Detecting surface freeze/thaw states in Northeast China with passive microwave data using an improved standard deviation method

Passive microwave remote sensing datasets are widely used to observe surface freeze/thaw (F/T) states. However, current algorithms are highly affected by snow cover and complex land cover types, compromising their performance. Therefore, this study proposes an improved algorithm for daytime detection of diurnal F/T states by using Advanced Microwave Scanning Radiometer 2 data. In the daytime F/T discrimination algorithm, a microwave spectral gradient index is applied to divide the surface into snow-covered and snow-free areas. In the snow-free area, the surface temperature index is optimised to improve the accuracy of the standard deviation method (SDM) in evaluating the accuracy of the F/T state. For the nighttime dataset, the microwave standard deviation index difference values between day and night are used to detect the F/T states based on the daytime results. The accuracy of the improved algorithm reaches 88.6% and 84.5% in the daytime and at nighttime, respectively. Compared with the SDM, the accuracy is improved by 10.2% in the daytime and 5.4% at nighttime. The results demonstrate that the proposed model is able to effectively distinguish the F/T states of snow-covered surfaces. Optimising the surface temperature index can significantly improve the accuracy of the SDM. The results reveal that the proposed surface F/T detection algorithm can be applied to regions with complex land cover types.