The Tibetan Plateau (TP) region experiences strong land‐atmosphere interactions, and as an elevated heating source, significantly influences the formation of the Asian monsoon. Those interactions are not well represented in current land‐surface models (LSMs), partly due to difficulties in representing heterogeneities in soil structures in LSM. Simulations using the Noah with multiparameterization options (Noah‐MP) LSM are employed to assess the relative importance of parameterizing vertical soil heterogeneity, organic matter, and soil rhizosphere and their impacts on seasonal evolution of soil temperature, soil moisture, and surface energy and water budgets at the sparsely vegetated Amdo site located in central TP. The LSM spin‐up time at the central TP depends on the complexity of the model physics, ranging from 4 years with simplest soil physics to 30 years with the addition of organic matter and spare to dense rhizosphere parameterization in Noah‐MP. Representing layered soil texture and organic matter does not improve low biases in topsoil moisture. Reducing the saturated conductivity from the mucilage in the rhizosphere produces better results. Surface sensible and latent heat fluxes are better simulated in the monsoon season as well. Adding layered soil texture and organic matter in Noah‐MP retard the thawing in deep soil layers, and the rhizosphere effect delays thawing even more in the transient season. Uncertainties in soil initialization significantly affect deep‐soil temperature and moisture, but uncertainties in atmospheric forcing conditions mainly affect topsoil variables and consequently the surface energy fluxes. Differing land‐surface physics cause 36% uncertainty in the accumulated evapotranspiration and subsurface runoff.

Precipitation is a critical input to land surface and hydrology modeling and prediction. Dynamical downscale modeling has added value to representing precipitation, when compared with the performance of coarse-resolution reanalysis and global climate models, over the Tibetan Plateau (TP). Convection-permitting modeling (CPM) may even outperform dynamical downscale models (DDMs). In this study, 4-km CPM results were compared to 28-km DDM results for a snow season (1 October 2013–31 May 2014) over the TP. The CPM- and DDM-simulated precipitation, as well as three merged gridded precipitation datasets, were evaluated against in situ observations below 4800 m. The five precipitation datasets (CPM, DDM, CMFD, COPRPH, and TRMM) showed large differences over the TP with underestimation of TRMM and overestimation of CPM and DDM compared to observations. The most significant difference occurred in the Brahmaputra Grand Canyon. Given the substantial uncertainty in observed precipitation at high mountains, snow cover simulated by a high-resolution land data assimilation system was used to indirectly evaluate the above precipitation data using MODIS observations. Simulated snow-cover fraction was greatly underestimated using all the merged precipitation datasets. However, simulations using the DDM- and CPM-generated precipitation as input outperformed those using other gridded precipitation data, showing lower biases, higher pattern correlations, and closer probability distribution functions than runs driven by the merged precipitation. The findings of this study generally support the assumption that high-resolution CPM-produced precipitation has an added value for use in land surface and hydrology simulations in high-mountain regions without reliable in situ precipitation observations.

Snow cover over the Tibetan Plateau (TP) plays an important role in Asian climate. State‐of‐the‐art models, however, show significant simulation biases. In this study, we assess the main uncertainty associated with model physics in snow cover modeling over the TP using ground‐based observations and high‐resolution snow cover satellite products from the Moderate Resolution Imaging Spectroradiometer (MODIS) and FengYun‐3B (FY3B). We first conducted 10‐km simulations using the Noah with multiparameterization (Noah‐MP) land surface model by optimizing physics‐scheme options, which reduces 8.2% absolute bias of annual snow cover fraction (SCF) compared with the default model settings. Then, five SCF parameterizations in Noah‐MP were optimized and assessed, with three of them further reducing the annual SCF biases from around 15% to less than 2%. Thus, optimizing SCF parameterizations appears to be more important than optimizing physics‐scheme options in reducing the uncertainty of snow modeling. As a result of improved SCF, the positive bias of simulated surface albedo decreases significantly compared to the GLASS albedo data, particularly in high‐elevation regions. This substantially enhances the absorbed solar radiation and further reduces the annual mean biases of ground temperature from −3.5 to −0.8°C and snow depth from 4.2 to 0.2 mm. However, the optimized model still overestimates SCF in the western TP and underestimates SCF in the eastern TP. Further analysis using a higher‐resolution (4 km) simulation driven by topographically adjusted air temperature shows slight improvement, suggesting a rather limited contribution of the finer‐scale land surface characteristics to SCF uncertainty.

Soil hydrothermal regime of the active layer in the permafrost regions of the Qinghai‐Tibet Plateau (QTP) is important to the underlying permafrost and the climate change dynamics in Asia. However, a large bias still exists in current land surface models in the representation of soil temperature and moisture. This study assessed and augmented the Noah land surface model with multiparameterization options (Noah‐MP) for simulating soil hydrothermal dynamics at the Tanggula (alpine meadow) and Beiluhe (alpine swamp) stations located in the permafrost regions of the QTP. The results showed that the default Noah‐MP tended to underestimate soil temperature and moisture. Specifically, the default model overestimated the snow depth and duration due to the low snow sublimation rate. This resulted in a cold deviation in the soil temperature at two stations. Such underestimation was reduced by introducing a scheme that considered the sublimation loss from wind. Moreover, the remaining cold bias in the soil profiles of two stations was greatly resolved by a combined scheme of roughness length for heat (Z_0h) and undercanopy aerodynamic resistance (r_a,g). A soil thermal conductivity scheme, which can produce more realistic soil thermal conductivity in frozen soil, further improved the deep soil temperature simulation. The consideration of soil organic matter could mitigate the underestimation of the shallow soil moisture to some extent, but this improvement was more obvious at the Tanggula station, which had coarser mineral soil than the Beiluhe station.

To meet the requirement of high-resolution datasets for many applications, a dynamical downscaling approach using a regional climate model (the WRF Model) driven by a global climate model (CCSM4) has been adopted. This study focuses on projections of future moisture flux changes over the Tibetan Plateau (TP). First, the downscaling results for the historical period (1980–2005) are evaluated for precipitation P, evaporation E, and precipitation minus evaporation P − E against Global Land Data Assimilation System (GLDAS) data. The mechanism of P − E changes is analyzed by decomposition into dynamic, thermodynamic, and transient eddy components. Whether the historical period changes and mechanisms continue into the future (2010–2100) is investigated using the WRF and CCSM model projections under the RCP4.5 and RCP8.5 scenarios. Compared with coarse-resolution forcing, downscaling was found to better reproduce the historical spatial patterns and seasonal mean of annual average P, E, and P − E over the TP. WRF projects a diverse spatial variation of P − E changes, with an increase in the northern TP and a decrease in the southern TP, compared with the uniform increase in CCSM. The dynamic component dominates P − E changes for the historical period in both the CCSM and WRF projections. In the future, however, the thermodynamic component in CCSM dominates P − E changes under RCP4.5 and RCP8.5 from the near-term (2010–39) to the long-term (2070–99) future. Unlike the CCSM projections, the WRF projections reproduce the mechanism seen in the historical period—that is, the dynamic component dominates P − E changes. Furthermore, future P − E changes in the dynamical downscaling are less sensitive to warming than its coarse-resolution forcing.