Emergency dispersion forecasts for east Tennessee: How best to utilize WRF?

As part of the atmospheric transport and dispersion program of NOAA's Air Resources Laboratory (ARL), we run WRF regularly on two levels of nest covering east Tennessee and parts of surrounding states. Our research intent is to enhance the operational utility of a mesoscale model in providing an immediate forecast of wind and mixing for emergency responders to an atmospheric spill. East Tennessee's terrain is a broad, shallow valley 300 m deep at Knoxville, 50 km wide and 450 km long. We present here an assessment of the quantitative performance of a base-line implementation of WRF in such terrain. This uses Version 2.2 with default options for all physical models and numerical schemes. It is initialized and bounded by NCEP's Rapid-Update-Cycle runs. Models are known to provide qualitative but not particularly good quantitative simulation of mesoscale features. Two such features will be examined in this preliminary exploration. First, large valleys are known to steer and channel wind, affecting the direction of atmospheric transport. Second, and perhaps more important in east Tennessee, there are long parallel ridges about 1 km wide and 50 m to 100 m high that corrugate the valley floor. Most human activity occurs between these ridges, but a 3.3-km inner grid typical of mesoscale models does not resolve them. We have been accumulating thrice-daily WRF runs for more than a year and have a parallel set of measurements of wind and temperature from 16 sites of the (east Tennessee) Regional Atmospheric Monitoring and Analysis Network. This network, which ARL maintains on and between the corrugating ridges, provides the primary comparison for this baseline test and subsequent upgrades.