Tobler’s Law and Spatial Optimization

In 1970, Tobler produced a movie simulating population growth in the City of Detroit. He argued that his model did not need to include terms for faraway places like Singapore, while still being relative accurate in his forecast by invoking what he called the first law of geography. In spatial optimization, like the general warehouse location problem (GWLP), it is assumed that all possible linkages need to be included, as arbitrarily dropping potential variables may prevent optimal solutions from being identified. In this article, it is demonstrated that it may be possible to meet such exacting standards in spatial optimization, while at the same time being guided by Tobler’s argument for being simple and frugal. This article gives a demonstration of how this might be achieved using the GWLP as an example. A new model form is proposed which distinguishes between “near†(low cost) allocations and “far†(high cost) allocations and uses both explicit and implicit variables for capacity allocations. Computational experience in using this new model is given which shows that optimal solutions can be identified and verified while eliminating a substantial number of allocation/transport variables and constraints. This article ends with a challenge for the reformulation and redevelopment of other spatial optimization problems in regional science.