Soil moisture influences many hydrologic applications including agriculture, land management and flood prediction. Most remote-sensing methods that estimate soil moisture produce coarse resolution patterns, so methods are required to downscale such patterns to the resolutions required by these applications (e.g. 10- to 30-m grid cells). At such resolutions, topography is known to affect soil moisture patterns. Although methods have been proposed to downscale soil moisture based on topography, they usually require the availability of past high-resolution soil moisture patterns from the application region. The objective of this article is to determine whether a single topographic-based downscaling method can be used at multiple locations without relying on detailed local observations. The evaluated downscaling method is developed on the basis of empirical orthogonal function (EOF) analysis of space–time soil moisture data at a reference catchment. The most important EOFs are then estimated from topographic attributes, and the associated expansion coefficients are estimated on the basis of the spatial-average soil moisture. To test the portability of this EOF-based method, it is developed separately using four data sets (Tarrawarra, Tarrawarra 2, Cache la Poudre and Satellite Station), and the relationships that are derived from these data sets to estimate the EOFs and expansion coefficients are compared. In addition, each of these downscaling methods is applied not only for the catchment where it was developed but also to the other three catchments. The results suggest that the EOF downscaling method performs well for the location where it is developed, but its performance degrades when applied to other catchments. Copyright © 2011 John Wiley & Sons, Ltd.