Calibration and Validation of a Regional Climate Model for Pan-Arctic Hydrologic Simulation

Abstract

A number of polar datasets have recently been released involving in situ measurements, satellite retrievals, and reanalysis output that provide new opportunities to evaluate regional climate in the Arctic. These data have been used to assess a 1-yr pan-Arctic simulation (October 1985?September 1986) performed by a version of the fifth-generation Pennsylvania State University?National Center for Atmospheric Research (PSU?NCAR) Mesoscale Model (MM5) that incorporated the NCAR land surface model (LSM) and a simple thermodynamic sea ice model to investigate interactions between the land surface and atmosphere. The model's standard cloud scheme using relative humidity was replaced by one using simulated cloud liquid water and ice water after a set of short test simulations revealed excessive cloud cover. Model validation concentrates on factors relevant to the water cycle: atmospheric circulation, temperature, surface radiation fluxes, precipitation, and runoff. The model captures general patterns of atmospheric circulation over land. The rms differences from the Historical Arctic Rawinsonde Archive (HARA) rawinsonde winds at 850 hPa are smaller for the simulation (9.8 m s?1) than for the NCEP?NCAR reanalysis (10.5 m s?1) that supplies the model's boundary conditions. For continental watersheds, the model simulates well annual average surface air temperature (bias <2°C) and precipitation (bias <0.5 mm day?1). However, the model has a summer dry bias with monthly precipitation error occasionally exceeding 1 mm day?1. The model simulates the approximate magnitude of spring runoff surge, but annual runoff is less than observed (18%?48% less among the continental watersheds). Analysis of precipitation and surface air temperature errors indicates that further improvements in both evapotranspiration and precipitation are needed to simulate well the full annual water cycle.