Key points are not available for this paper at this time.
Abstract National or regional grid‐based hydrological models are usually run at relatively fine spatial resolutions. But the meteorological data necessary to drive such models are often coarser resolution, so some form of spatial downscaling is generally required. A 1 km hydrological model for Great Britain is used to test the performance of a simple method of downscaling precipitation based on 1 km patterns of long‐term mean annual rainfall (Standard Average Annual Rainfall; SAAR). For a range of coarser resolutions (5, 10, 25 and 50 km), a 1 km grid of multiplicative scaling factors is derived as the ratio of the 1 km grid box SAAR divided by the mean SAAR of the coarser resolution grid box that contains it. A dataset of 1 km daily observation‐based precipitation is then degraded to the coarser resolutions, and application of SAAR scaling factors is compared to no downscaling and direct use of 1 km data, for simulating river flows for a large set of catchments. SAAR‐based downscaling provides a clear improvement over no downscaling. Using monthly rather than annual long‐term mean rainfall patterns provides minimal further improvement. There are no strong relationships between performance and catchment properties, but performance using 50 km precipitation without downscaling tends to be worse for smaller, steeper catchments and those with a more south‐westerly aspect; these benefit more from SAAR‐based downscaling. An assessment using high‐resolution convection‐permitting model data shows relatively small changes in derived SAAR scaling factors between a baseline and far‐future period, suggesting that use of historical scaling factors for future periods is reasonable. Applicability of this simple downscaling method for other parts of the world should be similarly assessed, for both historical and future periods. While use of annual patterns seems to be sufficient in Britain, areas where spatial rainfall patterns are more variable through the year may require use of sub‐annual patterns.
Kay et al. (Wed,) studied this question.