Machine learning-based mapping of fog water harvesting potential in Pithoragarh, Uttarakhand: Evaluating climate scenarios and geospatial influences

Fog is crucial to the hydrological processes in Pithoragarh district, Uttarakhand, significantly affecting local water systems and ecosystems. This study employed five machine learning (ML) models, Gradient Boosting Machine (GBM), AdaBoost (ADB), Model Averaged Neural Network (avNNet), Naive Bayes (NB), and Shrinkage Discriminant Analysis (SDA) to map fog water potential in the area. Using data from 100 foggy locations and 23 variables (hydro-climatic, topographical, and terrain), the models were rigorously evaluated. Results showed that GBM, ADB, avNNet, NB, and SDA models identified high fog water potentiality classes in 43.91%, 43.84%, 44.79%, 42.97%, and 34.16% of the area, respectively. ADB and GBM performed best (AUC = 0.999), followed by avNNet (AUC =0.967), SDA (AUC =0.967), and NB (AUC =0.941). Key factors influencing fog occurrence included elevation, wind speed, wind exposition index, relative humidity, and mean power density, identified using the ordinary least square (OLS), various nature-inspired algorithms (such as GA, PSO, GO, DFO, HHO, and GWO), Pearson correlation, and Boruta sensitivity analysis. In this investigation, the linear regression approach was also applied to utilize ensembles of EC-Earth3, NorESM2-LM, and MIROC6 CMIP6 models for fog water forecasting climate conditions spanning from 2025 to 2055. Approximately 22.81% and 21.52% of the study area consistently exhibit very high potential for fog water harvesting (FWH) under the ssp245 and ssp585 scenarios, respectively. This research lays a foundation for addressing environmental concerns related to FWH and represents a significant step towards mitigating water scarcity, contributing to water security in the eastern Himalayas in line with Sustainable Development Goal 6 (SDG 6). • Evaluated fog water potential in Pithoragarh using five ML models with geospatial factors. • GBM and AdaBoost models demonstrated the highest accuracy in mapping fog water potential. • Key factors influencing fog potential included elevation, wind speed, and humidity. • Projected future fog water harvesting potential under climate scenarios, supporting SDG 6.