Spatial Association Analysis Between Population Distribution and Ambient Gamma Radiation Dose Rate in the Raška Region, Southwestern Serbia

ABSTRACT

Integrating spatially explicit population data with environmental measurements can improve the interpretation of potential human exposure within inhabited areas. This study examines the spatial correspondence between dasymetrically modeled population distribution and ambient γ‐radiation dose rate in the Raška region of southwestern Serbia. An imperviousness‐based dasymetric model estimated population per pixel, expressed as residents per 10 × 10 m cell. Ambient γ‐radiation dose rates were measured at 42 field locations at ground level and 1 m height. Inferential analyses used 40 locations with complete data for population per pixel, terrain slope, land cover and geology. Spearman correlation, point‐level bivariate Local Moran's I (BiLISA) and ordinary least squares (OLS) regression were used to assess global and local associations. IDW‐interpolated γ‐radiation surfaces were used solely for descriptive spatial visualization of their co‐occurrence with populated areas. Spearman correlations showed no statistically significant monotonic association between measured γ‐radiation and either population per pixel or terrain slope. Exploratory BiLISA identified local associations at 10.0%–15.0% of locations, depending on measurement height and predictor, using an unadjusted pseudo‐ p ‐value threshold of 0.05. The OLS models explained a small proportion of the variance in log‐transformed γ‐radiation dose rate ( R ²  = 0.078 at ground level and R ²  = 0.075 at 1 m) and were not statistically significant overall. These findings provide limited evidence of correspondence between population per pixel and measured γ‐radiation across the sampled locations. The dasymetric model provides spatial context for interpreting potential exposure, while statistical analyses remain based on field observations rather than interpolated raster cells.