INTERANNUAL VARIABILITY AND REGIONAL CLIMATE SIMULATIONS

An assessment is made of a regional climate model's skill in simulatin g the mean climatology and the interannual variability experienced in a specific region. To this end two ensembles comprising three realizat ions of month-long January and July simulations are undertaken with a limited are a operational NWP model. The modelling suite is driven at its lateral boundaries by analysed meteorological fields and the compu tational domain covers Europe and the North-western Atlantic with a ho rizontal resolution of 56 km. Validation is performed against both ope rational ECMWF analyses and objectively analysed precipitation field f rom a network of similar to 1400 SYNOP rain gauge stations. Analysis o f the simulated ensemble-mean climatology indicates that the model suc cessfully reproduces both the winter and summer distributions of the p rimary dynamical and thermodynamical held, and also provides a reasona ble representation of the measured precipitation over most of Europe. Typically the domain averaged model-biases are below 0.5 K for tempera ture and 0.1 g/kg for specific humidity. Analysis of the interannual v ariability reveals that the model captures the wintertime changes incl uding that of the precipitation distribution, but in contrast the summ ertime precipitation totals for the individual years is not simulated satisfactorily and only partially reproduces the observed regional int erannual variability. The latter shortcomings are related to the follo wing factors. Firstly the model bias in the dynamical fields is somewh at larger for summer than winter, while at the same time summertime in terannual variability is associated with weaker effects in the dynamic al fields. Secondly the summertime precipitation distribution is more substantially affected by small-scale moist convection and surface hyd rological processes. Together these two factors suggest that summertim e precipitation over continental extratropical land masses might be in trinsically less predictable than wintertime synoptic scale precipitat ion.