Thus for vegetation, absorption in the red band is related to the reasons of chlorophyll. While reflectance in the NUR and depends on the leaf structure. Wet soils have a lower response in all bands than dry soils.
Figure 2. Simplified spectral reflecance curves for five surface in the visible and near infrared parts of the electromagnetic spectrum. Example channels of SPOT, Landsat and Noaa sateilite sensors are shown
Spectral classification techniques can be grouped into supervised and unsupervised classification. In out case of study we utilize the maximum likelihood supervised classification which is consist of two steps.
- The user defines training areas of each these or class to be classified
- The system then computer the values of all pixels of the image and allocate according to the maximum likelihood rule each pixels of the image to one of thematic classes defined by the values of the corresponding training areas.
4.2 Land suitability Mapping
Land suitability evaluation is the process of assessing the suitability of land for specific use. These may be major kinds of land use such as rainfed agriculture, livestock production and aforestations. The topographic characteristics, the climatic conditions and the soil quality of an area are the most important determinant parameters of the land suitability evaluations. In our study case we applied the rules used by the food and agriculture organization FAO published on the frame work of land evaluation.
Geographic information system methodology was used to evaluate the land suitability. It can be defined as a digital processing system of georeferenced data it's main functions concerning; an automatic editing of existing maps, provides the possibility, after digitizing of boundaries and themes on existing maps to produce derived maps by digital updating of boundaries, extracting selected themes or by superimposing different maps together in a desired projection and scale, and the possibility of combining different levels of geographical data using logic and arithmetic operators. These capabilities allow the construction of models from which a new thematic maps (e.g. land suitability map) can be produced from a set of thematic maps (e.g climate, soil topography and land cover).
Three land suitability maps were produced using GIS methodology:
- Land Suitability for annual crops.
- Land suitability for annual crops.
- Land suitability for agriculture use (synthetic result)
5. Digital Images Processing
5.1 Land use
The digital image processing used for land - use mapping are as follows:
- Geometric correction of the SPOT XS image using a top[ographic map scale 1/50000.
- Digitizing and registration of the aerial photographic to SPOT image
- Integration SPOT panchromatic image with the SPOT XS to produce a natural color image (XS + P).
- Producing the following indexes:
- Vegetation index (VI) = XS3 = XS2/XS3+XS2
- BRINGTHNESS INDEX (SI)2 = XS3)2 + (XS2)2
- COLOR INDEX (CI) = XS2 - XS1/XS2+XSI
- Enhancement of images :
- Stretching
- Filtering
- Ratio
- Principal component analysis
- Supervised classification using the maximum likelihood rule
Our methodology of supervised classification consisted of the following steps:-
- Introducing the "Training Samples" and using the 3 band of spot images to generate the signature file for each class.
- Masking the vegetation cover using the vegetation index on one theme mask.
- Applied the maximum likelihood classification in the previous masking area at this stage we distinguished the following classes:-
1 - Forest 2 - Orchards 3 - Vegetables 4 - Rangeland
- Masking the urban area of manual classification on the masked urban area, this allowed us to distinguish three classes.
- Applied the maximum likelihood classification on the masked urban area, this allowed us to distinguish three classes :-
1 - High dense urban Area 2 - Low dense urban area 3 - No urban area
- Extracting specifics urban area by manual interpretation :
1 - Industrial area 2 - Educational Area 3 - Transportation area 4 - Roads network
- Applied the maximum likelihood classification to the unclassified area (not masked) to find out the other classes
- Grouping all of classes in one file.
- Generalizing the classification to remove out the uncorrected classified pixels.
