Evaluation of the regression and selective principal component analysis methods for change...

Evaluation of the regression and selective principal component analysis methods for change detection study of salinity and land use in siahkooh- playa using remote sensing data

SHARE

A. R. Khavaninzadeh
Yazd University, Iran

Abstract
Using two series data of Landsat TM &ETM+, acquired on 1990 and 2000, changes of expansion of salinity and agriculture lands, have been investigated in Sihahkooh playa in central province of IRAN. These changes were evaluated in 250000 Ha. Relations between total anion and kations in Surface soil and satellite data were evaluated by regression method and suitable models were Introduced. Using these models and applying them in two periods of satellite data, changes trend of study area were investigated. Also using selective principal component analysis method change regions were identified. For this mean it was showed that false color composite of PC2 in 3,4,7 corresponding bands of two periods data are suitable for unsupervised classification and identifying of change regions and land evaluation. Results showed that we can achieve a suitable information about change regions and trend of expansion of salinity and desertification using selective principal component analysis whilst cost and time related to regression method were minimize. Also in the regression method we can identified the inherent of changes and it is possible for evaluation of change trends as quantitative and qualitative whilst results relation to SPCA method are more accurate.