Sensitivity Analysis of Stream Water Quality and Land Cover Linkage Models Using Monte Carlo Method

Abstract

Sensitivity analysis might be considered as one of inevitable steps in modelling since it would help to determine the behaviour of model, which was developed for further application. Sensitivity analysis was not paid much attention in studies that have been conducted for modelling the relationship between stream water quality and land cover except machine learning techniques such as artificial neural networks was applied for specifying the possible relationship between alteration in area (%) of land cover types and changes in water quality variable. Two linkage models for predicating stream water total nitrogen (r2= 0.70, p<0.01) and total phosphorus (r2=0.47, p<0.01) concentrations were developed using multiple regression approach in twenty-one river basins in the Chugoku district of west Japan. Application of Monte Carlo method-based sensitivity analysis indicated that TN regression model would be able to predict stream water concentration between 0.4-3.2 mg/L without any possibility for generation of negative value. For the TP regression model, predicting capacity would vary between 0.04, 0.32 mg/L. The results revealed that the Monte Carlo method-based sensitivity analysis would provide reliable information for determining output space in which the model would accurately respond.