The new spatial interpolator was developed as part of Nickl dissertation research (Nickl, 2012) and it is distinctive in that it takes into account important but often ignored topographic features (e.g. latitudinal and longitudinal components of slope, elevation and exposure to orography). These topographic features are evaluated at different spatial scales in order to identify optimal relationships with precipitation errors that arise when topographic influences are not taking into account in the interpolation. An adjustable-scale spatial ellipse is used to represent topographic features within a spatial scale. The method was evaluated for the San Joaquin Valley (in Western United States) showing relatively low errors when comparing with traditional interpolation.
Precipitation observations from the Global Historical Climatology Network (GHCN) are used to estimate monthly and seasonal climatologies at each station over contiguous U.S. region and a digital elevation model (DEM) at 2.5-minutes resolution is used to estimate topographic features at different spatial scales. Relationships among topographic variables and precipitation errors reveal underestimations/overestimations related to higher/lower elevations, west/east oriented slopes and protruding/below surface areas. Land surface precipitation estimates using the new spatial interpolator exhibit differences over the high topographic variability areas when comparing with NCEI/NOAA and PRISM estimates, especially over the Western U.S. New estimates show better performance than Cressman's traditional interpolator when applying cross-validation.
Results from this research help to enhance our knowledge of the important role of topography on precipitation estimations over mountainous regions in contiguous U.S. and open the door to the potential applications of this new approach to other regions of the world.