Patching and Disaccumulation of Rainfall Data for

Hydrological models are increasingly relied upon for decision-making in catchment planning and management. But the impacts of errors in input data are seldom fully considered in the analysis of these and many other environmental models, even though they can be a potentially large source of model uncertainty. Precipitation data are an essential input for many hydrological modelling investigations but there are potentially significant errors in catchment areal rainfall estimates. Estimating rainfall at catchment scales is a difficult process, due to variations of rainfall both in time and space. Identifying, and preferably reducing, the errors present in catchment rainfall estimates is a necessary step towards determining and improving the accuracy of hydrological models. Sources of errors within areal rainfall estimates can include: extrapolating point measurements to spatial estimates, missing data, accumulation errors (rainfall is allowed to accumulate in the gauge for several days before it is measured), along with other measurement and timing errors. The focus of this paper is on patching and disaccumulation errors. Patching errors occur when missing rainfall within a record is estimated. Disaccumulation errors occur when accumulated rainfall is allocated to the preceding days within the period. In this paper we describe different rainfall patching and disaccumulation methods for dealing with accumulated and missing rainfall. We focus on conceptually simple and computationally inexpensive methods. The patching methods considered in the paper include: • Nearest Neighbour by Distance (ND): selecting the closest gauge with data