Estimating amount and spatial distribution of groundwater recharge in the Jullemeden Basin (Niger) based on 3H, 3He and CFC-11 measurements

Full text: The Continental Terminal (CT) formation of the Jullemeden Basin is located in southwest Niger (Africa) between 12.5-14.5 deg E and 2.5-4.5 deg N. The formation comprises three vertically structured aquifer compartments, which are mostly well separated by impermeable silt and clay layers. In this study only the uppermost aquifer (CT3) was investigated. The CT3 aquifer is unconfined and has a thickness of maximally 180 m and it is well sealed against the underlying CT2 aquifer. The average depth to the water table at the selected bore holes is 37.6m. The regional climate is dominated by semi-arid conditions where average annual rainfall is about 500-600 mm/y. Precipitation has found to be mainly convective. Rainfall events are usually short, intensive and very variable over short distances. About 90% of the precipitation occurs from June to September whereas from December to February virtually none occurs. Groundwater recharge was only detected in the top aquifer of the Continental Terminal 3 (CT3). Today's estimations of recharge amounts range from 10 to 100 mm/y. Leduc (1997) proposed that the primary infiltration process occurs through temporary pools and flow channels, and that spatial recharge is very small or negligible. However, stable isotope ratios of CT3 groundwater samples mostly coincide well with the local meteoric water line indicating that the water did not remain at the surface longer than a few days because the high evaporation rates lead to rapid isotope fractionation. All together, the today's picture of amount and spatial distribution of groundwater recharge is rather vague even though this information is essential to provide a reasonable management of the available groundwater resources. 3H, 3He and CFC-11 results from 23 bore holes of the CT3 aquifer were analysed to determine the recharge rates more accurately and to get more information about the spatial distribution of infiltration. The model used to simulate the transport of the three transient tracers is composed of a model for the unsaturated zone followed by a dispersion box model to determine the age distribution of the sampled groundwater volume. For the unsaturated zone transport the model approach of Cook and Solomon (1995) was used and extended for the transport of 3H in the water and the gas phase. A similar approach was used by Brennwald, et al. (2002) to estimate the average recharge rate into the CT3 aquifer. The transport equation was numerically solved with an implicit Crank Nicholson scheme assuming known atmospheric tracer concentrations at the top of the soil column and zero diffusion at the groundwater table. The tracer concentrations at the groundwater table are then used as input functions for the dispersion box model. When we fix the soil parameters (porosity, tortuosity and dispersion) three unknown parameters are remaining, namely the recharge rate, the mean residence time and the dispersion parameter. Since a high dispersion coefficient tends to an exponential age distribution it can be seen as an indicator for spatially distributed recharge. A small dispersion coefficient, on the other hand, tends to a piston-flow distribution and can be seen as local recharge at a distant location and subsequent transport to the sampling site. Therefore, it is possible to outline regions where recharge preferably occurs. Since the soil parameters are highly variable in space they have to be estimated within a certain range. Additionally, the tracers are measured with a different accuracy. To include these uncertainties a Monte Carlo simulation was applied considering the uncertainties of the soil parameters and the measurement errors. The results show that estimated recharge rates of 50-200 mm/y are still quite rough because the soil parameters were set in a wide range. Particularly a better estimation of the average soil wetness would lead to more accurate results. Since some samples fit well for small dispersion parameters and some do for large ones indicates that it is possible to determine regions where groundwater recharge preferably occurs. In our project area recharge seems to occur at the rim of the ancient river beds where surface runoff is gathered and leads to spatially limited high recharge events and possibly to the observed pools, when the soil is not permeable enough. (author)