Premium
Quantifying the effects of uncertainty on optimal groundwater bioremediation policies
Author(s) -
Minsker Barbara S.,
Shoemaker Christine A.
Publication year - 1998
Publication title -
water resources research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.863
H-Index - 217
eISSN - 1944-7973
pISSN - 0043-1397
DOI - 10.1029/1998wr900005
Subject(s) - sensitivity (control systems) , range (aeronautics) , estimation theory , bioremediation , groundwater , mathematical optimization , uncertainty quantification , environmental science , biological system , computer science , mathematics , statistics , contamination , ecology , geology , engineering , geotechnical engineering , electronic engineering , biology , aerospace engineering
This paper describes a method for quantifying the economic and environmental effects of uncertainty in biological parameter values on optimal in situ bioremediation design. The range of uncertainty in model results associated with a range of input parameter values is quantified for both individual parameter errors and errors in combinations of parameters. Three measures of sensitivity are presented that quantify different aspects of the effects of model error on an implemented optimal policy. Numerical results are presented for an example site contaminated with phenol, with parameter ranges derived from values reported in the literature. For the example site, K s (the substrate half‐velocity coefficient in the Monod kinetic equation for biodegradation) was found to be the most sensitive biological parameter and this sensitivity was asymmetric; i.e., reductions in the value of K s have a much greater effect than increases in the value of K s . The methodology applied in this paper could also be applied to other water resource management problems, allowing the user to quantify the effects of wide ranges of possible parameter values on model results. The method is particularly useful for computationally intensive optimization models, as it requires a manageable number of model runs, and for the many situations where insufficient data are available to permit accurate estimation of probability distributions.