The Usefulness of Radarsat for Monitoring Land Use Change in the Multimedia Super Corridor, Malaysia
Results and Discussion
Spatial Analysis
Maps of the distribution of terrain classes are presented in figures 1 and figures 2. Figure 1 was produced based on the topography data alone by applying rasterization while Figures 2 illustrated the 3 D of digital elevation model.
Figure 1: DEM extracted from topographic map
Figure 2: 3D-DEM of the study area
Some spatial effects of incorporating geomorphometric parameters in the classification are immediately obvious when comparing the two maps. The most evident is in the spatial distribution of elevations where it varied from the east to west with highest environment (300m- 450m) concentrated on the north-east of study area with 300 to 400m . A similar findings are observed on Figure 2 which it is clear that the vegetation covers are concentrated on the lower limits of elevation. This is located on the north-west of study area. Figure 3 shows the DEM results overlaid with textures analysis's results of Radarsat. It is observed that the vegetation covers are around urban zone and highway are compared to south-west area of the valley. These results are similar to the supervised classifications results (Figure 4). Using geomorphometric parameters to discriminate classes, the problem is eliminated and water/lakes are correctly mapped only in the valleys. However, areas along the plains are spectrally similar to those in the valleys due to the vegetation composition which is primarily virgin forest and palm oil, despite this clearly illustrates that topographic information is a necessary if the class is to be mapped successfully in this region.
Figure 3: DEM overlaid with textures analysis
Another variation between the two maps (Figure 1 and Figure 4) occurs in the northern part in area referred as the urban zone as housing, highway and house court. The highest elevation in the area is 300m. This is far below the lower limit of elevation for an area to be classified as building in this area. When elevation and the other geomorphometric parameters are included are included in the classification, this area is correctly identified as urban area.
Figure 4: Supervised classification of Radarsat image
A third observable difference between the two maps is in the level of homogeneity of classes. In Figure 4, small local variations in supervise3d classifications show up as different classes. The standard deviation of vegetation cover was found less in Sri Serding as compared with urban building (Table 1).
Table 1. Statistical Classes of Supervised Classifications
| Classes |
Standard Deviation |
| Vegetation |
50.50 |
| Forest |
34.60 |
| Urban |
52.94 |
| Lakes |
30.23 |
Moulton (1989) investigated the vegetation covers change on satellite Remote Sensing data by more advanced sensor. Environmental variables related to terrain features were found to have a significant effect on the data because changes in elevation or topographic relief was presented in the area investigated. Further 5the results above are similar to results obtained by Collins (1975) and Crawley (1974). They found that DEM elevation can be easily generated from topographic map information.
