Overview and details of the sessions and sub-session of this conference. Please select a date or session to show only sub-sessions at that day or location. Please select a single sub-session for detailed view (with abstracts and downloads if available).
C3: ID.10557 Wetlands Monitoring
From ERS1 To Sentinel 1 and from MERIS to Sentinel2 13 years of gained experienCe based on DRAGON EO data on water resource monitoring over the Yangtze middle reaches
1Unistra, ICube-SERTIT, France; 2IWHR; 3NIGLAS; 4LIESMARS; 5ICF & University of Wisconsin–Madison; 6formerly LEGOS;
Water resources monitoring from space is an application of remote sensing under full development, with arriving satellite such as Sentinel constellation, and of course the development of inland applications of altimetry and the future mission as SWOT. Since 12 years within the framework of the DRAGON ESA MOST program, a monitoring of Chinese major fresh water lakes, Poyang and Dongting lakes, located on Yangtze reaches, was realized exploiting large sources of EO data. Thanks to EO data archive exploitation and ongoing missions, the covered period starts in 2000 up to 2016, over these two core lakes. In addition, since 2012, start of DRAGON3, a similar approach is carried in parallel over the small lakes of Anhui Province: Wuchang, Shengjin and Baidang lakes.
Along the years, a key question was how to access to convenient satellite data with sufficient resource to insure a high temporal frequency, ie ideally with about 10 days of revisit. Over the years, an opportunistic approach was followed in order to insure this data access thanks to a large ESA support, accessing ENVISAT and ESA TPM data, as well as Chinese data. From 2004 to April 2012, ENVISAT Medium resolution products, MERIS and ASAR WSM data, were the core of the monitoring system. Since the loss of ENVISAT, the data exploited moved from MR to HR (20-30 m), thanks to the access of large volume of HJ-1 images in 2012 and 2014, as well as coverage Cosmo-SkyMed data (Wide mode 30 m over Poyang Lake, StripMap ones over Shengjin Lake), thanks to bilateral agreement with ASI. In the same way, 2014-2015, multi resolute TerraSAR-X data have been acquired over Poyang in 2014-2015. In addition, on most sensitive areas such as the Poyang Lake Natural Reserve (Jiangxi Prov.) and Shengjin Lake (Anhui Prov.), a unique set of VHR data, Pleiades, Kompsat, as well as SPOT time series from Take Five experiment have been exploited.
May 2014 was a milestone in term of data acquisition with a first Sentinel-1 image acquired over Poyang Lake, PR China, less than 6 weeks after its launch. From September to December 2014 a first Sentinel-1 time series was acquired with a frequency of 5 to 7 days. After a gap of near 4 months, since April 2015, Sentinel-1 acquisitions are now ongoing in IWS mode, with an actual revisit of 12 days. In 2015, a second milestone corresponds of the first acquisitions of Sentinel-2 data with, in fall 2015, a first Sentinel-2 time series done and accessible. These free cloud images were acquired on the 13th of September, 3rd, 20th and 23rd of October 2015. Then, bad weather conditions limited Sentinel-2 acquisitions until the 29th of December 2015, and the 7th of February 2016.
In parallel of the images acquisitions, a huge work was done on water height retrieval exploiting near all available archived or ongoing mission, Topex, Jason, RA Envisat, Altika. Virtual gauge stations have been defined over Yangtze segments, Poyang and Dongting lakes as well as smaller lakes of Anhui Province.
Satellite imagery was exploited for complementary topics in order to characterise and monitor the wetlands:
- Water surfaces
- Vegetation on lakes’ banks and floating/submerged vegetation
- Water quality assessment of lakes, ie suspended material analysis
- Anthropogenic impacts, ie sand dragging and high consumption of natural space by fish traps
In term of water surfaces behavior, the time series analysis allowed to characterize trends on an intra and inter-annual basis. A key result is the increase of years with water default since the mid-2000, ie near one year every two years. Over the latest 15 years, there is an apparent increase of drought tendency. There are, since 2006, 6 periods, long or short, of drought. Theses have to be compared with the only two periods of drought observed from 2000 to 2005. Another disturbing or anomalous element is the timing of the Poyang Lake draw off which is increasingly early since the mid-2000. Whereas, removal of the water is known, from literature, to begin in fall, ie mid-September or mid-October; in 2006, it was observed for the first time in mid-summer, in mid-August to be more precise. In 2013, the flooding season was reduced to a few days, with a redraw in mid-July. Same behaviors were observed in 2015, with a maximum of water extent reached in mid-June, and a draw off starting in the beginning of July. Therefore, with a different staircase pattern and surprising plateau effect from mid-August to Mid-October followed by a “classical” rapid draw off of the water with a decrease of 500 km2 in 12 days between the 15th and 27th of October. The driven forces of these behaviors (meteorological parameters, large scale climate effect, local management of water resource) are still under investigations. The apparent tendency of succession of long periods of deficit of water of Poyang Lake and Dongting lake, will surely, in a very short time, induce serious degradation of ecological system in a context of drought disaster. The risk is that the observed drought trends continue over the years
The full development of the Sentinel constellation (ie integration of Sentinel-3 in the monitoring scheme, improvement of the revisit thanks to the future launches of Sentinel-1B and Sentinel-2B) will allow the monitoring of the variables and enables the build-up of long data time series at equidistant and short time intervals
Monitoring Seasonal Changes in the Water Surface Areas of Poyang Lake Using a Time Series of Cosmo-skymed Data
China Institute of Water Resources and Hydropower Research, China, People's Republic of;
Poyang Lake, the largest freshwater lake of China, is well known for its ecological importance as a wetland system.. Understanding dynamics change in wetlands may benefit from a joint variation analysis in changes of size, shape, position and extent of water bodies. In this study, we used Cosmo-Skymed ScanSAR mode (HH) data for one-year monitoring of seasonal changes in the water surface areas of Poyang Lake from January 2014 to December 2014. To delinate water surface areas, a histogram analysis with a global threshold classification was performed, and the results were validated using Landsat8 OLI data. The results showed: Poyang Lake water body area showed a significant seasonal variation, minimum value was about 690 km2 and maximum value reached 3500 km2. A unique spatial pattern occurred during those months that a dramatic flooding or recession happened. The Cosmo-Skymed time series could capture temporal changes in the water body spatial distribution. Comparison of the resultant classification of Cosmo-Skymed and Landsat8 OLI yielded 91% of flood mapping accuracy. These demonstrate X-band Sythenetic Apture Radar is applicable to detect the wetland inundation area ,and there is the great potential of Cosmo-Skymed constellation for water bodies monitoring, regarding relative great revisit frequency and high resolution.
Variations of Sediment Budget in Poyang Lake during 1955~2010
Key Laboratory of Poyang Lake Wetland and Watershed Research, Ministry of Education, Jiangxi Normal University;
Poyang Lake is the largest fresh water lake in China. The sediments disposition is important in providing medium for wetland glassland and benthic animals. In this paper, the water discharge and sediment content gauged by the hydrostations in the main five triburaries (named as Gan River, Fu River, Xin River, Rao River and Xiu River) in Poyang Lake watershed were used to investigate the sediment budget in Poyang Lake. The following results was showed: (1) The total sediments loaded into Poyang lake from the main five tribuaries was about 811.687Mt during 1955-2010, in which Gan River accounted for 59.7%, Xin River accounted for 13.7%, Xiu River accouted for 10.2%, Fu River accounted for 9.7 and Rao River accounted for 6.7%；(2) The annual sediments loaded into Poyang lake was mainly decided by water discharged into Poyang Lake. And the inputted sediments were synchronized to the water discharge seasonally. (3) Sediments loaded into Poyang Lake decreased sharply caused by reservoir interception, though forest coverage changes also contrbute to the decrease of sediments; (4) The total sediments loaded into Yangtze River from Poyang Lake was about 560.103Mt during 1955-2010. And it showed a decreasing trend during 1955-2000, but the trend was interupted by the practice of sand mining in the new century; (5) Affected by the backflow from Yangtze River in flood period, the sediments loaded into Yangtze River were asynchronized to water discharge. And the sediments loaded by the backflow from Yangtze River was reduced dramatically because of the emplacement of Three Gorges Dam;（6）The sediments budget in Poyang Lake was changed by sand mining. Sediments deposit was estimated about 1.408mm/a during 1955-2000, but outputed sediments showed greater than the inputed sediments during 2001 to 2010 beceause of the practice of sand dredging.
Combating Aggressive Macrophyt Encroachment on a Typical Yangtze River Lake: Lessons from a Long-term Remote Sensing Study of Vegetation Changes
1Life Sciences, University of Science and Technology of China, Hefei, Anhui 230026, China; 2State Key Laboratory of Urban and Regional Ecology, Chinese Academy of Sciences, Beijing 100085,China; 3ICube-SERTIT, Strasbourg University, BP10413 F-67412 Illkirch Graffenstaden, France; 4Department of Bioscience, Aarhus University, Kalø, Grenåvej 14 DK-8410 Rønde, Denmark;
Overabundant growth of emergent lacustrine plants can cause biodiversity, ecosystem service and economic loss. The two-basined Wuchang Lake is a typical small shallow lake within the Yangtze River floodplain. Expansion of the emergent macrophyte Zizania latifolia at Wuchang Lower Lake (to 49 km2 in area) has increasingly denied the local community open water for fishing since the 1980s. To better understand the causes of these changes and potential remediation, we used Landsat imagery from 1975-2012 to determine the effects of trophic status, lake bed elevation and changes in water level on the annual extent of macrophytes in the Lower Lake. These analyses showed that: 1) Z. latifolia progressively covered the Lower Lake while remaining confined to one inlet in the Upper Lake; 2) spring water level variation and lake bed elevation had significantly negative effects on expansion rate, while variation in autumn water level variation and trophic status had non-significant negative and positive effects on expansion rate respectively. Based on these results, to reduce the expansion in Z. latifolia, we recommend maintaining spring Lower Lake water levels above 13 m, cutting shoots in June and July, and maintaining autumn water levels above 13.5 m combined with simultaneous shoot removal. Further we propose to continue lake monitoring by utilizing new generation of sensors, like Chinese HJ-1 and European Sentinel-2 with more frequent revisit times comparing to Landsat archives, to guide the management efforts as well as appraisal the effects of certain management in future years.
Mapping water bodies exploiting multi resolution TerraSAR-X SAR data: gained experience from plain flood monitoring in Central China, the Meixihu case (Poyang Lake)
1ICube-SERTIT, University of Strasbourg, France; 2NIGLAS, P.R. China;
In 2013, the German Space Agency (DLR) extended the imaging capability of TerraSAR-X, adding new products in term of spatial resolution, and corresponding swath: the Wide ScanSAR and the Starring Spotlight (SPL) modes. By the way, TerraSAR-X offers the possibility to acquire at resolutions ranging from 30 m to 0.25 m, with intermediate ones at 16 m, 3.5 m and 1 m. The covered areas are ranging depending of the mode from 40 000 km2 to 12 km2 with intermediate surfaces of150 km2 and 50 km2. Complementarities in term of resolution versus swath were investigated over a specific target: water bodies’ recognition/extraction and monitoring over embedded sensitive areas.
One test site, located under the tropics, corresponds to the Poyang Lake, the first fresh water resource in China, under remote sensing monitoring since 15 years exploiting both Middle and High resolution satellite images. Over this area, all the modes have been investigated, from the Wide ScanSar allowing to capture in a single image the 24 000 km² of the AOI, focusing the high resolutions, StripMap and Spotlight (SM & SL) of the Poyang Lake Natural Reserve (PLNR) and for highest one, Starring Spolight mode, over Meixi Hu and Bang Hu, the most sensitive and rich elements in term of biodiversity, of the Poyang Lake Natural Reserve.
Both over Chinese and Alsatian (France) sites, data acquisition, in HH polarization, were requested to be in a shortest at possible time interval, ie within 3 to 5 days and this every month or twice a month from June 2014 to June 2015. In term of acquisition, over each site, the SPL failed over one of the smallest area of interest, Bang Hu, but functioned very well on the other one, ie Meixi Hu.
The multitemporal approach was as promising as expected, and thanks to the very high resolution and revisit, it was possible to derive, water masks from each image and then water occurrence maps. The multiresolution approach was in addition exploited for validating the water extraction results from one coarser image towards better resolute ones. In a global view, results are very convenient, therefore this approach also highlight the limits of the validation approach. Indeed, even if it is a plain flood phenomena, characterized by a slow dynamic, that are targeted, the results shown that even on a few days (ie 3 to 5) water redraw can be very noticeable limiting the possibility of comparison in term of accuracy between data acquired at 3 days interval.
The results highlight the importance of accessing to high revisiting data, as surface variations with are sensitive to river discharge and water table fluctuations, varies positively or negatively , with a within a few days (5 to 10) but also as water bodies are relative small (few ha to few km square) to high resolute imagery. This confirm and enhance the previous results obtained mostly based on successive mono sensors approach, SPOT4 Take Five, Pleiades-HR, Cosmo-SkyMed. Further steps will correspond to the synergy between SAR constellation, ie TerraSAR-X and Sentinel-1, both in term of revisit and accuracy, and secondly with optical Sentinel-2 constellation.
Water surface and volume monitoring over Yangtze: water stock monitoring of Sha Hu and Da Hu Shi within the Poyang Lake Natural Reserve
1ICube-SERTIT, University of Strasbourg, France; 2LEGOS – ICube-SERTIT – CS-SI, France;
On the continuity of Poyang Lake’s water surfaces monitoring, a next step concerns the Poyang Lake water volume estimation thanks to the intersection of water surfaces with a Digital Elevation Model (DEM). Until now, the SRTM DEM was the only model available over Poyang Lake, but with these 90 meters resolution, it was not appropriate for this application. Thanks to the Thematic User Commissioning for the Pleiades-HR valorization program of the French Space Agency (CNES) and to the German Space Agency (DLR) for an access to Pleiades-HR and Tandem-X imagery respectively, new VHR DSMs over Poyang Lake Natural Reserve were generated.
First request for this was the availability of a high resolution DEM which were generated exploiting 11 Tandem-X products. On the resulting DEM there was resulting artefacts such as outliers on water bodies due to a loss of coherence of SAR signal on water. In order to resolve these outliers, a watermask, containing 250,000 water bodies corresponding to lakes, pounds, reservoirs, paddy field and river courses has been extracted from the magnitude image using a threshold technique. The final DEM is obtained by subtracting the water elements to the DEM and providing an arbitrary height value to these water bodies.
For the Pleaides tri-stereo set, DSM was generated with MicMac, an open-source toolbox based on a multiresolution and multicorrelation approach, developed at IGN. Accuracy estimated by previous studies comparing DSM Pleiades with LIDAR gave a standard deviation of 1.5 m.
Two approaches have been explored for water volume calculation over a set of two lakes of the Poyang Lake Natural Reserve, the Da Hu Shi and the Sha Hu. First, a selection of coherent altimetry from Pleiades DSM, in a given buffer around each water bodies, allowed to derive water height of each lake. The same methodology was applied to linear water bodies (rivers) with a hydraulic coherence rule: for a given river segment, the maximum of altitude should be less than the minimum of the previous one.
Then, another approach consisted in the use of water bodies lines as altimetric isoline, which was validated over Sha Hu by comparison of water extent time series with reference bathymetric data. Regarding the complexity of the lake’s borders (mud, wet soil, flooded vegetation), only high correlation of altimetric data from Pleiades DSM have been selected to derive the altitude of each water bodies lines. Thus, it is possible to derive from this new extended DSM, water volumes. The results show for Sha Hu an average elevation of water bodies between 13.16 to 15.8 meters. Over this limit, the lakes is overflowing into the Bang Hu. For a water altitude of 15 m, the lake has a surface of 4.7 km² and an estimated volume of 1.4 km3. Regarding this promising result, the approach will be firstly applied on Da Hu Shi, and then to the lakes within the depression that form the Poyang Lake, allowing to extend the methodology to wider areas and bigger water volumes.
Wetland Feature extraction in Poyang lake from Multi-sensor and Multi-temporal images
1ESRIN,ESA,Frascati,Roma,Italy,00044; 2Key lab of Poyang Lake wetland and watershed Research (Ministry of Education), Jiangxi Normal university,Nanchang,Jiangxi,330022,China; 3State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing,Wuhan University, Wuhan 430079, China; 4ICube-SERTIT, University of Strasbourg, 300 bd Sébastien Brant, CS 10413, F-67412 Illkirch Cedex, France,;
Under the high dynamic hydrological variation in last decades, the nature wetlands of Poyang Lake faces a big challenge in biodiversity decline and wetland degradation. As the dominant vegetation is herbaceous vegetation and aquatic macrophytes in the Poyang Lake wetlands, feature description of dynamics of wetlands difficult to map by one source or one time phase of remote sensing image. Time-series Sentinel 1A (C-band) for Radar images and for multi-spectral optical satellite data Sentinel2 as well as Sentiel2 like, e.g., Spot 5- take5, are combined to investigate the spatial-temporal variation of wetlands in autumn and spring growing season, using vegetation index and scattering scheme among others.
Normalized different vegetation index (NDVI), soil-adjusted vegetation index (OSVAI), Simple Ratio (SR) and Modified Normalized different water index (MNDWI) are extracted from Spot5-take5 and Sentinel 2A to study the vegetation variation trends during the two periods for different vegetation types (Aquatic macrophytes, Hygrophilous vegetation and Emerged plants).
Backscattering coefficients, Shannon Entropy (SE) and H/Alpha wishart classification value are selected from the Sentinel 1A to study the scattering scheme and structure information of the wetland landscape. Water inundation frequency map were generated from Sentinel 1A data in 2015 and the first half year of 2016 to study the different water conditions of the wetlands.
The results showed that NDVI are saturated or not sensitive to each other when the index value get to 0.7. Variables discriminating reed marshes from other land cover types were the NIR band between March and June, the OSAVI of January, and the MNDWI of September. Submerged macrophytes beds were discriminated with the shortwave-infrared band of January, the NDVI & NDWI of September, the red band of September and the SR index of April and August.
For Sentinel 1A data, VH polarization is better to investigate the wetland vegetation data than VV polarization. Spatial-temporal Shannon Entropy and H/Alpha classification values could be a good way for the dual-polarization Radar to describe the different scatter scheme of the wetland object during different growing statues of different vegetation communities.
The obtained results over Poyang lake shows that multi-sensor and multi-temporal Sentinel series images are very efficient to improve the accuracy of wetland mapping and increase the class numbers which could be distinguished.
Expects from Sentinel-2 for tropical small wetlands monitoring based on 2013 and 2015 Take-Five experiment : case of Baidang Lake, Anhui Province
1ICube-SERTIT, University of Strasbourg, France; 2State Key Laboratory of Urban and Regional Ecology, Chinese Academy of Sciences, P.R. China; 3NIGLAS, P.R. China;
Wetlands are indispensable for the countless benefits or “ecosystem services” as they provide to humanity ranging from freshwater supply, food and building materials and biodiversity, to flood control groundwater recharge and climate change mitigation. Therefore, it is demonstrated that wetland area and quality continue to decline in most regions of the world. As a result, the ecosystem services that wetlands provide to people are compromised. Satellite data are widely exploited to monitor this very valuable ecosystem from global to local scale. Within this context, the arrival of Sentinel data marks a turning point in the Earth Observation community. The rapid and systematic dissemination of free images provided by the Sentinel system opens up important perspectives for operational monitoring territories from local and regional to global scales, with a high update frequency. The aim of this paper is to present the potential and expected contributions of Sentinel-2 data, for detecting and monitoring wetlands and inland water bodies and related applications.
In addition to ongoing works over large wetlands such as Poyang Lake ones, an innovative part of the work will focused its analysis over a relative small one, the Baidang Lake in the Anhui Province (P.R. China). Baidang Lake is one of the Anhui lakes, located in the southern part of Anhui Province, downstream on the Yangtze River, in a set of lakes more or less connected to this main river. In this area, major lakes are, in term of size, the Dagan and Huang lakes, the Bohu, the Caizi Hu. In a second rank there are the Wuchang, the Shengjin, Fengshu and the Baidang lakes. The size of the Baidang Lake is about 60 km2 including 20 km2 of wetlands. The lake receives water from its own catchment, supplemented by influx during maximum flows from the main Yangtze River via a sluice. The depth of water is relatively weak and the annual variation is about 4 meters (10 meters in case of major flood events such as in 1998-1999).
Looking forward to the next Sentinel-2 mission, CNES and CESBIO with ESA’s support have carried twice Take-Five experiment, to simulate the repetitiveness and associated data flow of Sentinel-2 satellite. At the end of their life, SPOT4 and SPOT5 were used as a simulator of the time series that ESA's Sentinel-2 mission will provide. Satellite orbits were modified in order to insure a five days revisit, from April to September 2015 for the second one based on SPOT5. It has to be noticed that SPOT data have also a similarity with the Sentinel-2 in term of spectral coverage, thanks to its three visible spectral bands, SWIR bands, as well as in term of spatial resolution, thanks to the 10 m ground sampling of the multispectral bands of SPOT5. Unfortunately, due to cloudy weather, unfavorable for optical acquisitions, in 2013 only 36% of Sentinel-2-like data were exploitable over Baidang Lake. In addition to complete and extent this temporal sampling, Landsat-8 images were also exploited. By the way, 8 SPOT5 images covering the April to September 2015 period have been analyzed, with 7 Landsat-8 images from December 2014 to October 2015 and 2 Sentinel-2 images of the 20th of October 2015 and the 7th of February 2016. The total of available images gives an idea of the actual frequency of exploitable images that could be reached by the Sentinel-2 sensor over a year in such tropical conditions, ie about 1.3 images by month.
The obtained results confirm the high interest of SWIR bands for water bodies’ extraction, wetland identification, even if some precautions have to be taken to avoid expected confusion, i.e. sand/mud banks which could cause overestimation of water surfaces. Moreover, thanks to the HR² (High Resolution and High Revisit) of Sentinel-2-like data, it was possible to monitor relative small hydrological systems, highlighting the variability of water surfaces within these areas. In regards to the SAR data, such as Sentinel-1, the Sentinel-2 like time series allowed to characterize the invasive floating vegetation (Trappa, Zizania species) and to monitor its extent over a year. On the Sentinel-2-like images acquired over the Anhui lakes it is possible to monitor precisely the development with the increase of the temperature of floating/submerged vegetation, based on a monthly to bi-monthly acquisition.
Finally the Sentinel-2, alone or associated synergistically with Sentinel-1, will be a powerful tool for mapping and monitoring rich, complex and sensitive large ecosystems such as the monsoon lakes in Asia. The DRAGON and Take5 experiments, illustrate the benefits and interests of these Sentinel data for global and local water bodies monitoring, but also for related applications such as biodiversity, epidemiology, or societal stakes (land/urban planning, risk management, etc). A next step will be the synergistic exploitation of Sentinel-2 time series with Sentinel-3 sensors, the Ocean and Land Colour Instrument and the SRAL altimeter to combine water extent, vegetation behaviors in relations with water height and water quality indexes.
Contact and Legal Notice · Contact Address:
Conference: 2016 Dragon 3 Final Results Symposium
|Conference Software -
ConfTool Pro 2.6.94
© 2001 - 2016 by H. Weinreich, Hamburg, Germany