Determining the Minimum Number of Stochastically Generated Point Precipitation Records Required to Model Basinwide Precipitation

Abstract

This paper presents a general expression that may be used to determine the minimum number of locations in a river basin at which cross-correlated point precipitation time series must be stochastically generated to adequately capture the temporal variance of precipitation over large areas, based on the spatial correlation structure of the precipitation field. Using fewer precipitation time series than necessary leads to an overestimate of the variance of the areal precipitation. Application of the derived expression is demonstrated by using the Anseba River basin in Eritrea, East Africa, as an example. This test case is validated using a series of stochastic precipitation fields that are generated at different spatial scales. The expression was derived by expanding the method of moments equation for the variance at the field scale.