Impact of the Canadian land surface scheme on monthly ensemble predictionsof water and energy budgets over the Mackenzie river basin

As part of the Mackenzie GEWEX Study (MAGS), Canadian global spectral forec ast model (SEF) monthly simulations of surface water and energy fluxes aver aged over the Mackenzie river basin are examined. We study the impact of th e more sophisticated Canadian Land-Surface Scheme (CLASS) on the predictabi lity of the fluxes, using the operational configuration of the global model with the force-restore land-surface scheme as a baseline for comparison. T he model, connected successively to the current operational force-restore l and surface scheme and to CLASS, generates nine-member ensemble simulations of one month duration from analyses perturbed with the Monte-Carlo method, for spring, summer, fall and winter cases. The perturbations are comparabl e in magnitude to observational errors. The flux predictability is measured by the degree to which the individual forecasts in the ensembles vary from one another. The mean of the ensemble forecasts is compared with monthly a ccumulations of surface energy and water fields constructed from 12-hour fo recasts of the operational Regional Finite Element (RFE) model as well as w ith satellite-derived observations of precipitation. We found that, overall , the replacement of the force-restore scheme with CLASS in the Canadian Me teorological Centre (CMC) global forecasting system has a beneficial effect on the predicted MAGS energy and water budgets. While their sensitivity to initial conditions increases only slightly, the budgets, forecast with CLA SS, compare better with their counterparts constructed from the short-term predictions of the CMC regional forecasting system than the output of the m odel coupled with the operational force-restore scheme.