Probabilistic comparisons of grey water footprint models for herbicide mixtures in Brazilian sugarcane

Abstract The grey water footprint (GWF) quantifies freshwater pollution from pesticide use. The H-model relies on regulatory concentration limits, while the P-model uses ecotoxicological endpoints and concentration addition (CA); independent action (IA) across modes of action (MOA) remains underexplored. We present a fully re-verified probabilistic Monte Carlo comparison of CA and HRAC-based IA P-models for 17 herbicides applied to Brazilian sugarcane, built exclusively from traceable, source-cited inputs, including a soil attenuation factor recomputed from first principles and maximum acceptable concentrations compiled from five source tiers. Total grey water footprint is $2.3\times 10^{12}$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> mml:mrow mml:mn2.3</mml:mn> mml:mo×</mml:mo> mml:msup mml:mn10</mml:mn> mml:mn12</mml:mn> </mml:msup> </mml:mrow> </mml:math> m $^3$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> mml:mmultiscripts mml:mrow/ mml:mrow/ mml:mn3</mml:mn> </mml:mmultiscripts> </mml:math> under both CA and IA, with HRAC group C1 dominating in 100% of Monte Carlo iterations. The P-model exceeds the H-model by a factor R of 7 to 20 depending on data quality. Carfentrazone’s score rises from 1.5 to 4.9 once group membership is considered, and group E2’s hazard is driven mainly by oxyfluorfen. We report this audit trail transparently, prioritising numerical traceability over a single headline figure.