Validations of the 10-km operational Regional Spectral Model (RSM) and the coupled Mesoscale Spectral Model (MSM) with an advanced land surface model (LSM) forecasts during a 1-month period from 20 May through 20 June 2002 are performed at three surface sites on the island of Oahu. One heavy rainfall case over the Hawaiian Islands is also simulated using the MSM–LSM. Over land with adequate representation of the terrain, the 1.5-km MSM provides better forecasts of surface variables than the 10-km operational RSM. However, there are still appreciable discrepancies between the MSM simulations and observations. Further improvements are achieved by coupling the MSM with the LSM. In particular, overestimation of the surface wind speed and daytime cold biases experienced by the MSM are largely corrected in the coupled MSM–LSM.;
Composite analyses of surface variables at three surface sites under different trade wind conditions show that the observed diurnal cycles in 2-m temperature, 2-m dewpoint temperature, and 10-m wind are better forecasted by the MSM–LSM than by the MSM. The observed daytime minima in 2-m dewpoint temperatures during the strong trade wind days at two urban sites are reproduced by the MSM–LSM. The heavy rainfall case studies presented herein indicate that the high-resolution MSM–LSM has better capability in simulating localized rainfall distributions and airflows associated with the heavy rainfall event than the 10-km RSM–LSM. A major model bias is that the MSM–LSM produces excessive rainfall on the windward side of the island of Oahu with no rainfall downstream of the mountain ridges, in contrast to the observed rainfall distribution that shows the maximum rainfall axis occurring slightly downstream of the mountain ridges.