Abstract
This study investigates the sensitivity of components of the hydrological cycle simulated by the Canadian Regional Climate Model (CRCM) to lateral boundary forcing, the complexity of the land surface scheme (LSS), and the internal variability arising from different models’ initial conditions. This evaluation is a contribution to the estimation of the uncertainty associated to regional climate model (RCM) simulations. The analysis was carried out over the period 1961–99 for three North American watersheds, and it looked at climatological seasonal means, mean (climatological) annual cycles, and interanual variability. The three watersheds—the Mississippi, the St. Lawrence, and the Mackenzie River basins—were selected to cover a large range of climate conditions. An evaluation of simulated water budget components with available observations was also included in the analysis.
Results indicated that the response of climatological means and annual cycles of water budget components to land surface parameterizations and lateral boundary conditions varied from basin to basin. Sensitivity to lateral boundary conditions is, in general, smaller than sensitivity to LSS and tends to be stronger for the northern basins (Mackenzie and St. Lawrence). Interannual variability was unaffected by changes in LSS and in driving data. Internal variability triggered by different initial conditions and the nonlinear nature of the climate model did not significantly affect either the 39-yr climatology, the climatological annual cycles, or the interannual variability. A comparison with observations suggests that although the simple Manabe-based LSS may be adequate for simulations of climatological means, skillful simulation of annual cycles require the use of a state-of-the-art LSS.
The Cloud_cci simulator for the ESA Cloud_cci climate data record
and its application to a global and a regional climate model
