A comparative assessment of grey water footprint estimation methods in paddy fields

Reliable estimation of the gray water footprint (GWF), a key indicator for assessing freshwater pollution, remains limited due to uncertainties in estimating leaching-runoff fractions and other assumptions in conventional methods. To address this gap, we applied three different methods for estimating the GWF during the rice production process in a subsurface-drained paddy field under mid-season drainage (MSD) and alternate wetting and drying (AWD) strategies: (i) a simple tier-1 approach (T1M), (ii) a field-based estimation method (FBM), and (iii) a modeling approach using the HYDRUS-2D model (HM). During four rice growing seasons in 2011–2015, nitrate concentrations were monitored in drainage water of three subsurface drainage systems with drain spacings and depths of 30 m and 0.90 m (D0.90L30), respectively, 30 m and 0.65 m (D0.65L30), and 15 m and 0.65 m (D0.65L15). The HYDRUS performance was evaluated using the root mean square error (RMSE), normalized root mean square error (nRMSE), normalized mean bias error (nMBE), and model efficiency (EF). The values of RMSE (0.25–0.99 mg L−1), nRMSE (5.3–19.1 %), nMBE (-2.14–10.47 %), and EF (0.6–1.0) were within the acceptable range in the calibration and validation periods under MSD and AWD. The GWF determined using FBM ranged from 17 to 2995 m3 t−1, from 701 to 1368 m3 t−1 using T1M, and from 19 to 2770 m3 t−1 using HM. The average GWFs determined using FBM, T1M, and HM were 158, 783, and 171 m3 t−1 under MSD, and 1808, 1053, and 1840 m3 t−1 under AWD, respectively. While a difference of up to 3961 % was observed between GWFs estimated by FBM and T1M, HM provided closer values to FBM, with relative errors ranging from 5 % to 12 %. The results indicate the HYDRUS-2D model performed well in simulating nitrate losses under MSD and AWD. Therefore, HYDRUS-2D can provide acceptable estimations of the GWF compared to the conventional and widely used T1M.