Reducing the High Mountain Asia cold bias in GCMs by adaptingsnow cover parameterization to complex topography areas

Abstract. The influence of topography on the snow cover fraction (SCF) is investigated in this study with 5 different parameterizations. These SCF parameterizations are evaluated using the High Mountain Asia Snow Reanalysis (HMASR). Then, they are implemented in the ORCHIDEE land surface model (LSM) of the Institut Pierre Simon Laplace (IPSL) general circulation model (GCM) to quantify their skill in global land-atmosphere coupled simulations. SCF varies as a function to snow depth (SD), with a relationship that differs between flat and mountainous areas in HMASR. SCF parameterizations that do not include a dependency on the topography lead to large snow cover overestimations. Furthermore, a hysteresis between SCF and SD is found in HMASR, with a rapid snow cover increase during accumulation and a slower retreat of patchy snow occurring during ablation periods, discarding parameterizations not considering this effect. The application of the parametrizations in global simulations shows contrasting results depending on the location because other processes also explain the snow biases. Nevertheless, the snow cover overestimation in mountain areas is reduced by about 5 to 10 % on average when we include a dependency on the subgrid topography in our SCF parameterizations, which in turn allows to decrease the surface cold bias from −1.8 °C to about −1 °C in the High Mountain Asia (HMA) region. However, persisting snow cover biases remain in these experiments, with a SCF overestimation in HMA, as well as a SCF underestimation in several other regions (e.g., the Rockies mountains). Further calibration considering other regions and multiple datasets would allow to improve the SCF parameterizations. Assessing SCF parameterizations is challenging as it requires both realistic snowfall and snowpack in model experiments, and combined SCF, SD, and SWE – or snow density – observations, that are generally limited and uncertain in mountainous regions.