Regional classification and logistic regression modeling for surface freeze/thaw detection on the Qinghai-Tibet Plateau using CYGNSS data

Surface soil Freeze-Thaw (F/T) transitions play a critical role in climate feedbacks and energy balance. Global Navigation Satellite System Reflectometry (GNSS-R) technology has emerged as a promising tool for detecting surface soil F/T states, particularly through threshold-based methods. However, these methods often struggle in regions with complex environmental conditions, leading to reduced accuracy. To overcome these limitations, this study proposes an innovative approach that integrates the application of GNSS-R data with regional classification and logistic regression modeling, establishing an efficient framework for detecting surface F/T states. The method involves pre-classifying the Qinghai-Tibet Plateau (QTP) into five regions based on characteristic data, followed by logistic regression to predict soil states. Quality-controlled GNSS-R observations from 1 September 2018, to 31 August 2023, were analyzed to compute surface reflectivity, which, along with vegetation index, informed the classification process. This facilitated the generation of 0.1° × 0.1° surface F/T state maps, providing detailed spatial insights into F/T processes. The results demonstrate that GNSS-R effectively captures the spatiotemporal dynamics of F/T processes on the QTP. Compared with ERA5_Land temperature data, the method achieved accuracies of 84.35% for weekly and 83.78% for daily soil states. Against the in-situ measurements, accuracies were 80.25% and 79.18%, respectively. These findings highlight the potential of GNSS-R-derived soil state maps as a valuable complement to existing F/T monitoring products.