Coagulation is the main process for removing natural organic matter (NOM), considered to be the major disinfection by-products (DBPs) precursor in drinking water production. In this work, k-means clusters analysis were used to classify influent waters from two different surface drinking water treatment plants (DWTPs) located in the Mediterranean region. From this, enhanced coagulation models based on response surface methodology (RSM) were then developed to optimise coagulation at two water catchments (river and reservoir). The cluster analysis classified the water quality of the raw waters into two groups related to baseline and peak organic loads. The developed enhanced coagulation models were based on the turbidity, total organic carbon (TOC) and UV₂₅₄ removals. Sensitivity analysis applied to the models (after predictors selection) determined the factors relative individual contributions for each DWTP scenario. Then, profile plots for enhanced coagulation were studied to identify the optimal levels for each case. Models mean R² were 0.85 and 0.86 in baseline and 0.85 and 0.84 in peak scenario for river and reservoir catchments, respectively. Results of this study indicate that the surface water quality variation in river DWTP is seasonal and is expressed by an increase of turbidity, while in the reservoir DWTP is related to extreme weather events showing high levels of dissolved organic load (TOC and UV₂₅₄). During baseline cases, where raw waters present low levels of organics, the three factors optimal adjustment should be ensured to optimise coagulation. Then, during peak scenarios, where influent waters present high organics, the optimal for enhanced coagulation relies on the correct adjustment of C_d. The presented work provides models for drinking water production aimed to propose the optimum conditions for enhanced coagulation, considering the influent water characteristics under different weather conditions.