Combining Deterministic and Probabilistic Methods to Produce Gridded Climatologies

There are nearly 8,000 sites in the United States for which detailed climatologies of various atmospheric variables have been produced. However, spatial interpolation of these point climatologies via traditional techniques (IDW, kriging, etc.) is often not accurate, especially in mountainous regions, coastal zones, or data sparse areas. This paper discusses NCDC's preliminary efforts to combine deterministic and probabilistic methods to produce improved gridded climatologies. Two methods to combine multiple linear regression information with deterministic climatologies produced by the Advanced Climate Modeling and Environmental Simulation (ACMES) modeling system are described. GIS is used to visualize intermediate and final grids during the development of these techniques.