Evaluating an Objective Wind Analysis Technique with a Long Record of Routinely Collected Data

Abstract

An automated system has collected meteorological data hourly in the San Francisco Bay Area since October 1995. Data from sites operated by the National Weather Service, Federal Aviation Administration, U.S. Navy, U.S. Air Force, the San Francisco Bay Area Air Quality Management District, and others have been integrated into a common database and archived by the U.S. Geological Survey. Data for 0300 and 1500 UTC during 1996 were taken from that archive and used to evaluate the performance of the Winds on Critical Streamline Surfaces objective analysis code in this area of complex terrain, where altitudes range from 0 to over 1000 m above mean sea level. The comparisons of observed and modeled winds are for a longer time period and a wider variety of meteorological conditions than are available from limited-duration field studies. The authors chose 4-, 7-, and 11-station ?networks? for analysis; there were 461 h when data were available from all the sites in the 4-station network, 439 h for the 7 stations, and 343 for the 11 stations. The essential features of the model are described, and examples are given to show how changing some of the features affects the comparisons between modeled and observed winds. The effects of increasing the number of stations used for the analyses also are discussed. The model evaluation techniques described here can be applied to wind analyses from any source, but application certainly is more feasible with simple diagnostic models that can be run repeatedly so that sensitivities and performance changes can be examined. The performances of more complex models also can be compared with these results to determine if their greater complexity produces significant improvements.