Using spatial heterogeneity to strengthen the neighbourhood effects of urban growth simulation models.

The transition rules of cellular automata (CA) werestrengthened by incorporating annual growth rate (AGR) to reflect spatially heterogeneous neighbourhoods. We conducted hotspot (AGRHST) and gradient (AGRGDT) analysis of AGR to generate two spatial heterogeneity layers, then constructed three CA models: AGR-CA, AGRHST-CA and AGRGDT-CA. The modeling results showed that AGR-CA performed best in overall accuracy, AGRGDT-CA was superior to other models in terms of landscape pattern, and AGRHST-CA produced unrealistically smooth boundaries. The simulation accuracy of these models exceeds 89%, indicating that logistic regression-based simulation methods can be substantially improved by strengthening the neighbourhood effects. [ABSTRACT FROM AUTHOR]