Decoding seasonal irrigation suitability of drain water using integrated indexical, statistical, graphical and...

Rapid urbanization and industrialization have exacerbated freshwater scarcity while simultaneously increasing wastewater generation, making the safe and judicious reuse of wastewater a viable strategy to address both challenges.

The present study aimed to evaluate the seasonal irrigation suitability of the Kala Sanghian drain using integrated (graphical, indexical, statistical and machine learning) approaches.

Drain water samples from two seasons (pre- and post-monsoon) were collected and analysed for physico-chemical parameters and heavy metals, and irrigation suitability was evaluated using different indices and machine learning models.

The results revealed a notable improvement in the physico-chemical quality of drain water during the post-monsoon season, enhancing its suitability for irrigation. Among the heavy metals analysed, 84.4% and 73.3% of water samples exceeded the Food and Agriculture Organization (FAO) permissible limit (0.10 mg L⁻¹) of chromium during the pre- and post-monsoon seasons, respectively. Whereas arsenic and cadmium remained within permissible limits during both seasons. Most samples were classified as excellent to good based on water quality index (WQI) in the post-monsoon season, whereas 73.3% were deemed unsuitable based on heavy metal pollution index (HPI) due to chromium contamination. Principal component analysis and correlation analysis revealed that drain water chemistry is jointly controlled by anthropogenic inputs, geogenic processes, and salinity, with pre-monsoon evaporation and post-monsoon dilution-leaching driving seasonal variations. Comparative evaluation of machine models for irrigation quality prediction showed that the Lasso regression model achieved the highest accuracy (R² ≈ 0.99 with the lowest errors), followed by Ridge and Gradient Boosting, while other models exhibited lower accuracy. Chromium (Cr) was the dominant predictor across all models, whereas most major ions contributed marginally to prediction performance.

Although the drain water showed improved post-monsoon irrigation suitability in terms of salinity and conventional indices, persistent chromium contamination poses a serious constraint to safe reuse.