Unravelling the Role of Socio-Physical Drivers for Potential Built-up Site Selection in the Kumaun Himalayas Using GIS-Based Fuzzy-AHP and Machine Learning

Abstract

Rapid and uncontrolled urban growth in the Kumaun Himalayas in absence of proper land use policy has pushed built-up areas towards the tectonically and ecologically sensitive regions, reducing the availability of suitable built-up land while simultaneously increasing the vulnerability of both communities and environment. The identification of areas for sustainable built-up growth is of paramount importance to address the challenges arising from unregulated urban expansion. In this study GIS-based Fuzzy-AHP technique and machine learning algorithms (SVM and BN) were employed to delineate the potential built-up sites selection in Hawalbagh Block, Uttarakhand (India) using nine socio-physical drivers, including slope, aspect, LU/LC, distance to road, distance to drainage, distance to lineament, distance to landslide, distance to settlement, and lithology. The suitability maps generated by the three methods were validated using AU-ROC analysis, which demonstrated that each approach produces outstanding results with AU-ROC values more than 0.90. The comparison of the approaches shows that SVM (AUROC=0.99) outperforms BN (0.95) and GIS-based Fuzzy-AHP (0.90). The suitability maps were classified into five suitability classes. Assuming that very high and high suitability classes are acceptable for built-up expansion, the study identified potential built-up locations in the study region covering an area of 148.86 km2, 85.23 km2, and 55.25 km2 according to the Fuzzy-AHP technique, SVM model, and BN model, respectively. The suitability zonation in this study can serve as a foundation for the development of land-use policy or the formulation of master plans aimed at achieving a sustainable mountain ecology in the Kumaun Himalayas.