GIS and Remote Sensing in Hydrology, Water Resources and Environment (Proceedings of ICGRHWE held at the Three Gorges Dam, China, September 2003). IAHS Publ. 289, 2004, 378-384

1 Institute of Geography, School of Geosciences, University of Edinburgh, Edinburgh EH8 9XP, UK

2 Department of Geography, University of Leicester, Leicester LE1 7RH, UK

Abstract We explore the relative accuracy of using spline interpolations to estimate daily precipitation over England and Wales at 5-km grid resolution. We compare surfaces estimated for each day in March 2002 from networks of 820 and 110 raingauges and assess whether including collateral data such as gridded topography and rainfall radar accumulations improves relative estimation accuracy. For each surface RMSE, MAE and bias are computed against other gauges not used in the spline estimation. Using 820 gauges, a 2-D partial thin-plate spline on (x,y) with elevation as a single linear covariate estimates daily precipitation with an RMSE averaged over all stations and dates of 1.9 mm. For a network of 110 points, the equivalent RMSE is 2.3 mm. Modelling precipitation as a non-linear function of elevation or using further collateral data does not improve the estimation accuracy for either volume of input data. Including gridded rainfall radar data results in higher RMSE (>5 mm) and larger maximum and minimum estimates.

Key words collateral data; daily rainfall estimation; England; rainfall radar; spline interpolation; Wales