Assessing the effect of integrating elevation data into the estimation of monthly precipitation in Great Britain

Abstract

This paper is concerned with mapping monthly precipitation in Great Britain from sparse point data using a range of interpolation methods. The paper applies techniques that make use of the relation between precipitation and secondary variables, such as elevation, and the concern is to assess if they provide more accurate estimates than methods that do not make use of such secondary variables. The techniques applied were: (i) moving window regression (MWR), (ii) inverse distance weighting (IDW); (iii) ordinary kriging (OK), (iv) simple kriging with a locally varying mean (SKlm) and (v) kriging with an external drift (KED). MWR, SKlm and KED make use of secondary information and in this paper elevation data were acquired to inform estimation of precipitation using these techniques. The relationship between precipitation and elevation locally, and its effect on the accuracy of estimates, was examined for each month of 1999. The performance of each interpolation method was assessed through examination of mapped estimates of precipitation and cross-validation. It was concluded that KED provides the most accurate estimates of precipitation for all months from March to December judging by the cross-validation estimation error summary statistics whereas for January and February OK provided the most accurate estimates.