Enhancement of precipitation data for small storm hydrologic prediction

Hydrologic simulation and prediction of runoff resulting from small to moderate storm events may be hampered by the resolution at which precipitation data are available. For example, since about the early 1970s, many of the precipitation data archived by the US National Climatic Data Center have been measured and reported at a resolution of 2.5 mm. This resolution is generally adequate for large storm runoff prediction where the precipitation depths of interest may be several inches or more, but it may introduce biases into runoff predictions (or into calibrated runoff model parameters) when the precipitation depths of interest are small or moderate. This paper presents a method whereby the resolution of an archived time series of incremental precipitation depths may be enhanced for small storm hydrologic prediction. Enhancement is achieved by perturbing cumulative totals of the coarsely resolved archival data to represent them at a finer degree of resolution, followed by differencing of the enhanced cumulative totals to obtain enhanced incremental precipitation depths. It is demonstrated that the enhanced precipitation depths are a much less biased representation of true depths than are the coarsely resolved data, and that the enhanced values also have a lower mean squared error than do the coarsely resolved depths.