Uncertainty of Precipitation Estimates Caused by Sparse Gauging Networks in a Small, Mountainous Watershed

Watershed modeling requires reliable climate input data that provide a reasonable representation of spatial variability. In many cases, limited access, complex terrain, and remoteness make it difficult to acquire good data. This work characterizes the variability of precipitation in a small, mountainous watershed and quantifies the uncertainty of precipitation estimates caused by sparse precipitation gauging stations. Spatial precipitation variability was found to be of particular concern during the summer months. When one gauge within the watershed is recording precipitation, integration times of more than 8 days are necessary for all gauges to record. In the study catchment, the absolute error in daily mean catchment precipitation exponentially decreased with the increased number of precipitation gauges compared with the best available estimate. The use of 4 or more gauges implicitly allowed a close approximation of the best available daily mean catchment precipitation estimates. Fuzzy multiple linear regression was applied to estimate the average basin precipitation. With this method, the differences in average basin precipitation between the dense experimental gauging network and typically available sparse gauging setups were quantified. Depending on the integration time, the median relative error for the most suitable sparse setup was between 0.08 and 0.40, equivalent to a median absolute error between 0.6 and 1.8 mm d-1.