How Bad Isn't the Agent-Based Model CATCHSCAPE?

Abstract: Due to mounting human pressure, stakeholders in northern Thailand are facing crucial natural resource management issues. The impact of upstream irrigation management on the downstream agricultural viability is a common source of conflict. It has often both biophysical and social origins. CATCHSCAPE has been developed as an Agent-Based model that enables us to describe the whole catchment: hydrology, farmers ’ behaviour and water management rules. It is meant to simulate scenarios based on assumptions about value of these features as well as some assumptions about context, such as levels of prices for various commodities or climate. The biophysical modules are made of a hydrological system with its distributed water balance, irrigated schemes management, crop and vegetation dynamics. The social dynamics are described as a set of resource management processes (water, land, cash, labour force). Water management is described according to the actual different levels of control (individual, scheme and catchment). Virtual experiments according to a first defined plan are made with two aims: sensitivity analysis of the model through variation of different parameters and extreme scenarios on one hand; overall behaviour of the basin under various realistic scenarios on the other hand. Both sets are meant to give more insight on the consequences of this very virtual catchment behaviour and improve the collective understanding on the real basin. Simulations show that the model is quite robust from a variation of results point of view and help to identify key factors such as farmers ’ representation on the expected amount of water for a cropping season or pluri-annual climatic trends