The study uses advanced metrics like the nugget-to-sill ratio (NSR) to evaluate spatial variability and covariability between auxiliary and response variables. Could the authors elaborate on the rationale for the selected NSR thresholds (e.g., 10%, 35%) in determining the sampling design? Additionally, given the potential variability in RS data quality across geographic regions, how might these thresholds adapt to ensure consistent applicability in real-world implementations?
