Premium
Global Sensitivity Analyses of Three Pesticide Leaching Models Using a Monte‐Carlo Approach
Author(s) -
Soutter Marc,
Musy André
Publication year - 1999
Publication title -
journal of environmental quality
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.888
H-Index - 171
eISSN - 1537-2537
pISSN - 0047-2425
DOI - 10.2134/jeq1999.00472425002800040033x
Subject(s) - latin hypercube sampling , monte carlo method , environmental science , soil science , sensitivity (control systems) , soil water , groundwater , hydrology (agriculture) , mathematics , environmental engineering , statistics , geotechnical engineering , engineering , electronic engineering
The modeling of large natural systems requires a lot of data, carries a considerable amount of uncertainties and involves a significant computational effort. A global parameter sensitivity analysis, accounting for simultaneous parameter variations over their entire admissible ranges, is therefore potentially useful at all stages of the modeling process. The sensitivity analysis presented in this paper were carried out within the frame of a regional vulnerability assessment regarding ground water contamination by pesticides. The three selected pesticide leaching models include an analytical solution of the convection‐dispersion equation (attenuation factor, AF), a tipping bucket model (LEACHA), and a numerical solution of the convection‐dispersion equation (LEACHM). Sensitivity analysis was performed with a Monte‐Carlo based technique involving Latin hypercube sampling along with multiple regression. The results show that the parameters controlling degradation and sorption processes are far more important than those accounting for the transport process. Actually, pesticide properties (degradation rate and partition coefficient), organic C content and water table depth are the most predominant variables regarding cumulated (10 yr) pesticide fluxes to the ground water. Unexpectedly, variables such as soil hydrodynamic properties, dispersivity, or pesticide solubility did not have a significant effect on the magnitude of cumulated pesticide fluxes. As a consequence, in a stochastic approach to regional‐scale ground water vulnerability assessments, the probability density functions of those variables may be chosen empirically or according to a classification of soils combined with pedotransfer functions, without affecting significantly the resulting assessment maps and their uncertainties.