Depth distributions of soil temperature: Seasonal sensitivity and simulation across dryness/wetness conditions

Exploring the response patterns of soil temperature to dryness/wetness at various depths enhances our understanding of the complex interactions between surface water dynamics and thermal transfer. Previous studies have focused primarily on analyzing the thermal evolution and response mechanisms of soil temperature with respect to various meteorological factors, neglecting its response across different timescales for dry/wet conditions. This research examines response patterns from four perspectives: trends, seasonal sensitivity, propagation, and response relationships. Based on these patterns, the random forest model was employed to simulate the cumulative soil temperature (CST) during dry and wet periods, using the duration and severity of these periods at different soil depths as input variables. Using long-term monthly observations (1966–2022) from the Wudaogou National Comprehensive Hydrological Observation Station in Anhui Province, China, we analyzed the soil temperature at depths of 0–320 cm paired with meteorological data. The standardized precipitation evapotranspiration index (SPEI) was employed to characterize dryness/wetness across various timescales (from 1-month to 12-months). The results indicate the following: (1) A distinct shift in soil thermal dynamics was observed, with a cooling trend prior to 1992 transitioning into a pronounced warming phase thereafter. The post-1992 warming rate was 3–4 times faster than the overall rate for 1966–2022, revealing critical temporal shifts in soil temperature responses. (2) Shallower soil layers (0–40 cm) exhibited heightened seasonal sensitivity to dryness/wetness, responding more rapidly and intensely than deeper layers (>80 cm), especially at short timescales (≤3 months). This underscores the critical role of surface soil interactions in thermal dynamics. (3) A ‘compression’ phenomenon in temperature transmission was identified, where the influence of dryness and wetness on soil temperature diminishes with increasing depth. Dry periods consistently elevated soil temperatures, whereas wet periods reduced them, providing insights into vertical thermal propagation mechanisms. (4) The random forest model showcased a strong capability to simulate CST achieving R² and Ens values above 0.91 and absolute PBIAS values below 5 %. These findings are essential for managing ecosystems and agricultural practices, as well as informing water management strategies in regions facing extreme climate events.