A Scatter Plot for Variation in Spatial Rainfall Data
by
M C Breslin
Mathematics Department, University of Queensland
Coauthors: J A Belward

Spatial interpolation is a problem ubiquitous in spatial modelling and database development. It involoves `filling in gaps' in spatial data: providing estimates of a quantity at co-ordinates with no corresponding data value using the values at other positions. Interpolation methods abound, but the accuracy of the estimates depends not only on the method used to interpolate, but also on the nature of the given data. For instance, accuracy of the interpolant is limited when the data is not spatially dense enought to capture variations inherent in the quantity. In order to investigate the adequacy of a given spatial data set for interpolating, a simple method is presented. A scatter plot is made of the value-differences of pairs from the data set and their separation distances. These plots are used to consider the smoothness of the data, especially in relation to the scale of data sampling and the effect of time averaging. The application considered here is rainfall data, but the method applies to any spatial data.

Date received: July 8, 1999