Groundwater vulnerability assessment is commonly conducted using the DRASTIC index based on fixed weights and ratings. However, some influential parameters, particularly groundwater depth and aquifer recharge, exhibit considerable temporal variability that may affect assessment results. This study investigated the effect of temporal uncertainty on the calibration of the DRASTIC vulnerability index in the Sari–Neka aquifer. For this purpose, 20 years of groundwater depth and precipitation time-series data were analyzed. Three statistical conditions, including Q25, median, and Q75, were defined to represent dry, average, and wet scenarios, respectively.



The results showed that under wet conditions, groundwater levels changed by approximately 20–50 cm across different parts of the aquifer, whereas under dry conditions, variations ranged from 30 to 70 cm. In the Q25 scenario, the extent of the very low vulnerability class increased, while the Q75 scenario led to a broader distribution of highly vulnerable areas.



The DRASTIC index was calibrated using RF-HHO and CNN-HHO models with weight and rating ranges of 1–7 and 1–10, respectively. The correlation coefficient between the original DRASTIC index and nitrate concentration was 0.37. After calibration, this value improved to 0.62–0.64 using RF-HHO and to 0.67–0.68 using CNN-HHO. The findings demonstrate that integrating temporal uncertainty with data-driven calibration can significantly enhance the accuracy of groundwater vulnerability assessment. Moreover, the sensitivity of the DRASTIC index depends on the combined effects of parameter weights, ratings, spatial distribution, and parameter integration methods.