Numerical Simulations of the Island-Induced Circulations over the Island of Hawaii during HaRP

Abstract

Abstract The fifth-generation Pennsylvania State University–NCAR Mesoscale Model (MM5)/land surface model (LSM) is used to simulate the diurnal island-scale circulations over the island of Hawaii during the Hawaiian Rainband Project (HaRP, 11 July–24 August 1990). The model is initialized with the NCEP–NCAR reanalysis data. The diurnal variations of the land–sea thermal contrast at the land surface and the planetary boundary layer at Hilo, Hawaii, are well simulated. The main discrepancy occurs on the leeside areas of mountains or ridges below the trade wind inversion (2 km), where the simulated afternoon land–sea thermal contrast at the surface is 1°–3°C lower than observed mainly due to the misrepresentation of lava rocks by the bare ground category in the U.S. Geological Survey (USGS) data and stronger than observed simulated sea breezes bringing in relatively cool maritime air. The flow deceleration and splitting of the incoming trade wind flow and the evolution of the diurnal circulation cells on the windward side, the thermally driven diurnal winds, and the wake circulations on the lee side are well simulated. The simulated diurnal variations in rainfall are also in good agreement with observations. However, the simulated winds in areas well exposed to the trade wind flow are weaker (1–3 m s−1) than observed mainly due to the underestimation of trade wind flow in the NCEP–NCAR reanalysis. The simulated rainfall over windward lowlands at night is underestimated and the maximum rainfall axis shifts farther toward the coast as compared with observations, due to an underestimation of orographic lifting aloft and a relatively large horizontal extent of the simulated katabatic flow because of the weaker- than-observed trade wind flow in the initial conditions. In the afternoon hours on the windward side, the strongest winds (anabatic/sea breeze and trade wind flow) are simulated in low levels over land in response to the surface heating, with a westerly wind deviation beneath the mean trade wind inversion (2 km) and sinking motion over the adjacent oceans. The simulated low-level flow deceleration of the incoming trade wind flow is most significant in the early morning as a combination of island blocking and nocturnal cooling over land. At that time, the simulated upward motion representing the rising branch of the thermally direct circulation extends more than 40 km offshore. Sensitivity tests show that with better surface conditions in the model coupled with the Oregon State University (OSU) LSM, the simulated thermal forcing over land is improved. The improvements in simulated ground temperature, land–sea thermal contrast at the land surface, and mixing ratio lead to better simulation of the strength of land/sea breezes over the island.