[Contrastive analysis on soil alkalinization predicting models based on measured reflectance and TM image reflectance].

Based on the monitored data of soil pH and measured Vis-NIR reflectance on spot in Qitai oasis alkalinized area in Xinjiang, as well as comparison of the relationship between measured reflectance and soil pH and the relationship between TM reflectance and soil pH, both of the reflectance multivariate linear regression models were built to evaluate soil alkalinization level, and the model accuracy of pH fitting was discussed with error inspection of post-sample. The results showed that there is a significant positive correlation between soil pH and reflectance. With pH rising the reflectance increased concurrently. So the alkalinization soil characterized by hardening had good spectral response characteristics. Both measured reflectance and TM image reflectance had good potential ability for change detection of the alkalinization soil. The pH predicting model of measured reflectance had higher accuracy and the major error was from different hardening state. If building model by TM reflectance directly, the accuracy of fitting was lower because of the vegetation information in image spectrum. With the vegetation factor removed with NDVI, the accuracy of TM predicting model was near the accuracy of measured reflectance predicting model, and both of the model levels were good.