Trend Extraction Using Average Interpolating Subdivision

Spatial averages are robust components of data sets, and therefore well suited as starting points for trend analysis. A natural way of constructing a global trend surface from average values, is to use a subdivision algorithm. This procedure will in some cases reproduce exactly the drift of the data set. Numerical experiments on multibeam echo sounder data show that residual data sets contain no long range correlations. In addition, a variogram analysis clearly identifies the amount of noise in the original data.