A Monitoring Method of Land Cover/Land Use Change in Naiman,
Inner Mongolia Autonomous Region, China Using Landsat Data
An extraction method of land cover changes
At first, we attempted to monitor the desertification process through land cover changes using Landsat data, and we identified decertifying regions by the following methods:
First, a vegetation index was used to identify unvegetated regions.
For Landsat TM data (TM 4 - TM 3) / (TM 4 + TM 3)
For Landsat MSS data (MSS 7 - MSS 5) / (MSS 7 + MSS 5)
Low values of this index represented unvegetated areas. Comparisons were made with composite images to derive the threshold, then unvegetated areas were identified. In addition, to consider seasonal in vegetation, we used autumn and spring data and common areas were identified.
Second, water bodies and non-made structures such as settlements, which were included in the non-vegetated regions, had to be removed. Fro that purpose, the ratio, (TM 5-TM 1)/ (TM 5 + TM 1), was obtained. Since bare land value in this band ratio was higher than that of water bodies and non-made structure, it was possible to separate the two. Therefore, the index derived from this ratio was designated as the structure index. Since there was no MSS sensor corresponding to the TM band 5, old data, were masked by water bodies and non-made structures derived from the new images.
Third, using the redness index (TM 3 - TM 1) / (TM 3 + TM 1) which reflects the amount of oxidized iron contained in the ground, desert areas were identified by the ground color in each region.
Finally, desertified areas identified based on data from different years were superimposed to obtain yearly changes.
Next, we verified the application of this method in the model district of Naiman A 30 km2 section was selected as the model district, and was analyzed for desertification patterns over the last 10 years. Decertifying regions have low vegetation indices. White-colored areas corresponding to the soil color of these areas were identified and the results from analysis of old and new images were superimposed to analyze the changes (Fig. 2).
Fig. 2 Composite photo of Landsat TM (left) and recent trend of desertification (right) in
Naiman
The areal proportions of the 3 model districts undergoing desertification were roughly 40%. However, the area of reclaimed land and the area of newly decertifying land were roughly 12% in the districts. This fact indicated that the areal extent of desertification was almost constant.
In Naiman, the areas around the settlements and around the N-S-running road and railway in the eastern part of the town showed some recovery from 1982-1991, in contrast to outlying areas, where desertification continued.
The above results indicated that desertification does not proceed unilaterally regardless of the state of degradation; rather, due to some of the measures implemented the progressing of desertification could be contained.
