Approach to Land-use Analysis in Hetao Irrigation Project of Inner Mongolia, China, Based on Satellite Image Data
Materials and Method
As satellite image data, Landsat TM data (spatial resolution: 30m x 30m, path-row: 129-32 , date of observation: 1989/9/17) were used.
The result of image classification was used to analyze land uses in Hetao irrigation district according to the following procedure.
- All satellite image data were geometrically corrected based on the GCP obtained by the field survey (2000/8/11, 12) using the analysis software.
- Subset image around Linhe City was extracted referring to Inner Mongolia Maps (published by Inner Mongolia Map Pub.).
- Supervised and unsupervised classifications were performed with five training classes of water, dune, urban area, vegetation and saline soil (including naked ground).
- From the results of classification, areas were calculated and compared with the results of the field survey.
Table 1 shows the detailed classifications performed in the present study; three classification in total including two supervised classifications and one unsupervised classification.
Table 1 Decision rules
| | Parametric Decision Rule | Non-Parametric Decision Rule | Other Decision Rule |
| Supervised classification 1 | Maximum Likelihood | Multi Level Slice | × |
| Supervised classification 2 | Maximum Likelihood | × | × |
| Unsupervised classification | × | × | ISODATA Method |
Fig.3 Signatures in Linhe
Fig4 Reflection characteristics(CCT value)
In the supervised classification 1, parametric most-likely classification method was employed for unclassified areas following the nonparametric classification. In the supervised classification 2, only parametric most-likely classification was performed. For unsupervised classification, ISODATA method was used.
Fig. 3 shows the areas extracted as training data for dune, salinized soil (including naked soil), vegetation, water and urban area based on the satellite image of the area. All the areas underwent ground-truth in the present survey.
Fig. 4 shows the mean spectral reflection characteristics of the respective bands in the training area.
