PRECIPITATION MEASUREMENT BIASES IN THE UNITED-STATES

Gage-induced biases in monthly precipitation are estimated and removed at 1818 stations across the continental United States from 1950 throu gh 1987. Deleterious effects of the wind and wetting losses on the int erior walls of the gage were considered. These ''corrected'' estimates were obtained using site-specific information including wind speed, s helter-height air temperature, gage height, and sheltering. Wind speed and air temperature were interpolated at stations for which these dat a were not available using a spherically-based, nearest neighbor inter polation procedure. Results indicate that, as expected, biases are gre ater in the winter than the summer owing to the increased problems (pa rticularly wind-induced) of measuring snowfall. In summer, percent err ors range between 4 and 6 percent over nearly three-quarters of the Un ited States with slightly larger errors over the Rocky Mountains. By c ontrast, winter biases are highly correlated with snowfall totals and percentage errors increase poleward, mimicking patterns of snowfall fr equency. Since these biases are not trivial, they must be accounted fo r in order to obtain accurate and reliable time-series. If these biase s are not properly addressed, serious errors can be introduced into cl imate change, hydrologic modeling, and environmental impact research.