84th AMS Annual Meeting

Tuesday, 13 January 2004: 2:15 PM
Geospatial Uncertainty Analysis and Gridded Forecast Verification
Room 3A
Michael E. Splitt, University of Utah, Salt Lake City, UT; and K. Cook and E. Petrescu
Poster PDF (554.3 kB)
Objective analyses of meteorological surface fields by a variation of the ARPS Data Analysis System (ADAS) are used to verify gridded forecasts that are part of the Western Region National Weather Service Interactive Forecast Preparation System (IFPS). Objective analyses are often acknowledged as having uncertainty, but the error is often considered spatially homogeneous. Kriging, an optimum interpolation objective analysis scheme used in geospatial statistics, can provide a spatially varying uncertainty estimate of a gridded analysis. Kriging uncertainty estimates based on variances between the Rapid Update Cycle (RUC) and MesoWest surface observations are used as a proxy for uncertainty estimates for ADAS, which is also an optimum interpolation scheme. The uncertainty estimates are used to modify forecast skill scores and an assessment is made of accounting for spatially varying uncertainty in the verification process.

Supplementary URL: