Estimation of long-term climate information at locations with short-term data records

The accurate mapping of climate has widespread interest and benefit and depends on the availability of long-term data. Accuracy increases with the amount of data, but, often, insufficient data exist for the desired accuracy. To produce accurate maps within a short time frame, a scheme for estimating long-term climate statistics at locations with only short records is required. A general method is described that involves thin-plate smoothing-spline interpolation. Using data from the long-term stations, two independent interpolations are made to the short-term sites: first, the climate element's distribution and, second, the percentile for the period of observation at the short-term sites. Together these imply a value for the period of observation, but actual values are, of course, available for this period. The difference between the actual and implied values at the short-term sites provides an adjustment to the estimated distribution. It is also demonstrated that for short-term station deployment the better strategy is to have stations at sites for only 1 yr and then to move them to other sites rather than to have them at fewer sites for more than 1 yr.