GMES LAND COVER MAPPING IN LITHUANIA

Transcription

1 GMES SDI INSPIRE Applied EO INFORMATION GOVERNANCE EU CITIZENS GMES LAND COVER MAPPING IN LITHUANIA Gedas Vaitkus AGI ARC, Kaunas, LT

2 DATA & SERVICE PROVIDERS DATABASES DATA PROVIDERS CLC90LT GDBLT50000 ORT10LT GDB10LT KDB10LT ORT5LT GDB5LT Institute of Aerial Geodesy, Ltd. SŽNS_DB10LT Dirv_DB10LT Mel_DB10LT Žinv_DB10LT PLB LT CLC2000LT CNG1995/00 CLC2006LT CNG2000/06 Forest Information System State Land Survey Institute Agricultural Information and Business Center Institute of Ecology (Vilnius University) Forest Inventory and Management Institute CLIENTS / SERVICE PROVIDERS State Enterprise GIS-Centras NATIONAL LAND SERVICE META DATA ENVIRONMENT PROTECTION AGENCY LGII STATE FOREST SURVEY SERVICE

3 European Level Datasets Scale 1:100,000 WHOLE COUNTRY: CORINE LC 1995 CORINE LC 2000 CORINE LC 2006 CLC Change CLC Change COASTAL AREA: CORINE LC 1975 CLC Change

4 National Level Datasets Scale 1:50,000 LTDBK50000 digital reference base-map with a backdrop of satellite imagery, created in 1994/96 in cooperation with Satellitbild (Sweden). Last update 2005.

5 National Level Datasets Scale 1:10,000 ORT10LT digital ortophoto reference raster base map (1995 panchrom., 2005 color images), pixel resolution 0.5 m. GDB10LT digital georeference database and base-map. Some thematic classes still under production. KDB10LT full thematic scale cartographic database (only several towns available so far).

6 Municipal Level Datasets Scale 1:5,000 ORT5LT very high resolution (0.25 m) digital orto-photo reference raster base map (2005/6 color images) of urban areas. GDB5LT digital georeference database and base-map of 94 urban areas, so far with limited thematics, but very high accuracy.

7 Land Information System Databases SŽNS_DB10LT database of land use restrictions; Dirv_DB10LT database of soil ; Mel_DB10LT database of irrigation system and wetlands Žinv_DB50LT land inventory database.

8 FP7 Geoland2

9 FP7 Geoland2

10 FP7 Geoland2

11 FP7 Geoland2 SACHMO Over 200 testing sites across Europe with detailed mapping of 10 LC classes; Statistical analysis of LC structure and changes; Modeling of trends across Europe.

12 PELKIŲ AUGMENIJOS BENDRIJŲ ERDVINĖ ANALIZĖ Pelkinės augalijos klasifikavimui panaudotas vienas iš spektrinės analizės algoritmų RGB klasteringas (RGB clustering). Vykdant šią funkciją, pasirenkami 3 spektriniai kanalai: R G B 4 (nearinfrared) 5 (infrared) 3 (red) ir jų pagrindu yra suformuojamas vientisas tematinis rastrinis sluoksnis, kuriame spalvų reikšmės kinta nuo 0 iki 255.

13 AUKŠTUMALĖ BOG Mišrus miškas Spectral analysis and RGB clustering of standard and pan-sharpened Landsat ETM images provide different levels of spatial and thematic accuracy in vegetation mapping: 28,5 m Spygliuočiai 14,25 m Lapuočiai Eglės Pušys Mišrus miškas Eglės Differences in detection of tree stands: Coniferous Mixed forest Pine and spruce Pine separated from mixed with deciduous Pušys Eglės Pušys Pelkinės bendrijos Eglės Pušys

14 ORT10LT_2005 LANDSAT ETM RGB 453 RGB 453 CLUSTERING RECODING

15 Satellite imagery: USGS Landsat 5 = 30 m

16 Satellite imagery: SPOT 4 = 20 m

17 Satellite imagery: RapidEye = 5 m

18 KOMPSAT 2 RGB: G N R 1. Vector layer with 14 representative areas ( signatures ) of 3 classes: - water - agricultural areas - urban areas 2. Convert vector layer into raster map BMU Bonn / Berlin, Oktober 2006

19 3. Preparing the signature file (grass: // i.gensigset) Calculating the signature statistics of the image using the training map. - bands combination of image - main classes of training map - defining maximum amount of sub-classes of each class - resolution of signature file 4. Image contextual classification (grass: // i.smap) The i.smap program is used to segment multispectral images using a spectral class model known as a Gaussian mixture distribution. Gaussian mixture distributions include conventional multivariate Gaussian distributions, this program may also be used to segment multispectral images based on simple spectral mean and covariance parameters. It uses the sequential maximum a posteriori (SMAP) mode. The SMAP segmentation algorithm attempts to improve segmentation accuracy by segmenting the image into regions rather than segmenting each pixel separately The SMAP algorithm exploits the fact that nearby pixels in an image are likely to have the same class. It works by segmenting the image at various scales or resolutions and using the course scale segmentations to guide the finer scale segmentations. In addition to reducing the number of mis-classifications, the SMAP algorithm generally produces segmentations with larger connected regions of a fixed class which may be useful in some applications. The amount of smoothing that is performed in the segmentation is dependent of the behavior of the data in the image. If the data suggests that the nearby pixels often change class, then the algorithm will adaptively reduce the amount of smoothing. This ensures that BMU Bonn / Berlin, Oktober 2006 excessively large regions are not formed.

20 KOMPSAT 2 5 classes BMU Bonn / Berlin, Oktober 2006

21 KOMPSAT 2 5 classes, 8 sub-classes, window 512 BMU Bonn / Berlin, Oktober 2006

22 SPOT 5 BMU Bonn / Berlin, Oktober 2006

23 GMES SDI INSPIRE Applied EO INFORMATION GOVERNANCE EU CITIZENS THANK YOU Gedas Vaitkus AGI ARC, Kaunas, LT