Evaluation of a regional climate model for paleoclimate applications in the Arctic

Citation
Ba. Felzer et Sl. Thompson, Evaluation of a regional climate model for paleoclimate applications in the Arctic, J GEO RES-A, 106(D21), 2001, pp. 27407-27424
Citations number
84
Categorie Soggetti
Earth Sciences
Volume
106
Issue
D21
Year of publication
2001
Pages
27407 - 27424
Database
ISI
SICI code
Abstract
The Paleoclimates From Arctic Lakes and Estuaries (PALE) project has been i nvestigating methods of doing high-resolution model-data comparisons for th e Arctic. As a prelude to a paleosimulation of the North Atlantic region, a modem simulation using observationally driven reanalysis data has been com pleted. The ARCSyM mesoscale model has been configured for the North Atlant ic region, including Labrador, Ungava, Baffin Island, Ellesmere Island, Gre enland, and Iceland, with a resolution of 70 kin. This high resolution is n ecessary to predict sub-GCM grid-scale climate processes, such as precipita tion and storm patterns that depend upon the detailed topography and coastl ines of the region. Experiments were performed for the time period of Septe mber 1987 to March 1990, driven by observational analyses. The model accura tely captures the major summer and winter circulation systems in the North Atlantic region. Comparisons with meteorological station data show high cor relations for winter and summer surface temperatures, with a cold bias in w inter and a warm bias in summer. Winter precipitation is well simulated by the model because it is driven by the large-scale circulation. The orograph ically driven summer precipitation is overrepresented and does not correlat e well with observations, although the overall pattern is correct. These re sults show that the model is capable of capturing the correct temperature a nd precipitation patterns, although grid-to-grid comparisons are not possib le. The mesoscale model is therefore useful for regionally based data-model comparisons, but should not be used to compare individual cores with speci fic model grids.