Sustainable water management in coastal irrigated plains.
Physical and probabilistic modelling

Application to the Cap Bon, Tunisia

Abstract

The Cape Bon is located 50 km South-East of Tunis. It is one of the most productive agricultural areas in Tunisia. But in the same time, it is suffering from scarce water resources and salinization of the groundwater due to seawater intrusion in the coastal aquifer. The local water management authority developed alternative sources of freshwater by constructing dams and surface water irrigation systems. However, because of the large variability of the climatic conditions, farmers still need to exploit the groundwater resource. The World Bank is noticing that productivity is still increasing while the available resource is already fully used. They expect major water problems in the next decade.

The aim of the research is to develop, through a case study in a regionally important aquifer, field and modelling techniques to monitor, analyze and forecast the behavior of mediterranean coastal aquifers.

The first step of the project is to improve the understanding of the groundwater situation in the area today. The last complete assessment was conducted in 1996. By collecting field data related to the exploitation, the piezometry, the chemistry, and the salinity, we want to define  precisely the status of the resource. It is also necessary to monitor the dynamic of the system by installing continuously recording instruments. This is required because we expect that the system may react at very different time scales depending on the forcing conditions (storms and rapid aquifer recharge, long term decay due to exploitation, salinization during exploitation, etc.). The field data will be complemented by remote sensing imagery in order to combine punctual field observations with distributed remote sensing cover.

The knowledge gained by this fieldwork will be used to model the system in a deterministic and probabilistic approach. The probabilistic approach represents in our point of view a requirement since many observations will be incomplete. The resulting uncertainty should be evaluated in a stochastic framework in order to be able to produce forecasts with error bars to the managers. This task is highly challenging since the system has many degrees of freedom and therefore the uncertainty analysis will be very demanding in terms of computer power.