Page12345

Unsupervised classification of Kathmandu

The aim of the following exercise is to learn more about Kathmandu using image classification. You might think it is just a question of pushing a button, but this is not the case. It requires research, knowledge and accuracy.

During the following exercises we will apply different classification methods. The aim is to understand the principles of image classification and compare the results produced by different methods.

Classification is a very useful tool used to retrieve information for planning, control and cartographic updates. It is a comparatively cheap and easy way to obtain information on land cover, land use and land changes, especially in remote or inaccessible areas. However, even in well-known parts of the world, satellite images are part of our daily lives. Just think of the daily TV weather forecast and its cloud animations. A good example of the importance of classifications is the European Union.

The farmers in the EU are monitored by satellites, and the classifications show the allotment of used and fallow land to the Agrarian Ministry. Financial grants are given in proportion to unused land. How could we monitor the statements made by farmers if we did not have satellites?

Satellite images and classification maps are also used in cartography. Just think of how accurately forest boundaries are drawn in topographical maps. They are taken from satellite images and their image processing products.

Any digital classification, be it supervised or unsupervised, is only a basis for further adaptation. There is no way of producing an accurate and useful classification without manual input. The information processing carried out by the human brain allows the application of more complex procedures than a computer programme. There are many complex relations between different types of surfaces which are not covered by spectral or geometrical differences.

We will only produce high-level classifications of Kathmandu during the following exercises, due to the enormous complexity involved. We will start with a manual exercise to point out the operation method and to get an overview of the city's surface types. You will also obtain a nice hand-drawn map you could call your personal artwork.

This map will be overlayed with a digital classification, so as to compare the results from the manual and low-level digital classification. To be able to overlay our manual classification on the digitally classified image, the hand-drawn map needs to be georeferenced. The easiest way to do this is to include Ground Control Points in the false-colour image 4,5,3 and copy them to the tracing paper.

In LEOWorks, open the Kathmandu_Landsat_Band_453.tif image produced in the False-Colour Combination exercise.

We will use the GIStool in LEOWorks to include the Ground Control Points. Zoom into the image until it fills the screen. Then select GIS from the tool bar. A pop-up menu opens. Choose File>New Theme, name the theme "GCP" and select Polyline. To start drawing, select Edit>Start Edit.

Print the image. Choose Print... change the scale to 100 000 and print the image.


Manual Classification

The next step is to classify the image. In general you are free to select classes of your own choice. However, the following is to assist you to select useful classes:
 
 

	Manual High Level Classification of Kathmandu

High Level Classes

• Rivers
• Densely built-up areas
• Less densely built-up areas
• Airport runways
• Forest & shrub
• Grass & crop land

An easy way to get a map showing Kathmandu is to search for the city using http://maps.google.com.

Download the Proba images as a ZIP-file.

For high resolution images, ESA provides its Proba images of Kathmandu. The ground resolution of the images is 5 metres.

ESA's Project for On-Board Autonomy (Proba) spacecraft is one of the most advanced small satellites ever flown in space. Launched in 2002, this technology demonstration mission performs autonomous guidance, navigation, control, onboard scheduling and payload resources management. Its payload includes a compact multi-spectral imager and high-resolution camera used to acquire Earth observation data.

For more information, see ESA's Proba website.
 
 

Unsupervised Classification of Kathmandu using a thermal image

If you have difficulty locating the rivers, use the thermal Kathmandu_Landsat_2001_Band_61.tif image. Water warms up much more slowly than soil, and appears darker in a thermal image which measures the Earth's temperature.

The thermal image needs to be enhanced in order to get a clearer view of the area in the image. Choose Enhance>Interactive Stretching. A histogram will appear. Shift the left blue bar in the Input Histogram to the left start point of the Input Histogram. Then shift the right red bar to the right start point of the Input Histogram and click Apply.

Now you should have enough good material to classify the false-colour image. Use different colours and textures to trace and fill in the various image features. Do not go into too much detail. The goal of this exercise is to help you understand the classification method.

To use the drawing again later, scan it and save it as "HLC_Kathmandu.tif" in your Kathmandu folder.

Page12345