THE VALIDITY OF USING COUNTRYSIDE SURVEY SAMPLE DATA FROM GREAT-BRITAIN TO ESTIMATE LAND-COVER IN SCOTLAND

It is possible to make environmental or ecological descriptions of geo graphical regions using data not only collected within the region conc erned but also from a wider area. Producing accurate descriptions of a region using broader datasets is financially appealing as it should r educe the intensity and hence the costs of regional survey However; it is essential that any information collected outside the region is onl y ever used to describe similar situations and conditions within the r egion. An appropriate stratification may offer sufficient structure wi thin a dataset to identify which parts to include and which to exclude , but, if not used carefully, it can produce inaccurate results. This problem is explored using the Institute of Terrestrial Ecology (ITE) L and Classification, a sampling stratification used to structure ecolog ical and environmental surveys in Great Britain (GB). Using this strat ification, national and regional estimates of land cover and estimates of the extent of rural features in England, Scotland and Wales have b een produced by ITE from sample surveys undertaken in 1978, 1984 and 1 990. The published estimates for Scotland, for example, make use of th e complete GB sample dataset for land classes occurring in that region . The work described here shows that the Land Classification offers a legitimate stratification scheme for sampling Scotland, setting it in the context of GB. Overall, it suggests that, in most cases, it is pre ferable to make national estimates for Scotland from the full GB sampl e. The consequences of restricting the sample set to those samples dra wn only from Scotland is to reduce the sample size and subsequently st atistical confidence in the result. The implications of these findings for other studies are examined. (C) 1998 Academic Press Limited.