Heating method shows considerable potential for mitigating frost heave of subgrade in cold regions. However, the water-heat-deformation characteristics of subgrade under the coupling effect of freezing-thawing and heating effect remain unclear, which hampers the optimization and widespread application of heating method. Therefore, this paper proposes a numerical model of subgrade water-heat-deformation considering heating effect. The influence and mechanism of heating effect on water-heat-deformation of subgrade is systematically analyzed. The results show that the heating effect changes the water-heat-deformation state of subgrade. Furthermore, the combined influence of shady-sunny slope effect and ballast layer ensures that ground temperature near the subgrade center remains above 0 degrees C, thereby preventing the formation of ice lenses and frost heave. However, the shoulders on both sides enter a freezing state, and freezing rate, freezing depth and frost heave are reduced by more than 45 %, 60 % and 60 % respectively compared with the comparison subgrade. The freezing depth, driving force and rate of water migration are significantly affected by heating effect, which increases the pathways of water upward migration and greatly weakens the segregated frost heave of subgrade. This is the primary mechanism through which the heating method effectively mitigates frost heave in subgrades.

Dynamical downscaling generally performs poorly on the Tibetan Plateau (TP), due to the region's complex topography and several aspects of model physics, especially convection and land surface processes. This study investigated the effects of the cumulus parameterization scheme (CPS) and land-surface hydrology scheme (LSHS) on TP climate simulation during the wet season using the RegCM4 regional climate model. To address these issues and seek an optimal simulation, we conducted four experiments at a 20 km resolution using various combinations of two CPSs (Grell and MIT-Emanuel), two LSHSs (the default TOPMODEL [TOP], and Variable Infiltration Capacity [VIC]). The simulations in terms of 2-m air temperature, precipitation (including large-scale precipitation [LSP] and convective precipitation [CP]), surface energy-water balance, as well as atmospheric moisture flux transport and vertical motion were compared with surface and satellite-based observations as well as the ERA5 reanalysis dataset for the period 2006-2016. The results revealed that the model using the Grell and TOP schemes better reproduced air temperature but with a warm bias, part of which could be significantly decreased by the MIT scheme. All schemes simulated a reasonable spatial distribution of precipitation, with the best performance in the experiment using the MIT and VIC schemes. Excessive precipitation was produced by the Grell scheme, mainly due to overestimated LSP, while the MIT scheme largely reduced the overestimation, and the simulated contribution of CP to total precipitation was in close agreement with the ERA5 data. The RegCM4 model satisfactorily captured diurnal cycles of precipitation amount and frequency, although there remained some differences in phase and magnitude, which were mainly caused by the CPSs. Relative to the Grell scheme, the MIT scheme yielded a weaker surface heating by reducing net radiation fluxes and the Bowen ratio. Consequently, anomalous moisture flux transport was substantially reduced over the southeastern TP, leading to a decrease in precipitation. The VIC scheme could also help decrease the wet bias by reducing surface heating. Further analysis indicated that the high CP in the MIT simulations could be attributed to destabilization in the low and mid-troposphere, while the VIC scheme tended to inhibit shallow convection, thereby decreasing CP. This study's results also suggest that CPS interacts with LSHS to affect the simulated climate over the TP.