Fig 1.
Map of the study area and spatial distribution of microhematuria (typical symptom of S. haematobium in school-aged children) prevalence created using the following data sources: Major rivers and town locations were obtained from CERSGIS, Accra, Ghana; hillshade relief surface was created from elevation data obtained from ASTER Global Digital Elevation Model (v2); mining locations were digitized from Sentinel-2 satellite imagery; microhematuria prevalence data were collected by A. Kulinkina.
Fig 2.
Spatial definitions associated with the analysis conducted at the “community” level.
Table 1.
Summary of surface reflectance data used in the study.
Table 2.
Six environmental indices computed with Landsat 8 (OLI) and Sentinel-2 data.
Table 3.
Environmental (top) and WASH (bottom) predictor variables.
Fig 3.
Modeling approach explaining raster data extraction methods to be matched to each point-prevalence location.
Fig 4.
Schematic images of study communities showing settlements and water bodies (A1 and B1), NDWI values (A2 and B2), and MNDWI values (A3 and B3) generated using Landsat 8 data.
Table 4.
Results of environmental models for various extraction masks showing the R2 value and Spearman’s rank correlation value (r) between model predicted and observed prevalence values.
Fig 5.
Predicted prevalence from Landsat 8 data (A) and Sentinel-2 data (B) for five extraction masks {1, 2, 4, 5, and 6}.
Surface water access points are shown as + symbols. The scale and extent of the image match Fig 4(B).
Fig 6.
Predicted prevalence for the entire study area; for two smaller zoom windows (A and B); and variable importance values for the final model conducted with Landsat 8 environmental, topographic, and WASH variables.
The scales and extents of A and B match Figs 4 and 5. Surface water access points are shown as + symbols.