Performance of the Canadian Land Surface Scheme at temperate Aspen-birch and mixed forests, and a boreal young jack pine forest: Tests involving canopy conductance parametrizations

The Canadian Land Surface Scheme (CLASS) was evaluated using off-line tests , with data collected at temperate aspen-birch and mixed forests located ne ar Chalk Rivet; Ontario, and at a young jack pine forest located near Thomp son, Manitoba. Measurements of stomatal conductance and leaf area index wer e used to develop canopy conductance parametrizations for the aspen-birch a nd the jack pine stands, and these were incorporated into CLASS in a modifi ed run. The behaviour of the residuals from the unmodified run with respect to environmental variables at the mixed stand were used as a guide to prod ucing a modified ad-hoc canopy conductance parametrization for use at that site. It was found that CLASS exaggerates the diurnal range in soil heat flux. Th is leads to an underestimation of the diurnal range in available energy, an d it is argued that this has implications for its partitioning into the tur bulent fluxes of sensible and latent heat. Over time, systematic errors wil l accumulate which, in the case of the latent heat flux, can affect the lon g-term water balance. Errors in the size of the turbulent fluxes over short time periods can limit the usefulness of CLASS for resolving mesoscale cir culations in a regional model. These problems are believed to be related to CLASS's assumption of zero wind speed under the canopy, and the lack of an insulating litter layer at the soil surface. An unmodified CLASS produced estimates of sensible and latent heat fluxes that broadly agreed with measu red values at the temperate stands but not at the jack pine stand, where th e latent heat flux was significantly overestimated The modified canopy cond uctance parametrizations improved the representation of the Intent heat flu x at all of the sites, particularly at the jack pine stand. Overestimation at the aspen-birch stand may be related to site heterogeneity nor being ade quately represented in the measurement of stomatal conductance, or to diffi culties in defining the vertical distribution of leaf area index for scalin g stomatal conductance to the canopy level. As expected, a positive or nega tive bias in the modelled latent heat flux is associated with an opposite b ias in the modelled sensible heat flux: The tendency towards positive inter cepts and slopes less than unity in linear regressions relating modelled an d measured fluxes is a symptom of the underestimation of the diurnal range in available energy. Recommendations are made to allow for vertical discontinuities in texture w ithin a soil layer and a variable permeable depth, and to allow transpirati on when the soil water suction exceeds -1.5 MPa. This is particularly impor tant when soil moisture or texture is heterogeneous. Finally, ii is suggest ed that volumetric soil moistures should be allowed tofall below the origin ally prescribed limiting value of 0.04. This is necessary for evaporation t o continue in very sandy, dry soils.