Kevin Roe (Maui High Performance Computing Center)
Keywords: Modeling, Analysis, Simulations
Abstract:
The Hawaiian Islands consist of large terrain changes over short distances, which results in a variety of microclimates in a very small region. Some islands have rainforests within a few miles of deserts; some have 10,000+ feet summits only a few miles away from the coastline. Because of this, weather models must be run at a much finer resolution to accurately forecast weather changes in these regions. NCAR’s Weather Research and Forecast Model (WRF) is run, on a nightly basis, using a coarse 54 km resolution grid (encompassing an area of approximately 7000 by 7000 km) nested down to a 2 km grid over each Hawaiian county. Since the computational requirements are high to accomplish this in a reasonable time frame (as to still be a forecast) WRF is run in parallel on MHPCC’s Cray 2.4 GHz Opteron based Linux system, “Hoku”. Utilizing 32 nodes (64 processors) the WRF model is run over the above conditions in approximately 4 hours. Although WRF forecast have been in place for over a year now, a lot of experience has gone behind its setup. MHPCC has been running NCAR’s Mesoscale Model version 5 (MM5) since 2000, which continues to be utilized by operators at the telescope facilities on Haleakala, Maui. Currently, the forecast produced is for a 48-hour simulation, but will most likely be extended to a 72-hour simulation; this forecast is available to operators by 8 AM and produces forecasts out until the next day at 8 PM. This is enough time to give operators and managers time to reschedule their operations if unacceptable conditions are predicted. The products we currently provide are: temperature, wind speed & direction, relative humidity, and rainfall. Additional products to be produced over Haleakala, include clear air turbulence, the Richardson number, and a measure of optical turbulence for the telescope operators using the Dewan and Jackson models.
Date of Conference: September 12-15, 2007
Track: Modeling, Analysis and Simulations