winter there are unfrozen wetlands identified by SMAP, which is plausible considering that the southern boundary of West Siberia (50°N) ramians thawed. Third, the extensive flooding of the New Orleans in 2015 was examined using the maps generated by SMAP, Radarsat, and Japanese spaceborne SAR. The three agree fairly well spatially and the SMAP maps provide a realistic description of the flood’s temporal evolution.
The future evaluations will be performed between the SMAP-derived extent and the Sentinel SAR observations over Alaska and Northern Canada acquired during the ABoVE (Arctic Boreal Vulnerability Experiment) field campaign of 2017.
Acknowledgements: Brian Brisco (Canadian Centre for Remote Sensing), Sangho Yun (JPL), and Danica Shaffer-Smith (Univ. N. Carolina) kindly offered the inundation maps generated using Radarsat, ALOS, and Landsat, respectively.
Automatic Water Mapping Algorithm using Sentinel-1 Data within the ESA Hydrology Thematic Exploitation Platform
Garcia Robles, Javier
Garcia Robles, Javier (1); Blanco, Pablo (1); Balagué, Xavier (1); Gili, Albert (1); Koudogbo, Fifame (1); Novali, Fabrizio (2) - 1: TRE Altamira, Spain; 2: TRE Altamira, Milano
The Hydrology Thematic Exploitation Platform (TEP-Hydrology) is an ESA Portal providing large-scale Earth Observation (EO) products and services customized for hydrological applications. TRE ALTAMIRA has developed an automatic algorithm to obtain hydrological products (water mask, flood maps and statistics) employing SAR images in two main services within Flood Monitoring and Small Water Body Mapping Thematic Applications.
The Flood Monitoring enables detecting and delineating the extent of the area affected by flooding at high resolution take in consideration a water extent reference using data from Sentinel-1 and other SAR sensors as well as the possibility of refining the results using Sentinel-2 Optical Data. Additionally, the service includes further analysis to assess the evolution of water extent and floods over time: water and flood frequency maps, maximum floodable area and time series. The Small Water Body mapping is related to the identification and mapping of permanent and non-permanent water bodies and its temporal surface evolution.
TRE-Altamira has generated within the ESA TEP-Hydrology framework project, several results over two massive pilot basins: Niger River and Red River. On the scope of the project, it has also been generated flood mapping over Disaster Charter Alerts with this methodology. Two case studies are presented: a flood event in Costa Rica after Otto Hurricane in November 2016 and a flood event in the Caribbean Islands and the United States after Irma pass in September 2017.
The identification and delimitation of floods from EO data is a need for many hydrological communities and experts as well as Water Authorities. Additionally, there is a need to have an instrument to manage and share all the available information. Because of the high volumes of EO data, the water community needs to have access and process the large amount of data in a rapid, flexible and customized way so they can understand better which are the trends and impacts.
Synergies of Landsat, Sentinel-2, and -1 for improved characterization of surface water dynamics
Huang, Chengquan (1); DeVries, Ben (1); Huang, Wenli (1); Lang, Megan W. (2); Jones, John W. (3); Creed, Irena F. (4); Carroll, Mark L. (5,6) - 1: University of Maryland, United States of America; 2: US Fish and Wildlife Service, National Wetlands Inventory, USA; 3: US Geological Survey, USA; 4: University of Saskatchewan, Canada; 5: NASA Goddard Space Flight Center, USA; 6: Science Systems and Applications Inc., USA
The spatial and temporal variability of surface water has broad impacts on earth system processes, ecosystems, and human well-being. Publicly available global decameter observations provided by a virtual constellation of the Landsat and Sentinel-2 optical and Sentinel-1 synthetic aperture radar (SAR) satellites provide an opportunity to greatly improve the characterization of surface water dynamics from local to global scales. Here, we present a suite of highly automated algorithms for mapping surface inundation using these observations. We show that there are trade-offs between optical and SAR based methods in monitoring surface inundation dynamics. On the one hand, optical data from the Landsat and Sentinel-2 satellites allow for finer-resolution mapping of inundation features via sub-pixel unmixing algorithms, giving rise to better representation of numerous small but biogeochemically important water bodies and the highly heterogeneous surface water cover typical of most wetlands. On the other hand, optical images are frequently cloud-obscured, resulting in an irregular time-series and the inability to map critical events (e.g., peak floods), while SAR satellite sensors can provide imagery in nearly all weather conditions. Using historical optical data to train surface water classifiers, we show that while open surface water cover can be mapped efficiently using Sentinel-1 SAR imagery, generating regular estimates of inundation dynamics over time, complex backscatter signatures and noise due to SAR speckle present challenges to mapping mixed water-vegetation surfaces. We demonstrate that fusion of optical and SAR inundation estimates is often necessary to reliably capture rapid inundation dynamics. Our findings suggest that integration of optical and SAR data is necessary to produce consistent, reliable estimates of surface water dynamics, especially for regions dominated by small water bodies or highly variable inundation patterns. Continued efforts to improve multi-source inundation algorithms will be crucial to facilitate optimal use of observations from current and future Landsat, Sentinel-1, and -2 missions, as well as the forthcoming NASA-ISRO SAR mission (NISAR) and the Radarsat Constellation Mission (RCM), to advance surface water studies.
The findings and conclusions in this article are those of the author(s) and do not necessarily represent the views of the U.S. Fish and Wildlife Service.
Water Body Mapping and Monitoring in Arid Wetlands Based on Optical Satellite Imagery. A Case Study of the Lower Volga
Kozlova, Maria Vladimirovna
Kozlova, Maria Vladimirovna (1); Baig, Muhammad Hasan Ali (2); Kozlov, Alexander Vladimirovich (3) - 1: State Oceanographic Institute, Russian Federation; 2: Institute of Geo-Information & Earth-Observation (IGEO), Arid Agriculture University Rawalpindi, Pakistan; 3: Lomonosov Moscow State University, Faculty of Mechanics and Mathematics, Russian Federation
The Lower Volga arid wetlands maintain the high-productive ecosystems amongst the surrounding desert and dry steppe being the only source of fresh water for drinking, fishery, irrigation. It plays an important role in the local economy. Due to the vastness of the study area, which covers over 9000 (Volga-Akhtuba floodplain) and 20000 square kilometers (Volga delta), continuous in-situ observations on the numerous water bodies are challenging. The water body dynamics in arid wetlands being an important indicator of wetland ecosystem state, can be estimated by the satellite multispectral image time series analysis. Since that, satellite data combined with meteorological, hydrological, and other in-situ data is the only source of information for balanced and broad view on the wetland hydrographic network dynamics to develop the proposals for proper water resource management.
The key attributes of water bodies in arid climatic zones are its high rates and ranges of temporal and spatial variability, and its considerable heterogeneity in optical features caused by salinization, plant overgrowth, algal blooms, pollution, sediment load, etc., which arise from the drastic seasonal and long-term environmental variability.
Many of conventional methods based on optical satellite data analysis for such water bodies were often found to produce artifacts or unreliable results. Thus, the methods for satellite imagery processing need to be specifically adjusted to water bodies of arid wetlands, in order to reduce a shoreline, and an area estimation errors.
This study represents the methodological approaches to arid wetland water body mapping in Lower Volga, and estimation of water body dynamics in 1984-2017, based on the Landsat 4-8, and the Sentinel-2 imagery. The study includes the comparison of different kinds of image classification, tasseled cap transformations (wetness component), and the indices based on the selected spectral bands. The application results of conventional methods, such as supervised, and unsupervised image classification, and MNDWI index, together with the advanced approaches, such as Tasseled cap wetness component derived by the locally adjusted Jackson's algorithm and DFI index, were tested on different types of water bodies located in arid Lower Volga wetlands. The combined application of water indices with vegetation indices was also found to be productive.
Our approaches showed a great potential in water body mapping and analysis of the wetland hydrographic network dynamics. The methods undergo the continuous development. Some new results on their applications are to be reported.
Hyper-temporal Water Body Dynamics Mapping Using Sentinel-1 Time Series Clustering
Truckenbrodt, John (1); Schmullius, Christiane (1); Weise, Kathrin (2) - 1: Friedrich-Schiller-University, Jena, Germany; 2: Jena-Optronik GmbH, Jena, Germany
This study aims at providing water body dynamics maps at highest temporal resolution for better wetland characterization from space and as such is a mapping component of H-2020 project Satellite-based Wetland Observation Service (SWOS) (Grant No 642088). By depending on freely available data and demanding a high acquisition frequency across a large number of test sites from the North of Sweden to Central Africa, the Sentinel-1 SAR mission was chosen as the primary source of satellite images. Data from this mission acquired in Ground Range Detected (GRD) Interferometric Wide Swath (IW) mode VV polarization was selected as suited best for the mapping task due to storage volume (as compared to SLC data) and acquisition frequency (as compared to VH and HH polarization).
While SAR offers great possibilities for mapping water bodies from space, due to its insensitivity to clouds and independence from sunlight, difficulties arise from varying sea states. A roughened sea surface, returning more radiation to the sensor than a smooth surface, might be indistinguishable from land in the image. This effect of reduced contrast between land and water is particularly prominent in images from short wavelength SAR systems like X-Band and C-Band. Thus, common thresholding approaches like Otsu and Kittler-Illingworth, which assume a bimodal distribution of gray values in the image for distinguishing between dark water pixels and brighter land pixels are likely to be inaccurate. However, by incorporating e.g. interferometric and polarimetric techniques in combination with change detection approaches, a high accuracy for flood detection can be achieved regardless of wavelength. Yet, many of these methodologies are impossible to use on large spatial scales and high temporal frequency due to data scarcity and/or cost.
The choice of data, a single polarization C-Band VV setup, is not the most favorable for per-image water mapping due to the explained sensitivity to sea state, precluding land-water thresholding as described above for a large number of acquisitions. Thus, it was decided to perform the classification purely in the temporal domain making use not only of image intensity but also its variability over time. In this context a bright pixel resulting from high sea state can be seen as an outlier in an otherwise series of dark backscatter pixel acquisitions and thus be discarded from being classified as land unless it is succeeded by other bright pixels indicating a change from water to land. This combination of pixel value and time series neighbor relationships is transformed into a synthetic data space and subsequently clustered using a simple K-means approach for all pixels. After restoring the original time series with the clustering result a classification of land and water is achieved for each pixel in time.
Though being clustered/classified completely independent from each other, the pixels from the same acquisition form a low-noise and accurate classification map even at highest sea states. This approach has proven highly successful under different conditions in the observed wetlands while being conceptually simple and completely automated.
14:00 - 15:20
Lake monitoring in Siberia with Sentinel-1 and 2 data
Bartsch, Annett (1,2); Pointner, Georg (1); Widhalm, Barbara (2) - 1: b.geos, Korneuburg, Austria; 2: Zentralanstalt für Meteorologie und Geodynamik, Vienna, Austria
Lakes and ponds are ubiquitous features of the arctic tundra landscape. They are often associated with thermokarst phenomena (thaw lakes). Their changes over time are expected to reflect changes in permafrost. Satellite data are commonly used to analyze their spatial and temporal patters. High spatial as well as temporal resolution is required. Both SAR as well as optical data can be utilized. Features of interest are not only size and shape but also bathymetry and associated properties. The latter specifically includes ground fast lake ice along shallow shelfs and can be derived from SAR.
The Yamal peninsula is mostly underlain by continuous permafrost. The distribution of lakes is largely related to the patterns of marine terraces. We have utilized Sentinel-1 as well as Sentinel-2 data for mapping recent properties. A lake map has been derived from Sentinel-2 within the framework of the ESA DUE GlobPermafrost project as part of the landcover prototype development. Major challenges have been frequent cloud cover, seasonal lake change, sediments in lakes of floodplains and with erosional features as well as reflectance properties of high arctic lakes within wetland areas. Two years of data have been required for complete coverage of the area of interest. A post processing scheme has been developed for treatment of artefacts due to cloud masking effects based on time series analyses. SAR data can reveal differences in lake bathymetry among the terraces. There are however a range of phenomena on larger lakes which impact the ground-fast ice detection. These features can be identified in both, Sentinel- 1 and Sentinel-2 data. We have tested several algorithms to deal with these patterns. Improvements can be made under the assumtion that ground fast ice can only occur along the rims of the lakes.
Our results exemplify the challenges of lake mapping in permafrost regions, but also detail the information gain from using Sentinel-1 and Sentinel-2 data.
Automated Extraction of Time-Variable Water Surfaces based on Google Earth Engine
Schwatke, Christian; Scherer, Daniel - DGFI-TUM, Germany
Cloud-based data storage and geospatial processing as provided by the Google Earth Engine brings new potential for remote sensing and its applications. Long lasting downloads of large satellite imagery and raster files decreased by using the Google Earth Engine which allows user to work on larger area of interest. In this contribution, the new „Automated Water Area Extraction Tool“ (AWAX) for inland waters is presented. AWAX provides water surface area time series for users independently of their remote sensing knowledge and experience.
The methodology of AWAX is based on preprocessed Landsat images (1985-today) from which the best possible cloud-free monthly composite are used for the processing. Four different indicies - Modified Normalized Difference Water Index (MNDWI), New Water Index (NWI), Automated Water Extraction Index for Non-Shadow Areas (AWEInsh) and Automated Water Extraction Index for Shadow Areas (AWEIsh) — are applied to the composite and adaptive thresholds are used to classify water. Furthermore, data gaps caused by clouds, snow or missing data are filled by using statistial information of all existing Landsat images. This leads to reliable water area time series without monthly gaps. AWAX provides a long-term inundation probability mask, monthly maps and a water area time series for the investigated inland water. Finally, extracted water area time series are validated by using in-situ data of water level time series for different inland waters.
Mapping temporary excess water ponding on agricultural fields with Sentinel-1 & 2
Vekerdy, Zoltán (1,2); Qiu, Yun (1); van Lieshout, Arno (1); Czakó-Gál, Edina (2) - 1: ITC Faculty of University of Twente, the Netherlands; 2: Faculty of Agricultural and Environmental Sciences, Szent István University, Hungary
When precipitation or snow melt exceeds the infiltration capacity of soils on agricultural lands, the excess water is poning in the local depressions even in relatively flat areas. These ponds remain on the fields for shorter or longer periods, causing damage to agricultural production. This can be detected by timeseries of SAR and optical images. It is important to identify the location, spatial extension and severity of excess water inundation in agricultural areas. Due to cloud cover frequency in the wet periods, when excess water ponding occurs, a combinatio of optical and radar methods is needed. Because of the improved revisit time, Sentinel-1 & 2 provide optimal temporal coverage of agricultural areas in Europe with a high-enough resolution for identifying both the inundation and the damaged areas in the crops. Radar time series, supported by field information were used for monitoring the wetness and the water cover in the rural areas. Sentinel-2 and multi spectral images derived from drones, as well as digital elevation models were used for identifying the ponding areas. Polarization ratio of the radar images (normalized for incident angle) was analyzed with NDVI maps for dates without cloud cover, and the resulting regression was used to monitor the crop growth conditions with radar images under cloud cover. High resolution images and field measurements were used for the validation of the results. The developed methods were applied for two test areas in the Great Plain of Hungary to map the spatial and temporal distribution of surface ponding of excess water and to assess the damage caused by it.
Dynamic Water Surface Detection Algorithm Applied on PROBA-V Multispectral Data
Van De Kerchove, Ruben
Bertels, Luc; Van De Kerchove, Ruben; Reusen, Ils; Smets, Bruno; Wolfs, Davy - VITO, Belgium
A global scale multi-temporal and multi-spectral image analysis method for water body detection was developed using PROBA-V imagery with a spatial resolution of 333 m. The Red, NIR and SWIR bands of the atmospherically corrected 10-daily synthesis images are first HSV color transformed and subsequently used in a decision tree classification for water body detection. To minimize commission errors four additional data layers are used: the normalized difference vegetation index (NDVI), maximum water extent mask (MWEM), permanent glacier mask (PGM) and volcanic soil mask (VSM). Threshold values on the HUE and VALUE bands, expressed by a parabolic function, are used to detect the water bodies. Beside the water bodies layer a quality layer, based on the water bodies occurrences, is available in the output product. The performance of the water bodies detection algorithm (WBDA) was assessed using Landsat 8 scenes over 22 regions selected worldwide. A mean commission error (CE) of 1.3 % was obtained while a mean omission error (OE) of 13.4 % was obtained for minimum Water Surface Ratio (WSR) = 0.5 and drops to 8.6 % for minimum WSR = 0.6. Here WSR is defined as the fraction of the PROBA-V pixel covered by water as derived from high spatial resolution images, e.g. Landsat 8. Both the CE = 1.3 % and OE = 8.6 % (WSR = 0.6) fall within the user requirements of 15 %. The WBDA algorithm is being ported to Sentinel-3 and can be applied on Sentinel-2. The WBDA at 300m is fully operational in the Copernicus Global Land Service and products are freely available.
15:20 - 16:10
Chairs, All - .