[Objective] The mechanisms underlying the responses of vegetation coverage measures to hydrothermal patterns and temporal stability of surface soil on Benggang collapsing gully walls were analyzed， in order to provide a theoretical basis and predictive parameters for evaluating the efficacy of ecological restoration of stable Benggang collapsing gully walls. [Methods] Collapsing gully walls in Hetian Town， Changting County， Fujian Province， were subjected to two treatments， namely， planting Paspalum wettsteinii on the gully walls （grass-growing treatment） and maintaining the original bare collapsing gully walls （control）. To compare differences in the spatial distribution characteristics of soil water and heat under the two treatments， as well as indicators of soil moisture temporal stability， we monitored the dynamic changes in soil moisture and temperature in the 0—20 cm soil surface layer. [Results] Throughout the period of monitoring， the average soil temperature under the grass treatment with P. wettsteinii was 20.93 ℃， which was significantly lower than that of the bare control soil （22.26 ℃）. Moreover， the ranges of temperature fluctuation on the upper and lower slopes under grass cover （23.76 ℃ and 21.95 ℃， respectively） were both lower than corresponding values for the bare slope （25.32 ℃ and 26.36 ℃， respectively）. In addition， the annual average soil moisture content under the grass treatment （18.77 g/kg） was approximately 38.11% higher than that of the bare control （13.59 g/kg）. Spearman rank correlation analysis revealed that the mean correlation coefficient （ρ） of soil moisture under grass cover （0.41） was significantly higher than the 0.18 value obtained for bare land， thus indicating enhanced spatial consistency. However， vegetation coverage was found to alter soil moisture cycling， increasing the frequency of dry-wet alternations， as indicated by higher values for the mean relative difference （MRD）， standard deviation， and temporal stability index （ITS） under the grass treatment compared with those of the bare land control. Specifically， in response to the grass treatment， we obtained values of 19.27%， 22.57%， and 29.68% for the average MRD， standard deviation， and ITS respectively， on the downslope， whereas on the upslope， the corresponding values were 10.48%， 12.47%， and 16.29%. In contrast， under the control treatment， we obtained MRD， standard deviation， and ITS values of -13.43%， 16.05%， and 20.93%， respectively， on the upslope， and corresponding values of -16.32%， 9.33%， and 18.80% on the downslope. [Conclusion] By reducing temperature， increasing moisture content， and enhancing spatial similarity， the grass-growing treatment assessed in this study contributed to considerable improvements in the water and heat dynamics of the surface soil on collapsing gully walls. However， this treatment also intensified short-term fluctuations in soil moisture and reduced soil temporal stability.