Neural Network Based Sensitivity Analysis of Natural Resource Models

The management of hydrologic catchments typically faces the challenge of keeping a reasonable balance between water quality demands and farming restrictions. In order to handle this problem it is most important to identify farming areas whose land use have a high influence on the nutrient leaching into the receiving stream. This identification can be regarded as a sensitivity analysis (SA). Changes in the land use of farming areas with high sensitivity usually promise an adequate relation between costs and benefits. In order to indentify sensitive farming areas, the processes responsible for nitrogen cycling and transport must be modelled first. Then a sensitivity analysis of the resulting model can be performed.