The diurnal variations of surface winds, temperature, and rainfall under summer trade-wind conditions during the Hawaiian Rainband Project (HaRP, July 11 – August 24, 1990) were simulated by using the MM5 model coupled with the Oregon State University land surface model (OSU-LSM), and validated with the 50 HaRP Portable Automated Mesonet (PAM) surface station data. The LSM has four layers underground: 10, 40, 100, and 200 cm. The land use and vegetation cover from the U.S. Geological Survey (USGS) 30' resolution data were used. A two-way nesting procedure with four nested domains is used with horizontal resolutions of 81 km, 27 km, 9 km and 3 km, respectively. The MM5/LSM was run for two months prior to the HaRP period with the initial soil moisture specified according to the soil type to generate the required soil moisture and soil temperature fields for HaRP simulations. Starting from July 10, each simulation during HaRP for each day was run for 48 hours initialized at 1200 UTC using the NCEP/NCAR reanalysis data and the 24-h forecasts of the soil moisture and soil temperatures of the previous day. The output from the 12th hour and the 36th hour was used to represent the simulated diurnal cycle of the following day.

Overall, the island blocking, orographic lifting, and the diurnal cycles of the surface winds, temperature and rainfall over the Island of Hawaii are well simulated. The simulated splitting airflow, strong winds over the northern and southern tips, Waimea Saddle, and Humu’ulu Saddle of the island of Hawaii are in good agreement with observations. At most areas on the windward side and the Kona coast on the leeside, the upslope (downslope) flow duration and the morning (evening) transition time from downslope (upslope) flow to upslope (downslope) agree well with observations. The diurnal rainfall maxima were well simulated with a maximum in the afternoon hours on the windward slopes and the leeside slopes, and a nighttime maximum over the windward coastal areas.

Some discrepancies exist between the model simulations and the observations. The simulated daytime upslope flow is 1 – 2 m s^-1 smaller than observations. The simulated rainfall accumulation during HaRP is 10 - 50% smaller than the observations with larger relative errors on the leeside areas and the windward slopes. The simulated surface temperature during HaRP is ~ 0.5º - 2.5º C lower than observations on most areas of the island except on some areas of the windward lower slopes, where the simulated surface temperature is ~ 0.5º higher. Factors attributed to the simulation errors will be discussed.