Estimating Observational Uncertainty in Surface Temperature over Complex Terrain
John D. Horel, University of Utah, Salt Lake City, UT; and D. T. Myrick and C. D. Whiteman
Forecasts of temperature from the National Weather Service are being issued on a 5 km x 5 km grid for the entire United States. Verification of these temperature forecasts against observations is hampered in regions of complex terrain by observational uncertainty arising from the instrumentation and representativeness errors. In addition, data assimilation systems require specification of the observational error covariance relative to the unknown truth. The goal of this study is to improve estimates of observational uncertainty of surface temperature in mountainous areas. Attention is placed on the representativeness error of temperature observations as a function of terrain characteristics (e.g., elevation, slope, roughness within a 5 km x 5km grid box, vegetative cover).
Observations from pairs of stations located within 5 km of one another throughout the western United States are used to examine observational uncertainty. Data from 1997-2006 were obtained from the MesoWest data base. Diurnal and annual variations in observational error estimates are examined as well as sensitivity to synoptic regimes and characteristics of the underlying terrain.
Surface temperature observations across the Owens Valley, CA, that were collected as part of the Terrain-Induced Rotor Experiment (TREX) during March-April 2006 are examined as well. HOBO temperature sensors were spaced at roughly 50 m increments of elevation along a line through Independence, California from the western slope to the eastern slope of the valley. Several pairs of sensors were colocated to allow estimation of measurement errors distinct from the representativeness errors arising from the sloping terrain.
Session 16, Forecasting Mountain Weather: Part II
Friday, 1 September 2006, 10:30 AM-12:00 PM, Ballroom South
Browse or search entire meeting
AMS Home Page