In some situations in urban modeling practice, data cannot be preserved at the highest possible level of resolution. For example, when data from different sources are collated, the areal partitioning systems may not be compatible with each other. In other cases, related but separated models (e.g., urban transportation–land use and environmental models) may have been designed to operate at different spatial scales, posing a challenge to any efforts to link them. In these and similar situations, a method for interpolating data is required to produce compatible zoning systems or data at a desired level of resolution. A nonstationary (location-specific) spatial interpolation method, which has various potential applications in transportation as well as urban modeling, is proposed. The method combines the concept of location-specific parameters of a geographically weighted regression model and the concept of variogram function of kriging used to model spatial autocorrelation. Two case studies are presented to illustrate the application of the method in situations that are common in urban and transportation analysis. The results suggest that the method can be a useful alternative for spatial interpolation when nonstationarity and spatial autocorrelation appear co-incidentally in the analysis. The model is therefore expected to help to improve the performance of urban models by providing more accurate data at desired levels of resolution.