On the use of the Garin-Lowry activity allocation model in metropolitan area planning : a case study of Winnipeg

Thumbnail Image
Crews, Robert James
Journal Title
Journal ISSN
Volume Title
The current debate concerning the potential utility of large-scale, quantitative simulation models to the predicting of urban development processes is familiar to urban planners and analysts alike. This research examines the operation of the Garin-Lowry activty allocation model, and its ability to describe the structure of the Winnipeg urban system for a 'base-year' of 1971. A doubly-constrained gravity submodel is used to simulate work and shopping-trip behaviours as the foundations for the allocation of employment and population to subareas of the bounded metropolitan spatial landscape. A second economic base submodel is operated on functional-economic criteria via the minimum requirements technique. The zone system partitions the landscape into 9 superzones and 125 zones wherupon spatial resolution is increased relative to other similar research on the model; this design then constitutes a compartively rigorous test of the model's capabilities. The definition of the economic base, and the sensitivity of the associated multiplier, are found to be vital determinants of model performance. Winnipeg's spatial isolatedness and the nature of its role as regional service centre, introduce certain unique elements into the 'base' concept whererin the criticality of manufacturing, transportation, financing, and administrative functions to the metropolitan economy is suggested. Although the minimum requirements technique derives the most conceptually satisfying description of the base, the associated subdivision of individual industrial sectors into basic and service components is largely incompatible with the nature of the interaction data governing the internal allocation of employment. It is suggested that the economic base be realigned in accordance with a method, on either functional or locational criteria, wherein individual sectors are wholly classified as basic or service. This would be proved the most immediately frutiful means of effecting improvement on the comparatively poor correspondence in activity distributions derived in this case.