Coupled hydrological–economic modelling for optimised irrigated cultivation in a semi-arid catchment of West Africa

Abstract

A coupled model system, consisting of a distributed hydrological model and an economic optimisation model, communicating via model interfaces, is developed and applied to investigate regional interdependencies between irrigated agriculture and regional water balance and to identify optimised cultivation strategies. The coupled model is employed in the context of a case study in the Atankwidi catchment in the northern Guinea Sudan zone of West Africa. The physically based hydrologic model WaSiM [Schulla, J., Jasper, K., 2001] is used to simulate the water balance of this catchment including a small reservoir–irrigation system complex. The economics of irrigated crop cultivation are optimized under hydrological constraints using the non-linear optimisation model GAMS-ECIM, encoded in GAMS (General Algebraic Modelling System).

The coupled model system is utilized for empirical model testing under fictitious scenarios involving variable water availability for irrigation. Interdependencies between irrigation water application quantities and the water balance were identified and maximum agricultural profit was calculated using the coupled model system. The coupled model system is designed as decision-support tool for local authorities and agricultural stakeholders in the Ghanaian Upper East Region (UER).

Introduction

In the Volta Basin of West Africa, water availability is of great importance in supporting economic and social development. Population growth and increasing water demand in different sectors, in combination with high spatial and temporal variability of limited rainfall, underscore the need for sound interdisciplinary management of water distribution and allocation. The use of coupled hydrological–economic models permits integration of socio-economic behaviour and objectives into the physical water balance framework, and hence improves rational and sustainable water management. The coupled hydrological–economic model CMS-HYDROECON was developed by IMK-IFU and ZEF within the scope of the international project GLOWA-Volta (www.glowa-volta.de) to investigate regional interdependencies between economy and hydrology and to identify optimised irrigation strategies. It is applied within the Atankwidi catchment in the Guinea Sudan zone of West Africa, an important project research area, as a case study and prototype development, to be integrated within a prospective decision-support system for the planning, management and use of water resources in the Volta Basin. The scope of this study was to test the models' suitability and the operability of the model interface using fictitious scenarios involving variable levels of water availability. The coupled model system will provide policy simulation capacity in the Upper East Region where irrigation is expected to expand significantly (FAO, 2005a, FAO, 2005b, Van Edig et al., 2003). Policy-makers seek to know which development strategies are most cost-effective in a traditional sense (capital cost per ha of irrigation development), and which are most hydrologically efficient (yield per unit water). The well-situated Atankwidi catchment was chosen as an appropriate study area due to the availability of climatic, hydrologic and socio-economic data.

In the last decades, development and application of interdisciplinary coupled models has increased considerably, with several applications in water resources management. Against the background of limited water resources, growing populations, and projected increases in the demand for water in different sectors, problems of inter-sectoral water distribution and availability need to be solved (Young et al., 1994).

McKinney et al. (1999) evaluate a range of integrated hydrological–economic models, and subdivide them into holistic and compartment modelling approaches. While the holistic approach tightly combines both disciplines in a consistent model, the compartment modelling approach is based on a loose connection between the economic and the hydrological components (McKinney et al., 1999). Early compartment modelling approaches to integrating hydrology and socio-economic aspects are reported by Noel and Howitt (1982). To determine optimal spatial and temporal water allocation, a quadratic economic welfare function is incorporated in a multi-basin conjunctive use model modified and advanced by Lefkoff and Gorelick, 1990a, Lefkoff and Gorelick, 1990b. Economics of externalities in irrigated agriculture in the Colorado River basin were analysed by Lee and Howitt (1996) using a hydrology model and the Cobb–Douglas production function for regional production models. More recent studies developed combined models within the framework of decision-support systems for agro-economic and agro-ecological problems, for example Fohrer et al., 2001, Qiu and Prato, 2000, and Frede et al. (2002). Gillig et al. (2001) analysed different water use scenarios with respect to groundwater and surface water in Texas using a complex coupled hydrological–economic model for profit optimisation. Ecological and economic consequences of simulated flood scenarios were analysed by Duvail and Hamerlynck (2003). Quinn et al. (2004) coupled the hybrid hydrological model CALSIM II (“California Water Resource Simulation Model”) with the non-linear economic model APSIDE (“Agricultural Production Salinity Irrigation Drainage Economics Model”) for an integrated water management in the San Joaquin Basin (California, USA). An integrated hydrological–economic model was developed by Lanini et al. (2004) for the Hérault Basin (France). Bhuiyan (2005) coupled the hydrological model SWAT (“Soil and Water Assessment Tool”) with the linear economic model MIDAS (“Model of Dryland Agricultural System”) for the investigation of different land use scenarios and the development of optimal cultivation strategies in arid areas. Letcher et al. (2007) provide a generalised conceptual framework for representing the interactions of the stream system with water allocation, agricultural production and other water use decisions in integrated water allocation models.

As an early example of holistic model integration, the “Colorado River Network Model” (CRM) was developed to investigate hydrological consequences of droughts and different political water management strategies for the Colorado River (USA). The holistic model was studied by Harding et al. (1995) and upgraded by Booker and Young (1994) and Booker (1995). For water management in California, Howitt et al. (1999) developed the economic optimisation model CALVIN (“California Value Integration Model”). Jenkins et al. (2004) summarise the results of the long-term study of optimal water allocation in California. Optimal reservoir management and irrigation strategies for profit optimisation in Karnataka (India) were analysed with an integrated agro-economic-hydrological model by Vedula and Kumar (1996), which was improved by Mujumdar and Ramesh (1997). A GIS-based model for the investigation of regional water allocation strategies in the Aral Sea basin was developed by McKinney and Cai (1996). The optimisation model consists of a network flow hydrologic model containing source (inflow) and demand nodes, and representing important hydraulic (e.g., dams) and economic components; and optimizes water allocation while preserving mass balance. An advanced version was used by Rosegrant et al. (2000) in the Maipo Basin (Chile) to evaluate crop choice and irrigation technology options. To analyse alternative water management and policy strategies, the model approach was also used on the Mekong Basin in China (Ringler, 2001) and in the Dong-Nai Basin in Vietnam (Ringler and Nguyen, 2004). Cai et al., 2001, Cai et al., 2002, Cai et al., 2003 reapplied the advanced version of the agro-economic-hydrological model to the Aral Sea basin, focusing on the effects of salinization. Cai (2008) stresses the difficulties involved in large-scale holistic modelling for integrated river basin management. An alternative configuration is the spatial-dynamic model of structural change in agriculture developed by Balmann, 1995, Balmann, 1997, advanced and refined by Berger (2001) which utilizes a multi-agent/cellular automata approach by using heterogeneous farm-household models. Communication of the hydrological component with the agent-based economic component occurs via a raster-based GIS. The model was applied in Chile and selected regions of Baden-Wuerttemberg, Germany. Krol and Bronstert (2007) analyse climate change impacts on water resources in semi-arid Northeast Brazil by an integrated model that dynamically describes the relationships between climate forcing, water availability, agriculture and selected societal processes. Van Delden et al. (2007) present a Policy Support System (PSS) that incorporates socio-economic and physical processes in a strongly coupled manner and allows policy making in the fields of water management and sustainable farming.

In contrast to the holistic coupling approaches (above) based on comparatively simple conceptual hydrological models, this study utilizes a complex, physically based distributed hydrological model. The modelling approach corresponds with compartment modelling and facilitates continuous improvement of the model components. The study emphasizes the sustainable use of water resources and the preservation of a stable regional water balance. Results of the interdisciplinary hydrological–economic modelling approach will be presented for two fictitious scenarios in the Atankwidi catchment of West Africa.