PNNL

Abstract

This paper compares the wind field from a diagnostic model (CALMET) over complex terrain in the Phoenix region in the USA with observations that are gridded by a state-of-the-art Four-Dimensional Data Assimilation (FDDA) system. The wind difference between the CALMET and FDDA wind fields is larger at night than in the day. The magnitude of the wind difference can be smaller than 5% of the mean wind speed at low levels in areas with dense observational stations, while it can be larger than 80% in areas without observational stations or at high altitudes. The vector-mean wind direction difference over the domain is 15 deg on the surface level and 25 deg between 10 and 1500 m. To evaluate the effects of the wind difference on dispersion calculations, dispersion of a hypothetical passive tracer released from surface point sources is simulated by the second-order closure integrated puff (SCIPUFF) model driven by the CALMET and FDDA wind fields, respectively. Differences in the two simulated tracer concentration fields increase with time due to accumulation of effects of the wind differences both near the surface and at higher altitudes. Even for the release in the area with the densest distribution of surface stations, the relative difference in the peak surface concentration from CALMET-SCIPUFF and from FDDA-SCIPUFF is less than 10% only within 0.5 hr after the release in the afternoon, and increases to 70% at 1.5 hr; this is because of large differences in wind above the surface. For the release in the area with few stations, the difference can be larger than 100% or even larger after 1.5 hr from the release. To improve dispersion simulations driven by the CALMET wind in the region, observations at upper-air stations are needed and the current surface observation network needs to be reorganized or more stations are needed to account for the influence of terrain.