Resolution Errors Associated with Gridded Precipitation Fields

Spatial-resolution errors are inherent in gridded precipitation (P ) fields – such as those produced by climate models and from satellite observations – and they can be sizeable when P is averaged spatially onto a coarse grid. They can also vary dramatically over space and time. In this paper, we illustrate the importance of evaluating resolution errors associated with gridded P fields by investigating the relationships between grid resolution and resolution error for monthly P within the Amazon Basin. Spatial-resolution errors within gridded-monthly and average-monthly P fields over the Amazon Basin are evaluated for grid resolutions ranging from 0.1° to 5.0°. A resolution error occurs when P is estimated for a location of interest within a grid-cell from the unbiased, grid-cell average P . Graphs of January, July and annual resolution errors versus resolution show that, at the higher resolutions (<3°), aggregation quickly increases resolution error. Resolution error then begins to level off as the grid becomes coarser. Within the Amazon Basin, the largest resolution errors occur during January (summer), but the largest percentage errors appear in July (winter). In January of 1980, e.g., resolution errors of 29, 52 and 65 mm – or 11, 19 and 24% of the grid-cell means – were estimated at resolutions of 1.0°, 3.0° and 5.0°. In July of 1980, however, the percentage errors at these three resolutions were considerably larger, that is, 15%, 27% and 33% of the grid-cell means. Copyright  2005 Royal Meteorological Society.