Dynamically Coupled Socioeconomic system dynamics models integrated with physically-based Environmental Models (DCSEM) can capture relationships between complex environmental and socioeconomic systems, and are promising tools for participatory environmental management involving the integration of various viewpoints, disciplines and processes for sustainable water resources management. However, the application of DCSEMs has been limited for many reasons, including the complexity of the model coupling process (i.e., the lack of a flexible model coupling approach), issues of over-parameterization, high parameter uncertainty, intensive computational requirements (due to many interacting parameters), and the possibility of eliciting behaviour-pattern oriented model outputs. As such, most conventional (numerical) procedures for model evaluation (i.e., analysis of point values of modelled results) that have been implemented for physically-based models are not applicable to DCSEMs. In order to address these challenges, this study developed a novel automatic behaviour-pattern global sensitivity analysis (GSA) procedure to determine the influence of input parameters on the general behaviour trends (rather than numerical point values) of coupled model outputs. The developed behaviour-pattern GSA procedure was implemented in an existing software (Tinamït), developed previously by the authors, to ensure ease-of-use. This study investigates the suitability of the proposed behaviour-pattern GSA procedure for the analysis of DCSEMs by comparing the proposed procedure with the conventional numerical procedure. The numerical and proposed behaviour-pattern procedures, coupled with the Morris (qualitative) and EFAST (quantitative) GSA methods, were applied to a DCSEM to rank and screen parameters in a water table depth simulation in Pakistan. The determination of important parameters facilitates subsequent model calibration and groundwater management. It was found that the Morris and EFAST methods achieved similar parameter ranking results in the numerical and behavioural procedures, respectively. The results also indicated that the behaviour-pattern GSA procedure offers more information about several important parameters, and a variety of parameter ranking orders, compared to the numerical GSA procedure, regardless of which GSA method (Morris or EFAST) was used. The increased information obtained through the application of the developed behaviour-pattern procedure confirms that the DCSEM model outputs are behaviour-oriented. It is recommended that the proposed behaviour-pattern GSA procedure be used with the Morris method, which has a higher computational efficiency (>150 times) than the EFAST method, to detect important parameters in DCSEMs.