Improving the calibration of an integrated CA-What If? digital planning framework

Cities continue to grow under the global urbanisation trend, and planners require enhanced evidence to understand the spatiotemporal evolutionary patterns of urban built-up land, as well as their uncertain future growth scenarios. Given these issues, a digital planning framework is here proposed that integrates Cellular Automata (CA) with What If? sub-models for simulating future urban growth scenarios. The XGBoost algorithm is used to calibrate the CA transition rules, while spatial partitioning mechanisms are used to mitigate the impacts of heterogeneity on modelling accuracy. The verified and optimised CA sub-model is then integrated with the What If? planning support system (PSS) sub-model to generate and analyse three representative built-up development scenarios. The modelling results demonstrate that the hybrid partitioning method, combining administrative regions using K-means clustering, can improve the simulation performance of this integrated digital planning framework.This dataset includes (1) all required data for reproducing the materials within the manuscript, (2) detailed Python codes of the proposed CA-What If? model, and (3) a step-by-step instruction document.