 | | | | |
| | | | | |
|
Coastal change Danube Delta change detectionOil spillsDeforestation Bardia National ParkCongo River BasinKameng-Sonitpur Elephant ReserveKilimanjaroRondoniaShillong and GuwahatiIce Antarctica 2003Climate change and glaciersGlacier analysis using radar imageryGlacial retreat in the AlpsGlacier Ice FlowMonitoring of glaciers in the HimalayasRemote sensing of ice and snowUrbanisation CairoCity of KathmanduCórdobaHimalayasKathmandu ValleyLagosVegetation Annapurna Conservation AreaLost in the AndesNgorongoro Conservation AreaNiger Inland DeltaVegetation in South America
| | | | | | | Bardia Region Overview - Image processing - Part 7
This exercise is divided into eight parts and requires the use of LEOWorks.
Supervised Classification
Supervised classification is based on a statistical method using training fields representing the different land cover classes. Each land cover class has its own specific spectral fingerprint. The supervised classification method integrates human knowledge in selected training fields of specific land cover classes. At least one training field should be selected for each land cover type to be classified. LEOWorks has facilities to evaluate training fields and allocates each image element (pixel) to one of the given land cover classes.
More than one training field can be selected for each land cover type to make the classification result more accurate. Consult the following picture for the different land cover and its location.
Land cover of western part of Bardia National Park (2002)
We will use the whole spectrum of Landsat bands. To do this, open the following images:
- Bardia_landsat_2002_band_1.tif
- Bardia_landsat_2002_band_2.tif
- Bardia_landsat_2002_band_3.tif
- Bardia_landsat_2002_band_4.tif
- Bardia_landsat_2002_band_5.tif
- Bardia_landsat_2002_band_7.tif
….and to select the training fields, open:
- Bardia_landsat_2002_and_453.tif
Click Bardia_landsat_2002_band_453.tif and choose Multivariate Analysis>Supervised Classification>Select Training Fields.
Select Draw Polygon from the toolbar.
Start with the Sal forest (Shorea robusta) for training field selection. Draw a polygon in the north western part, and name the class 'Sal Forest'. Draw some more polygons in the south and the east, and name the class 'Sal Forest'. Do the same with all the other classes.
The land cover classes are:
- Sal Forest
- Mixed Hardwood Forest
- Dry River Bed (white in the Landsat image)
- Flood Plain Grassland
- Riverine Forest
- Wooded Grassland
- Agricultural Field
- River
- Shadow (mainly Hill Deciduous Forest)
- Hill Deciduous Forest
Choose Multivariate Analysis>Supervised Classification>Maximum Likelihood and select all bands of the 2002 Landsat image.
Set the Threshold Value to "5%".
Here the colours of the classification results should be changed again for easier interpretation.
Choose Image>Add Legend and check the combination of colours and class names.
Choose Edit>Look Up>Table and select the colour for the class "Sal forest". Change the colour to a dark green. Change the other colours in the same way.
Save the classification as "Bardia_Landsat_2002_maximlike.tif" in your folder Bardia.
1. Have a close look at the classified image and describe your impression.
2. How does the classified image fit with the false colour images?
3. Which land cover is very well represented, and which are less well represented?
4. Compare the supervised and the unsupervised classification. Describe the advantages of the supervised classification. Which parts in particular are better?
5. Which type of land cover is missing or under represented in the supervised classification?
| |
| | Bardia National Park IntroductionLocationGeology and soilWeather and climatePeople and settlementsExercises Worksheet introductionOverview - working on paper printsBardia Region Overview - Image processingLandscape and land cover dynamics in the Karnali FloodplainHabitat suitability evaluation for rhinoceros in BardiaEduspace - Software LEOWorks 3ArcExplorerEduspace - Download bardia.zipbardia_paperprints.zipResources Useful links
|
Previous
