Detection of land surface freeze-thaw status on the Tibetan Plateau using passive microwave and thermal infrared remote sensing data

Abstract

The freeze/thaw (F/T) cycle plays an important role in climate change and ecology research. Currently, soil F/T monitoring is restricted by low satellite spatial resolution or a relative long revisit cycle, which is one of the main problems in improving F/T monitoring resolution using available satellite data. Because temperature is a key parameter in determining soil F/T status, in this study, relatively high-resolution merged land surface temperature data were obtained using the Bayesian Maximum Entropy (BME) method by blending LSTs retrieved from passive microwave and infrared remotely sensed data. The merged temperature data were then used to downscale the passive microwave brightness temperature from 0.25° to 0.05°. Finally, the merged temperature and downscaled brightness temperature data were applied to discriminate the surface freeze/thaw status. A comparison with in situ data turned out that the downscaled brightness temperature could be used to determine soil F/T status with a total classification accuracy higher than 80%. The total freeze/thaw classification accuracy using merged temperature data was only 59.7%, which can be attributed to the temperature difference between the land surface and soil. After the adjustment with a relationship between soil temperature and land surface temperature, the classification accuracy reached 89.7%.