Enhancing the interpretability of a GIS method with ordered weighted averaging (GIS-OWA): A case study on ecological constraints on urban sprawl

The integrated application of geographic information systems (GIS) and ordered weighted averaging (OWA) has attracted significant attention because it reflects the preference to risk in decision-making and controls the trade-off between indicators. However, the GIS-OWA algorithm still lacks effective indicators to assist evaluators in judging which key inputs affect decision-making scenarios at specific spatial scales. We thus propose location contribution (LOCC) and attribute contribution (ATTRC) to explore the interpretability of the GIS-OWA algorithm at the pixel-to-local scale. Taking the fast developing Golden Triangle urban agglomeration of Southern Fujian Province of China as the study area, the new approach identifies the contribution of ecological factors from GIS-OWA-based ecological constraints. Results show that the aggregation phenomenon of ecological factors in the OWA-ranked layers provides a novel perspective for interpreting the principle of spatial heterogeneity discussed by Anselin and Goodchild. The set of key factors observed at the pixel scale indicates that spatially stratified ordered heterogeneity is ubiquitous. The key LOCCs and ATTRCs for cells in different OWA-equivalence areas differ significantly, and the set of key ATTRCs observed in a specific window suggests that the attribution of the GIS-OWA algorithm is feasible. LOCC and ATTRC efficiently probe key inputs to the GIS-OWA algorithm.