Modified snow algorithms in the Canadian land surface scheme: Model runs and sensitivity analysis at three boreal forest stands

Abstract

Version 3.1 of the Canadian Land Surface Scheme (CLASS) contains a number of new algorithms of significance for snow simulations in the boreal forest. In particular, mixed precipitation is now modelled, the density of fresh snow varies with temperature and the maximum snowpack density varies with snow depth. A model for canopy interception and unloading of snow developed in the Canadian boreal forest has also been implemented. In this paper, nine‐month column runs of CLASS 3.1 are compared with CLASS 2.7, the current operational version. The model runs span the winter of 2002–03 at three boreal forest sites: a mature aspen stand, a mature jack pine stand and a mature black spruce stand, all located in central Saskatchewan. The focus is on the winter performance and the representation of snow. More accurate (lower) values of modelled snow density improve the modelled snowpack depth. The accuracy of the canopy interception algorithm could not be tested directly with respect to measured interception, but results suggest that the ability to unload intercepted snow is important for accurate estimates of sublimation loss, and that simulated snow water equivalent is sensitive to perceived canopy gap fraction, interception capacity, and unloading rate. Underestimation of the canopy gap fraction increases canopy interception and sublimation losses, and decreases snow water equivalent in the snowpack, and vice versa. Employing modified gap fraction values improved the modelled snow water equivalent at two of the sites. Modifications to the model are suggested to allow the total albedo to respond to changes in the modelled sub‐canopy albedo.