Large-area land surface simulations in heterogeneous terrain driven by global data sets: application to mountain permafrost

Numerical simulations of land surface processes are important in order to perform landscape-scale assessments of earth systems. This task is problematic in complex terrain due to (i) high-resolution grids required to capture strong lateral variability, and (ii) lack of meteorological forcing data where they are required. In this study we test a topography and climate processor, which is designed for use with large-area land surface simulation, in complex and remote terrain. The scheme is driven entirely by globally available data sets. We simulate air temperature, ground surface temperature and snow depth and test the model with a large network of measurements in the Swiss Alps. We obtain root-mean-squared error (RMSE) values of 0.64 degrees C for air temperature, 0.67-1.34 degrees C for non-bedrock ground surface temperature, and 44.5 mm for snow depth, which is likely affected by poor input precipitation field. Due to this we trial a simple winter precipitation correction method based on melt dates of the snowpack. We present a test application of the scheme in the context of simulating mountain permafrost. The scheme produces a permafrost estimate of 2000 km(2), which compares well to published estimates. We suggest that this scheme represents a useful step in application of numerical models over large areas in heterogeneous terrain.