Abstract
A geostatistical framework for integrating lower-atmosphere state variables and terrain characteristics into the spatial interpolation of rainfall is presented. Lower-atmosphere state variables considered are specific humidity and wind, derived from an assimilated data product from the National Centers for Environmental Prediction and the National Center for Atmospheric Research (NCEP–NCAR reanalysis). These variables, along with terrain elevation and its gradient from a 1-km-resolution digital elevation model, are used for constructing additional rainfall predictors, such as the amount of moisture subject to orographic lifting; these latter predictors quantify the interaction of lower-atmosphere characteristics with local terrain. A “first-guess” field of precipitation estimates is constructed via a multiple regression model using collocated rain gauge observations and rainfall predictors. The final map of rainfall estimates is derived by adding to this initial field a field of spatially interpolated residuals, which accounts for local deviations from the regression-based first-guess field. Several forms of spatial interpolation (kriging), which differ in the degree of complexity of the first-guess field, are considered for mapping the seasonal average of daily precipitation for the period from 1 November 1981 to 31 January 1982 over a region in northern California at 1-km resolution. The different interpolation schemes are compared in terms of cross-validation statistics and the spatial characteristics of cross-validation errors. The results indicate that integration of low-atmosphere and terrain information in a geostatistical framework could lead to more accurate representations of the spatial distribution of rainfall than those found in traditional analyses based only on rain gauge data. The magnitude of this latter improvement, however, would depend on the density of the rain gauge stations, on the spatial variability of the precipitation field, and on the degree of correlation between rainfall and its predictors.
Corresponding author address: Phaedon C. Kyriakidis, Dept. of Geography, University of California, Santa Barbara, Ellison Hall 5710, Santa Barbara, CA, 93106-4060. phaedon@geog.ucsb.edu
