Sensitivity analysis of a simple layer‐equilibrium model for the one‐dimensional leaching of solutes

Abstract

Simulated solute concentration profiles in soils are affected by inherent model limitations and by estimation and experimental errors in input parameter values. Sensitivity analysis can help investigators rank the relative importance of these errors, and focus on those parameters that have the greatest affect on profile estimates. A sensitivity analysis is presented for the simple layer‐equilibrium model Trace Element Transport (acronym TETrans). The sensitivity of the output to changes in the following physical, chemical and biological input parameters is shown: chemical adsorption coefficients (i.e., Langmuir affinity coefficient and Langmuir adsorption maximum), mobility coefficient (a term specific to TETrans which represents the fraction of soil water which is displaced strictly by piston‐type flow), field capacity water content, minimum water content (i.e., the water content at the wilting point), bulk density, maximum root penetration depth and root water uptake distribution. In addition, sensitivity to the soil layer thickness, and to the initial and boundary conditions is presented. Sensitivity of the simulated output to a given parameter was determined by two approaches: (1) by setting all other parameters to their mean value and letting the selected parameter vary by one standard deviation above and below its mean value and (2) by setting all other parameters to their mean value and letting the selected parameter vary by a fixed percentage of the mean (i.e., +5% of the mean). The mean values are determined from measured values taken from a soil lysimeter column study. Results show that the simulated output over a fixed range of input parameter values for a reactive solute are most sensitive to the adsorption parameters and the bulk density; less sensitive to the maximum root penetration depth, the field capacity and the mobility coefficient; and least sensitive to the plant root water uptake distribution and the minimum water content. For input parameters ranging from plus‐or‐minus one standard deviation of their mean measured value, TETrans is most sensitive to the Langmuir affinity coefficient and the mobility coefficient; less sensitive to the Langmuir adsorption maximum, the bulk density, and the field capacity; and least sensitive to the minimum water content.