ESTIMATING VILLAGE-LEVEL DRINKING-WATER ACCESSIBILITY IN SINDH, PAKISTAN, FROM OPEN GEOSPATIAL DATA: WHY ANNUAL RAINFALL INVERTS THE RANKING

Abstract

Household surveys are the main way Pakistan tracks progress toward Sustainable Development Goal 6.1, but they are expensive, run infrequently, and cannot resolve drinking-water access below the district level. This study builds the Sindh Water Access Index (SWAI), a reproducible composite index scoring relative drinking-water accessibility for 5,159 villages in Sindh from open geospatial data alone: HydroRIVERS and OpenStreetMap hydrography, WorldPop population, CHIRPS precipitation, Sentinel-2 NDWI, and SRTM elevation. Indicators were log-transformed, winsorized at the 1st and 99th percentiles, min-max normalized, then combined with rainfall and canal proximity treated as substitutes rather than as additive terms. The main finding is methodological. Using annual precipitation, which is standard practice in composite water indices, inverts the district ranking and places Tharparkar - the most water-insecure district in the province - among the best served. Tharparkar is not dry by annual total. Mithi receives about 277 mm a year against 148-169 mm in the canal-command districts, but it has no perennial supply, deeply saline groundwater, and no storage, so the rain it does get evaporates or soaks away unused. Precipitation totals are not a weak proxy for availability here; they are close to an inverted one. Substituting dry-season precipitation, which stands in partly for storage-independent supply, moved Mithi from rank 1 to rank 25 of 25 and reversed the rainfall-proximity correlation from −0.62 to +0.34. District rankings hold across six weighting variants (Spearman ρ = 0.906-0.958). A second finding is structural: for 88.6% of villages the nearest mapped perennial feature is an irrigation canal rather than a natural river. The index measures accessibility, not potability. Coastal Thatta and Badin score among the highest-ranked districts even though field studies report 100% of Thatta groundwater samples above WHO chloride limits, a decoupling confirmed here at district scale and against 21 georeferenced groundwater samples. The pipeline and dataset are openly archived (DOI: 10.5281/zenodo.21422759).