Energy budget-based distributed modeling of snow and glacier melt runoff is essential in a hydrologic model to accurately describe hydrologic processes in cold regions and high-altitude catchments. We developed herein an integrated modeling system with an energy budget-based multilayer scheme for clean glaciers, a single-layer scheme for debris-covered glaciers, and multilayer scheme for seasonal snow over glacier, soil, and forest within a distributed biosphere hydrological modeling framework. Model capability is demonstrated for Hunza River Basin (13,733 km 2 ) in the Karakoram region of Pakistan on a 500 m grid for 3 hydrologic years (2002)(2003)(2004). Discharge simulation results show good agreement with observations (Nash-Sutcliffe efficiency = 0.93). Flow composition analysis reveals that the runoff regime is strongly controlled by the snow and glacier melt runoff (50% snowmelt and 33% glacier melt). Pixel-by-pixel evaluation of the simulated spatial distribution of snow-covered area against Moderate Resolution Imaging Spectroradiometer-derived 8 day maximum snow cover extent data indicates that the areal extent of snow cover is reproduced well, with average accuracy 84% and average absolute bias 7%. The 3 year mean value of net mass balance (NMB) was estimated at +0.04 myr À1 . It is interesting that individual glaciers show similar characteristics of NMB over 3 years, suggesting that both topography and glacier hypsometry play key roles in glacier mass balance. This study provides a basis for potential application of such an integrated model to the entire Hindu-Kush-Karakoram-Himalaya region toward simulating snow and glacier hydrologic processes within a water and energy balance-based, distributed hydrological modeling framework.

Abstract. In this study, the snow physics of a distributed biosphere hydrological model, referred to as the Water and Energy Budget based Distributed Hydrological Model (WEB-DHM) is significantly improved by incorporating the three-layer physically based energy balance snowmelt model of Simplified Simple Biosphere 3 (SSiB3) and the Biosphere-Atmosphere Transfer Scheme (BATS) albedo scheme. WEB-DHM with improved snow physics is hereafter termed WEB-DHM-S. Since the in-situ observations of spatially-distributed snow variables with high resolution are currently not available over large regions, the new distributed system (WEB-DHM-S) is at first rigorously tested with comprehensive point measurements. The stations used for evaluation comprise the four open sites of the Snow Model Intercomparison Project (SnowMIP) phase 1 with different climate characteristics (Col de Porte in France, Weissfluhjoch in Switzerland, Goose Bay in Canada and Sleepers River in USA) and one open/forest site of the SnowMIP phase 2 (Hitsujigaoka in Japan). The comparisons of the snow depth, snow water equivalent, surface temperature, snow albedo and snowmelt runoff at the SnowMIP1 sites reveal that WEB-DHM-S, in general, is capable of simulating the internal snow process better than the original WEB-DHM. Sensitivity tests (through incremental addition of model processes) are performed to illustrate the necessity of improvements over WEB-DHM and indicate that both the 3-layer snow module and the new albedo scheme are essential. The canopy effects on snow processes are studied at the Hitsujigaoka site of the SnowMIP2 showing that the snow holding capacity of the canopy plays a vital role in simulating the snow depth on ground. Through these point evalua-Correspondence to: M. Shrestha (maheswor@hydra.t.u-tokyo.ac.jp) tions and sensitivity studies, WEB-DHM-S has demonstrated the potential to address basin-scale snow processes (e.g., the snowmelt runoff), since it inherits the distributed hydrological framework from the WEB-DHM (e.g., the slope-driven runoff generation with a grid-hillslope scheme, and the flow routing in the river network).

Snow and frozen soil are important factors that influence terrestrial water and energy balances through snowpack accumulation and melt and soil freeze-thaw. In this study, a new land surface model (LSM) with coupled snow and frozen soil physics was developed based on a hydrologically improved LSM (HydroSiB2). First, an energy-balance-based three-layer snow model was incorporated into HydroSiB2 (hereafter HydroSiB2-S) to provide an improved description of the internal processes of the snow pack. Second, a universal and simplified soil model was coupled with HydroSiB2-S to depict soil water freezing and thawing (hereafter HydroSiB2-SF). In order to avoid the instability caused by the uncertainty in estimating water phase changes, enthalpy was adopted as a prognostic variable instead of snow/soil temperature in the energy balance equation of the snow/frozen soil module. The newly developed models were then carefully evaluated at two typical sites of the Tibetan Plateau (TP) (one snow covered and the other snow free, both with underlying frozen soil). At the snow-covered site in northeastern TP (DY), HydroSiB2-SF demonstrated significant improvements over HydroSiB2-F (same as HydroSiB2-SF but using the original single-layer snow module of HydroSiB2), showing the importance of snow internal processes in three-layer snow parameterization. At the snow-free site in southwestern TP (Ngari), HydroSiB2-SF reasonably simulated soil water phase changes while HydroSiB2-S did not, indicating the crucial role of frozen soil parameterization in depicting the soil thermal and water dynamics. Finally, HydroSiB2-SF proved to be capable of simulating upward moisture fluxes toward the freezing front from the underlying soil layers in winter.

In this study, a distributed biosphere hydrological model with three-layer energy-balance snow physics [an improved version of the Water and Energy Budget-based Distributed Hydrological Model (WEB-DHM-S)] is applied to the Dudhkoshi region of the eastern Nepal Himalayas to estimate the spatial distribution of snow cover. Simulations are performed at hourly time steps and 1-km spatial resolution for the 2002/03 snow season during the Coordinated Enhanced Observing Period (CEOP) third Enhanced Observing Period (EOP-3). Point evaluations (snow depth and upward short-and longwave radiation) at Pyramid (a station of the CEOP Himalayan reference site) confirm the vertical-process representations of WEB-DHM-S in this region. The simulated spatial distribution of snow cover is evaluated with the Moderate Resolution Imaging Spectroradiometer (MODIS) 8-day maximum snow-cover extent (MOD10A2), demonstrating the model's capability to accurately capture the spatiotemporal variations in snow cover across the study area. The qualitative pixelto-pixel comparisons for the snow-free and snow-covered grids reveal that the simulations agree well with the MODIS data to an accuracy of 90%. Simulated nighttime land surface temperatures (LST) are comparable to the MODIS LST (MOD11A2) with mean absolute error of 2.428C and mean relative error of 0.778C during the study period. The effects of uncertainty in air temperature lapse rate, initial snow depth, and snow albedo on the snow-cover area (SCA) and LST simulations are determined through sensitivity runs. In addition, it is found that ignoring the spatial variability of remotely sensed cloud coverage greatly increases bias in the LST and SCA simulations. To the authors' knowledge, this work is the first to adopt a distributed hydrological model with a physically based multilayer snow module to estimate the spatial distribution of snow cover in the Himalayan region.

An enthalpy-based frozen soil model was developed for the simulation of water and energy transfer in cold regions. To simulate the soil freezing/thawing processes stably and efficiently, a three-step algorithm was applied to solve the nonlinear governing equations: (1) a thermal diffusion equation was implemented to simulate the heat conduction between soil layers; (2) a freezing/thawing scheme used a critical temperature criterion to judge the phase status and introduced enthalpy and total water mass into freezing depression equation to represent ice formation/melt and corresponding latent heat release/absorption; and (3) a water flow scheme was employed to describe the liquid movement within frozen soil. In addition, a parameterization set of hydraulic and thermal properties was updated by considering the frozen soil effect. The performance of the frozen soil model was validated at point scale in a typical mountainous permafrost basin of China. An ice profile initialization method is proposed for permafrost modeling. Results show that the model can achieve a convergent solution at a time step of hourly and a surface layer thickness of centimeters that are typically used in current land surface models. The simulated profiles of soil temperature, liquid water content, ice content and thawing front depth are in good agreement with the observations and the characteristics of permafrost. The model is capable of continuously reproducing the diurnal and seasonal freeze-thaw cycle and simulating frozen soil hydrological processes.Significant progress has been achieved over recent decades in modeling ground freezing and thawing and coupling these processes to hydrological and land surface models. A semiempirical permafrost model BAO ET AL.AN ENTHALPY-BASED FROZEN SOIL MODEL 5259

In distributed hydrological modeling, surface air temperature (Tair) is of great importance in simulating cold region processes, while the near‐surface‐air‐temperature lapse rate (NLR) is crucial to prepare Tair (when interpolating Tair from site observations to model grids). In this study, a distributed biosphere hydrological model with improved snow physics (WEB‐DHM‐S) was rigorously evaluated in a typical cold, large river basin (e.g., the upper Yellow River basin), given a mean monthly NLRs. Based on the validated model, we have examined the influence of the NLR on the simulated snow processes and streamflows. We found that the NLR has a large effect on the simulated streamflows, with a maximum difference of greater than 24% among the various scenarios for NLRs considered. To supplement the insufficient number of monitoring sites for near‐surface‐air‐temperature at developing/undeveloped mountain regions, the nighttime Moderate Resolution Imaging Spectroradiometer land surface temperature is used as an alternative to derive the approximate NLR at a finer spatial scale (e.g., at different elevation bands, different land covers, different aspects, and different snow conditions). Using satellite‐based estimation of NLR, the modeling of snow processes has been greatly refined. Results show that both the determination of rainfall/snowfall and the snowpack process were significantly improved, contributing to a reduced summer evapotranspiration and thus an improved streamflow simulation.

Abstract. Adequate estimation of the spatial distribution of snowfall is critical in hydrologic modelling. However, this is a well-known problem in estimating basinscale snowfall, especially in mountainous basins with data scarcity. This study focuses on correction and estimation of this spatial distribution, which considers topographic effects within the basin. A method is proposed that optimises an altitude-based snowfall correction factor (C fsnow ). This is done through multi-objective calibration of a spatially distributed, multilayer energy and water balance-based snowmelt model (WEB-DHM-S) with observed discharge and remotely sensed snow cover data from the Moderate Resolution Imaging Spectroradiometer (MODIS). The Shuffled Complex Evolution-University of Arizona (SCE-UA) automatic search algorithm is used to obtain the optimal value of C fsnow for minimum cumulative error in discharge and snow cover simulations. Discharge error is quantified by NashSutcliffe efficiency and relative volume deviation, and snow cover error was estimated by pixel-by-pixel analysis. The study region is the heavily snow-fed Yagisawa Basin of the Upper Tone River in northeast Japan. First, the system was applied to one snow season (2002)(2003), obtaining an optimised C fsnow of 0.0007 m −1 . For validation purposes, the optimised C fsnow was implemented to correct snowfall in 2004, 2002 and 2001. Overall, the system was effective, implying improvements in correlation of simulated versus observed discharge and snow cover. The 4 yr mean of basin-average snowfall for the corrected spatial snowfall distribution was 1160 mm (780 mm before correction). Execution of sensitivity runs against other model input and parameters indicated that C fsnow could be affected by uncertainty in shortwave radiation and setting of the threshold air temperature parameter. Our approach is suitable to correct snowfall and estimate its distribution in poorly gauged basins, where elevation dependence of snowfall amount is strong.

Mountain snowpack and its distribution both have intimate connections to regional hydrology by preserving winter precipitation to sustain streamflows during the summer months. One of the key knowledge gaps in mountainous region is the interplay of precipitation and temperature with changing altitudes. Three‐dimensional temperature distribution is pivotal for the realistic temporal and spatial distribution of precipitation with pattern (rain/snow). The environmental/linear lapse rates are inadequate to address snow processes, resulting in significant uncertainties. An effort is made in this study to develop a vertical profile of temperature (VPT) and apply it as a dynamic temperature lapse rate to curtail uncertainties. The VPT was used for the spatiotemporal bias correction of precipitation by targeting accessible data sources based on the quantitative and spatial analysis in a distributed hydrologic modeling framework with a logical calibration and validation. The water and energy budget‐based distributed hydrological model with snow was utilized to simulate the streamflows and spatial distribution of snow cover based on VPT and corrected precipitation. During calibration and validation phase, the simulated discharge resulted with Nash‐Sutcliffe Efficiency over 0.76 and 0.71, respectively. Moreover, the output for the spatial distribution of snow cover evaluated against Moderate Resolution Imaging Spectroradiometer‐derived 8‐day maximum snow cover extents by employing pixel‐by‐pixel analysis with average model accuracy over 88.28% and 85.89%. To the authors' knowledge, it is the first study to integrate VPT in hydrologic modeling with robust potential for optimal water resource management in the data scarce region.