Borehole validation dataset for a fuzzy logic-AHP groundwater potential model, Minna Basement Complex, north-central Nigeria
Description
Independent validation data for a GIS-based fuzzy logic-AHP groundwater potential model of the Minna area, Niger State, Nigeria. The dataset records 103 boreholes with documented yield outcomes (69 productive, 34 non-productive), together with the composite groundwater potential index and the value of each of the seven model criteria sampled at every borehole location. The boreholes were withheld entirely from model construction, so the dataset constitutes an independent test of both the model and its expert-assigned weights. Criteria sampled at each site are composite lineament density, Source Parameter Imaging depth to magnetic basement, geomorphon landform class, slope, drainage density, lithology and land use/land cover. Coordinates are supplied in WGS84 decimal degrees and in UTM Zone 32N. A README gives full column definitions, units, class-code tables and provenance. An accompanying Python script reproduces every validation statistic reported in the associated article from the CSV alone, including ROC analysis with bootstrap confidence intervals, criterion-level rank correlations with multiple-comparison correction, a permutation test of the landform association, and per-zone success rates.
Files
Steps to reproduce
All statistics reported in the associated article can be regenerated from minna_gwpz_borehole_validation.csv using the included script. Requirements: Python 3 with numpy, pandas and scipy. 1. Place minna_gwpz_borehole_validation.csv and reproduce_validation.py in the same directory. 2. Run: python reproduce_validation.py The script prints the mean groundwater potential index by yield class; Mann-Whitney U and point-biserial tests; ROC-AUC for the integrated model with a 5,000-replicate stratified bootstrap confidence interval; single-criterion AUCs and Spearman correlations for all seven criteria, with Benjamini-Hochberg and Bonferroni correction for multiple testing; a paired bootstrap comparison of SPI against the integrated model; a 20,000-permutation test of the landform-yield association; success rates for each mapped potential zone with Wilson score intervals and a trend test; and a land-cover comparison. Column definitions, units, class-code tables and provenance are given in README.md. The map-removal sensitivity analysis reported in the article additionally requires the criterion raster grids and is not reproducible from this file alone.