An appraisal of SPOT imagery to investigate landuse in the River Main drainage basin, N. Ireland

Abstract

The study described here was designed to test the use of SPOT imagery for the mapping and assessment of landuse. Unlike England & Wales, no landuse maps are available in N. Ireland and as such all testing is achieved by Field ground truth.

Numerous techniques, ranging from the use of single band (3) through ratios and vegetation indices to maximum-likelihood, were used to enable a successful classification. The major problem was the similarity of bare soil (arable landuse) and heather. The solution to this problem involved the use of a 3/2 ratio to give landuse classes augmented by an “edited” box classification which gives an estimate of the bare soil/arable component. This problem may not be significant in future studies, provided the fly-over occurs during the period of growth for arable crops.

Testing of the classification developed from SPOT data was undertaken using field ground truth and agricultural statistics for the year 1984. Landuse derived from the classification was shown to agree with ground truth on 97% of all cases, while the areal estimates of landuse types produced an r2 value of 0.9607 when plotted against estimates derived from official 1984 Agricultural Statistics.

Production of this classification would have been virtually impossible with other available imagery for several reasons, the prime reason being the relatively small field size in Northern Ireland. Standard Landsat-3 imagery was found to be of little use in comparison with that from SPOT imagery.