Montane snowpacks are extremely complex to represent and usually require assimilating remote sensing images at very fine spatial resolutions, which is computationally expensive. Adapting the grid size of the terrain to its complexity was shown to cut runtime and storage needs by half while preserving the accuracy of ~ 100 m snow estimates. This novel approach will facilitate the large-scale implementation of high-resolution remote sensing data assimilation over snow-dominated montane ranges.