The relative accuracy of forecasting land use by five residential land use simulation models was tested. Explicit model statements and descriptions of the procedures used in calibrating the forecasts are given. All conditions for the test models were held constant except the interrelations among variables, so that differences among the forecasts would be caused only by inherent differences in the models themselves. Results of the study reported here cannot be considered definitive because of the differences in amount of information required and the differences in the degree to which the models were fitted to data. Also, the study represents only one of man possible test conditions.
Abstract