ABSTRACT

An important problem in land resource analysis is concerned with finding ways of linking the available data sources, which take regard of their particular accuracy and resolution, and which present results for relevant spatial frameworks (Walker and Mallawaarachchi, 1998). This problem is particularly highlighted when the input data are drawn from census surveys, for which data are generally provided for irregular and arbitrary tracts, since often these do not match the geography of environmental resources. GIS-literate quantitative geographers in UK universities have paid considerable attention to studying this problem, and indeed their knowledge is now proving valuable in planning outputs for the 2001 population census (see, for example, Martin, 1997; Martin, 1998). It is thus somewhat ironic that the criticisms which have been levelled at the annual agricultural census by their predecessors and peers for more than a quarter century (Coppock, 1960, 1965; Clark, 1982; Robinson, 1988) remain valid today. Whereas the ‘GIS revolution has changed the entire context in which users see and use [population] census data’ (Openshaw, 1995, p. 133), the agricultural census (which records crop/non-crop areas and livestock and labour numbers) has been ‘made GIS-able’ very much as an incidental to the ongoing production of annual ‘text-and-table’ summary reports (see, for example, MAFF, 1997; SOAEFD, 1998). This is achieved by the attachment of farmers’ returns summed as parish-level totals to a parish boundaries coverage.