M.S. Applied Data Science - Capstone Chronicles 2025

16

Figure 13 Lead Forecast in the next 10 years

The Cadmium forecast shows a slow increase in

The Lead forecast remains remarkably low centered around 10–20 µg/L through 2035 with very little seasonal fluctuation. The wide confidence interval in early years (driven by rare high outliers up to 14,000 µg/L) narrows rapidly once the model settles on a largely flat trajectory. Although the EPA’s action level is 15 µg/L, the median forecast hovers slightly above that threshold, suggesting many systems will need continued corrosion ‐ control optimization to avoid intermittent regulatory action. These results reinforce the importance of maintaining orthophosphate dosing and pH control in distribution systems and indicate that full lead service ‐ line replacement remains the only sure path to eliminating episodic exceedances. Figure 14 Cadmium Forecast in the next 10 years

background levels from about 2 µg/L in 2025 to 6 µg/L by 2035 well below the U.S. EPA MCL of 5 µg/L but approaching it in the upper confidence band by the end of the decade. Historical data are sparse (few hundred samples) but include sporadic peaks to 12 µg/L; the model’s confidence intervals accordingly remain wider relative to the trend. This projection suggests that while routine compliance is likely, under ‐ sampling may mask local exceedances especially in areas with legacy industrial contamination. To reduce uncertainty, we recommend bolstering cadmium sampling frequency in counties with known mining or metal ‐ processing history and integrating sediment and soil geochemistry data to identify at ‐ risk water systems.

88