Statistical methods in physical geography

Abstract

This essay presents a state of the art review of statistical applications in physical geography. Initially, methods drawn from traditional statistics based upon classical probability theory are described and evaluated. It is argued that explicitly, or, more usually, implicitly such methods assume independence of the observations in both space and time making it incumbent upon statistical physical geography to reject what one authority has called its former 'timebound preoccupation' (Chorley, 1962, 6) in favour of time-less 'equilibrium' morphological systems analysis. Next, I go on to examine the development and potential of models based upon the theory of stochastic processes as it relates to the evolution of systems in space and time in which observations are not assumed to be independent of each other. It is suggested that many problems are better cast in this, or a mixed stochastic/deterministic framework which will allow the time dimension to be reincorporated into physical geography..In a review of this kind, traditional subdivisions of physical geography do not form an adequate organizational framework and are becoming increasingly irrelevant to what physical geographers actually do. Instead, statistical concepts and techniques will be discussed according to the method of systems analysis to which they are most suited. According to Chorley and Kennedy ( t g7 i ), four levels of systems analysis are possible. Morphological systems deal with networks of structural relationships between parts of systems; these have been analysed by classical statistical methods. Cascading systems are defined by the paths followed by throughputs of energy and mass, and like process-response systems in which these cascade flows are linked to morphological transformations, can be analysed using stochastic process theory. Finally, physical geographers are also interested in control systems resulting from the interaction of human decision-making and physical process-response for which statistical decision theory provides a useful analytic framework but which will not be considered in the present work. Although this classification was designed for geomorphological work, a similar scheme for climatological studies has been proposed and exemplified by Terjung (1976). I The analysis of morphological systems At this level of analysis, statistical methods are used to investigate the

Percy Crowe: a forgotten pioneer quantitative geographer and climatologist

“…Stan Gregory not only adopted and adapted his methods but also, and very significantly, stimulated the teaching of rigorous statistical description through a pioneering textbook and promoted it throughout British universities and high schools (Johnston, 2018). As Unwin (1977: 186; 1999: 2) has recorded, those who think thatUK quantitative geography…was all to do with Haggett and Chorley on a supposed ‘Cambridge (Chorley) to Bristol (Haggett) axis’ should note that there was also another axis from Manchester to London with two unsung and largely forgotten influences from climatologists Crowe and Gregory.The counterfactual – without Crowe and Gregory, how might British geography have developed in the 1950s–1960s? – is unanswerable, but there can be no doubt of their importance.…”