Wind Direction Bias in Generating Wind Roses and Conducting Sector-Based Air Dispersion Modeling

Abstract

Certain widely used wind rose programs and air dispersion models use an overly simple data-transfer algorithm that induces a directional bias in their output products. The purpose of this paper is to provide a revised algorithm that corrects the directional bias that occurs from the aliasing that occurs when the sector widths used to report wind direction data are on the same order of magnitude, but not equal, to the sector widths used in the wind direction summaries. The directional bias issue arises when output products in 16 direction sectors (22.5° each) are produced from wind direction data reported in terms of 36 sectors (10° each). The result directional bias affects the results of simulations of air and surface concentrations using widely applied air dispersion models. Datasets or models with the directional bias discussed here give consistent positive biases (～30%) for cardinal direction sectors (north, south, east, and west) and consistent negative biases for all of the other sectors (around ?10%). Data summary and air dispersion programs providing outputs in direction sectors that do not match the observational sectors need to be checked for this bias. A revised data-transfer algorithm is provided that corrects the directional bias that can occur in transferring wind direction data between different sector widths.