Validation of the Canadian land surface scheme (class) for a subarctic open woodland

Abstract

This study reports on testing of the peatland version of the Canadian Land Surface Scheme (CLASS) for simulating the energy balance of subarctic open woodland terrain. Model results are compared against several years of measured data from a site near Churchill, Manitoba. In contrast to most forest environments, the floor of the open forest plays a large role in total ecosystem energy exchange. This behaviour presents a significant challenge for land surface models like CLASS and their simplified treatment of vegetation canopies. Simulations of summer energy balance for seven years encompassing a wide range of meteorological conditions produced consistent results. Root mean square errors for sensible and latent heat fluxes fell between 11 and 28 W m−2. CLASS consistently underestimated slightly the daily latent heat flux and overestimated the sensible heat flux, average mean bias errors being ‐7.6 and 9.1 W m−2, respectively. The soil heat flux was less well represented. In general, CLASS was able to capture the diurnal and seasonal behaviour of the measured fluxes under a range of conditions with reasonable accuracy. In a full year simulation, CLASS reproduced the annual variations in energy balance with some discrepancies associated with snow accumulation and ablation periods. The model performance was sensitive to both snow density and specification of the surface cover. Recommendations for improving the model for subarctic woodlands and terrain types with similar features are discussed.