A new method for estimating the spatial distribution of mean seasonal and annual rainfall applied to the Hunter Valley, New South Wales

Abstract

An empirical analysis of the spatial distribution of long-term mean seasonal and annual rainfall in the Hunter Valley for the period 1901 to 1975 is presented. Isohyet maps have been derived, using a recently developed method, from Laplacian smoothing spline surfaces fitted to rainfall data at 203 stations. The surface fitting method represents a substantial improvement over previously available subjective, least squares polynomial and weighted interpolation methods. The smoothing spline surfaces are functions of latitude, longitude and elevation with the degree of data smoothing determined by minimising the generalised cross validation (GCV). The method does not require prior standardisation of short-term means to long-term means, affords error estimates and allows testing of the validity of the fitted model. The fitted rainfall surfaces estimate long-term mean annual rainfall at any point in the valley with a percentage root mean square predictive error of 9.2 per cent and estimate long-term mean seasonal amounts at any point with percentage root mean square predictive errors ranging from 11.9 per cent in winter to 18.2 per cent in summer.