Kathmandu - Now and then with optical images - Urban detection - Continued
|  | Housing in Kathmandu | |
Detection of urban areas
Detection of urban areas is sometimes not that easy. Reflection from urban areas may be similiar to reflection from the surrounding environment.
This phenomenon appears in the satellite images of Kathmandu, due to the natural materials used to build the city. Most buildings are constructed using timber and clay.
Because parts of what look like pretty meagre vegetation are actually built-up areas, the city and its surroundings are difficult to distinguish.
However, remote sensing does provide some ways of solving this problem. The arithmetic calculation Landsat band 1 - Landsat band 4, for example, leads to a very strong reflection of developed areas and a lesser reflection of undeveloped ones.
This is due to the strong reflection of urban features in the blue band 1, which leads to bright areas, and the slight reflection of these surfaces in the near infrared band 4, which leads to dark areas in the image. Subtraction of these two bands produces an image with accentuated urban features and absorbed non-developed surfaces.
| | | New calculated image (unstretched) |
We will try this in the following exercise and see if we are able to identify the urban area of Kathmandu.
Be aware, however, that this is not universally valid. For precise, high quality products, many manual working hours have to be put into the work. And sometimes this is not even sufficent to obtain a satisfactory final result.
Open the images Kathmandu_Landsat_2001_Band_1.tif and Kathmandu_Landsat_2001_Band_4.tif.
Choose Image>Arithmetic... and select the image Kathmandu_Landsat_2001_Band_1.tif for "Image #1" and Kathmandu_Landsat_2001_Band_4.tif for "Image #2". Select the Operation "Subtract" and the Modifiers "Scale pixel values".
This arithmetic operation leads to an empirical index which is very sensitive to urban features. But it is still hard to seperate built-up areas from soil. The image needs to be enhanced in order to seperate both classes.
| | New calculated image (stretched) | |
Choose Enhance>Interactive Stretching. A histogram will appear. Shift the left blue bar in the Input Histogram to 200. Then shift the right red bar to 220 in the Input Histogram. Now the urban area is clearly seperated from its surroundings.
Save the image as Kathmandu_2001_1sub4 (TIF) in your Kathmandu folder.
The enhanced image is now very useful to detect built-up areas.
We also need the arithmetically subtracted images of Kathmandu in 1989 and 1976 for the following exercise. To obtain them, repeat this exercise but use the images Kathmandu_Landsat_1989_Band_1.tif,
Kathmandu_Landsat_1989_Band_4.tif,
Kathmandu_Landsat_1976_Band_7.tif,
Kathmandu_Landsat_1976_Band_4.tif, and save the enhanced images.
| 
