Dongting Lake Dynamics ID-10697 Current Advances in Monitoring Inundation and Land Use Dynamics of Dongting Lake Wetlands Juliane HUTH, YANG Bo, LI Jing, Claudia KUENZER ESA-Dragon-3-Symposium June 22-26, 2015, Interlaken, Switzerland
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Study Site – Dongting Lake, Hunan, China
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Challenges at Dongting Lake Economic Use of Wetland Areas Floods and Droughts Water Quality Urban Growth Cash Crops
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Objectives of the Project ID 10697 "Assessing Wetland Dynamics and Land Resources of Dongting Lake, China" - ESA-Dragon3 project (2012 – 2016) – focussing on joint data use of Chinese and European satellite sensors - Assessment of flood pulse, wetland development, quantification of wetland degradation, derivation of land use change in direct surrounding of the Dongting Lake - Methods: remote sensing data analyses based on radar and optical data with preferably automated derivation of surface water, and further parameter to describe flood events and land cover change - Data used: Envisat ASAR, Terra-SAR-X, Sentinel-1, SPOT, HJ-1 - Data to be used in future: Sentinel-2 and further Chinese data…
Dongting Lake Flood Monitoring
WaMaPro – Principles and Techniques Input data Watermask (DN) Input: SAR amplitude DN values Medianfilter Remove ‚islands’ (3x3) and ‚lakes’ Filtering (island size, lake size) Threshold definition (confident Threshold Definition (water and land water and confident land) threshold) Image processing in buffer zone B5 Convert to binary of unconfident land/water images Morphological operations for Morphological Closing edge smoothing with fixed value (fix value) B1 B2 (water) (land) Removing of artefacts with user given size Comparison Image Dilation B4 (B3 && B2) || B1 (sesize) software tool implementation B3 GSTAIGER, V., GEBHARDT, S., HUTH, J., WEHRMANN; T. and C. KUENZER, 2012: Multi-sensoral derivation of inundated areas using TerraSAR-X and ENVISAT ASAR data. International Journal of Remote Sensing, Vol. 33, 22, 7291-7304, DOI:10.1080/01431161.2012.700421
WaMaPro – open-source based tool implementation WaMaPro – open-source based tool implementation
WaMaPro – open-source based tool – GUI
Accuracy Assessment Watermask derived from TSX stripmap compared to digitization from high-resolution optical data: Producers Accuray: 96,7 % Users Accuracy: 89,9 % Overall Accuracy: 91,3 % aquaculture ocean MARTINIS, S., KUENZER, C., WENDLEDER, A., HUTH, J., TWELE, A., ROTH, A., DECH, S.: Comparing four different approaches for operational SAR- based water and flood detection. Submitted to International Journal of Remote Sensing.
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Sentinel-1A Data – available since Spring 2015
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Sentinel-1A Watermask Processing Results Sentinel-1A 2015-04-28 Watermask result from Sentinel-1A 2015-04-28 data set
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Comparison of spatial resolution TSX 3m vs. Sentinel-1A 10m vs. ASAR 150m TerraSAR-X stripmap mode 2012-04-16 Sentinel-1A 2015-04-28 Envisat ASAR-WSM 2010-05-17
Derivation of Inundation Frequency Products KUENZER, C., GUO, H., LEINENKUGEL, L, HUTH, J., LI, X., and S. DECH, 2013: Flood mapping and flood dynamics of the Mekong Delta: An ENVISAT-ASAR-WSM based Time Series Analyses , Remote Sensing 5 (doi:10.3390/rs5020687), 687-715 Slide 13
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken • ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken TerraSAR-X Data Footprint
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Inundation Frequency at East Dongting Lake derived from high resolution TerraSAR–X Data 2012/2013 2012 2013 • 2012 – 14 scenes (Jan–Dec) – 2013 – 8 scenes (Jan–Mar) Yellow to orange – flood prone Blue – permanent water bodies •
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Flood Extent for 2012 – from TSX – Detail Views 1 4 4 5 2 3 5 1 2 3
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Sentinel-1A Data Footprint
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Inundation Frequency for Dongting Lake derived from NEW Sentinel-1A Data – April and May 2015
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken ASAR-WSM Data Subset Footprint
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Annual Flood Distribution at Dongting Lake derived from ASAR–WSM Data – 2010/2011 2010 – 8 scenes (May–Sep) flood year - compared to water level data of Chenglingji • hydrological station – measured between 27 m up to 33 m in this period 2011 – 9 scenes (Apr–Nov) dry year – measured 21 m to 28 m in this period •
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Sentinel-1A- Watermask Processing for Flood Detection Flood event between 2015-04-28 and 2015-05-02 advantage of Sentinel-1 repeat cycle •
Dongting Lakes Monitoring of Land Resources Quantification of Economically Important Vegetation Types within Wetlands of DTL
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken SPOT Data Footprint
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Flowchart of data processing for SPOT image classification based on seasonal information
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Processing SPOT data using seasonal information - Seasonal information for classification of 2-season-dataset - NDVI Difference – already applied - TC Greenness Difference and TC Brightness Difference – will be included as next step - Common Features: - Spectral bands 1-3/4 of SPOT2/4 - NDVI - Tasseled Cap Greenness and Brightness - Principal Components
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Classification of 2007 and 2010 SPOT data – including seasonal information
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Validation Matrix 2007 – with absolute numbers of validation samples and User’s Accuracy (UA) and Producer’s Accuracy (PA) in [%] CLASSIFIED agric. area agric. area agric. area poplar tree reed reed evergreen wet-land artif. mud-flat water Total PA type 1 type 2 type 3 cover fresh cut old cut trees veget. surface agricultural area type 1 R 62 13 1 7 5 88 70,45 agricultural area type 2 E 7 74 1 1 1 84 88,1 agricultural area type 3 F 4 227 9 22 262 86,64 poplar tree cover E 17 457 4 478 95,61 reed- fresh cut R 25 242 6 273 88,64 reed- old cut E 23 302 325 92,92 evergreen trees N 4 1 1 20 1 127 3 157 80,89 wetland vegetation C 3 31 34 91,18 artif. surface E 3 3 1 94 101 93,07 mudflat 1 4 177 9 191 92,67 water 99 15 24 313 451 69,4 Total 79 95 371 492 280 330 128 42 104 201 322 UA 86,2% 78,48 77,89 61,19 92,89 86,43 91,52 99,22 73,81 90,38 88,06 97,21
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Validation Matrix 2010 – with absolute numbers of validation samples and User’s Accuracy (UA) and Producer’s Accuracy (PA) in [%] CLASSIFIED agric. area agric. area agric. area poplar tree reed reed evergreen wet-land artif. mud-flat water Total PA type 1 type 2 type 3 cover fresh cut old cut trees veget. surface agricultural area type 1 R 46 17 6 7 76 60,53 agricultural area type 2 E 5 80 2 87 91,95 agricultural area type 3 F 5 257 6 2 1 8 279 92,11 poplar tree cover 1 E 272 6 66 345 78,84 reed- fresh cut R 1 203 3 207 98,07 reed- old cut E 30 166 196 84,69 evergreen trees N 1 22 10 75 108 69,44 wetland vegetation C 7 23 4 2 36 63,89 artif. surface E 2 2 2 8 8 77 6 1 106 72,64 mudflat 3 9 1 211 8 232 90,95 water 23 1 46 348 418 83,25 Total 62 104 268 332 251 170 141 40 96 267 359 UA 84,1% 74,19 76,92 95,9 81,93 80,88 97,65 53,19 57,5 80,21 79,03 96,94
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Classification of HJ-1A data of 2013-01-01 Classification more difficult compared to SPOT etc., since no atmospheric correction can be applied for HJ-1 with standard procedures like ATCOR, therefore seasonal combinations cannot be applied single dataset classification
• ESA-Dragon-3-Symposium • June 22-26,2015 • Interlaken Validation Matrix 2013 – with absolute numbers of validation samples and User’s Accuracy (UA) and Producer’s Accuracy (PA) in [%] CLASSIFIED agric. area agric. area agric. area poplar tree reed reed evergreen wet-land artif. mud-flat water Total PA type 1 type 2 type 3 cover fresh cut old cut trees veget. surface agricultural area type 1 R 318 1 319 99,69 agricultural area type 2 E 7 173 11 34 13 1 9 248 69,76 agricultural area type 3 F 6 171 6 2 185 92,43 poplar tree cover 43 E 245 10 67 34 399 61,4 reed- fresh cut 8 R 20 110 14 152 72,37 reed- old cut E 11 438 449 97,55 evergreen trees N 3 9 110 122 90,16 wetland vegetation C 20 10 1 27 58 46,55 artif. surface E 3 4 2 3 102 114 89,47 mudflat 217 217 100 water 1 358 359 99,72 Total 402 215 182 281 158 444 191 62 112 217 358 UA 86,5% 79,1 80,47 93,96 87,19 69,62 98,65 57,59 43,55 91,07 100 100
Recommend
More recommend
