Observed controls on resilience of groundwater to climate variability in sub-Saharan Africa.

Groundwater in sub-Saharan Africa supports livelihoods and poverty alleviation1,2, maintains vital ecosystems, and strongly influences terrestrial water and energy budgets3. Yet the hydrological processes that govern groundwater recharge and sustainability-and their sensitivity to climatic variability-are poorly constrained4,5. Given the absence of firm observational constraints, it remains to be seen whether model-based projections of decreased water resources in dry parts of the region4 are justified. Here we show, through analysis of multidecadal groundwater hydrographs across sub-Saharan Africa, that levels of aridity dictate the predominant recharge processes, whereas local hydrogeology influences the type and sensitivity of precipitation-recharge relationships. Recharge in some humid locations varies by as little as five per cent (by coefficient of variation) across a wide range of annual precipitation values. Other regions, by contrast, show roughly linear precipitation-recharge relationships, with precipitation thresholds (of roughly ten millimetres or less per day) governing the initiation of recharge. These thresholds tend to rise as aridity increases, and recharge in drylands is more episodic and increasingly dominated by focused recharge through losses from ephemeral overland flows. Extreme annual recharge is commonly associated with intense rainfall and flooding events, themselves often driven by large-scale climate controls. Intense precipitation, even during years of lower overall precipitation, produces some of the largest years of recharge in some dry subtropical locations. Our results therefore challenge the 'high certainty' consensus regarding decreasing water resources4 in such regions of sub-Saharan Africa. The potential resilience of groundwater to climate variability in many areas that is revealed by these precipitation-recharge relationships is essential for informing reliable predictions of climate-change impacts and adaptation strategies.

Measurements of water vapor isotope ratios with wavelength-scanned cavity ring-down spectroscopy technology: new insights and important caveats for deuterium excess measurements in tropical areas in comparison with isotope-ratio mass spectrometry.

The new infrared laser spectroscopic techniques enable us to measure the isotopic composition (δ(18)O and δ(2)H) of atmospheric water vapor. With the objective of monitoring the isotopic composition of tropical water vapor (West Africa, South America), and to discuss deuterium excess variability (d=δ(2)H - 8δ(18)O) with an accuracy similar to measurements arising from isotope-ratio mass spectrometry (IRMS), we have conducted a number of tests and calibrations using a wavelength-scanned cavity ring-down spectroscopy (WS-CRDS) technique. We focus in this paper on four main aspects regarding (1) the tubing material, (2) the humidity calibration of the instrument, (3) the water vapor concentration effects on δ, and (4) the isotopic calibration of the instrument. First, we show that Synflex tubing strongly affects δ(2)H measurements and thus leads to unusable d values. Second, we show that the mixing ratio as measured by WS-CRDS has to be calibrated versus atmospheric mixing ratio measurements and we also suggest possible non-linear effects over the whole mixing ratio range (~2 to 20 g/kg). Third, we show that significant non-linear effects are induced by water vapor concentration variations on δ measurements, especially for mixing ratios lower than ~5 g/kg. This effect induces a 5 to 10‰ error in deuterium excess and is instrument-dependent. Finally, we show that an isotopic calibration (comparison between measured and true values of isotopic water standards) is needed to avoid errors on deuterium excess that can attain ~10‰.

Estimate of recharge of a rising water table in semiarid Niger from 3H and 14C modeling.

A hydrodynamic survey carried out in semiarid southwest Niger revealed an increase in the unconfined ground water reserves of approximately 10% over the last 50 years due to the clearing of native vegetation. Isotopic samplings (3H, 18O, 2H for water and 14C, 13C for the dissolved inorganic carbon) were performed on about 3500 km2 of this silty aquifer to characterize recharge. Stable isotope analyses confirmed the indirect recharge process that had already been shown by hydrodynamic surveys and suggested the tracers are exclusively of atmospheric origin. An analytical model that takes into account the long-term rise in the water table was used to interpret 3H and 14C contents in ground water. The natural, preclearing median annual renewal rate (i.e., recharge as a fraction of the saturated aquifer volume) lies between 0.04% and 0.06%. For representative characteristics of the aquifer (30 m of saturated thickness, porosity between 10% and 25%), this implies a recharge of between 1 and 5 mm/year, which is much lower than the estimates of 20 to 50 mm/year for recent years, obtained using hydrological and hydrodynamic methods and the same aquifer parameters. Our study, therefore, reveals that land clearing in semiarid Niger increased ground water recharge by about one order of magnitude.

Occurrence of 1,1'-dimethyl-4,4'-bipyridinium (Paraquat) in irrigated soil of the Lake Chad Basin, Niger.

Increased use of agrochemical products to improve yields for irrigated crops in sub-Saharan Africa has been accompanied by a significant increase in the risk of environmental contamination. Detailed examples of the fate of pesticides after initial spreading on crop fields are scarce in tropical regions, where safe practices and related health risks are poorly understood by smallholder farmers. In the semi-arid environment of the Lake Chad Basin, SE Niger, both intrinsic properties of pesticides and extrinsic factors such as soil and climate helped to characterize processes leading to an accumulation of pesticides in soils. Analysis by HPLC-UV of a 6 m deep soil profile showed the presence of Paraquat at concentrations from 953 ± 102 µg kg(-1) to 3083 ± 175 µg kg(-1) at depths between 0.80 and 2.75 m below the land surface. Soil analysis revealed that up to approximately 15 % of the total soil matrix consists of smectites, a clay mineral capable of retaining cationic pesticides such as Paraquat, and a very low content of organic matter (<0.15 wt.% TOC). Paraquat could be stored and not bioavailable in a clayey barrier at approximately 2-m depth and therefore does not represent an immediate risk for populations or environment in this form. However, if the Paraquat application rate remains constant, the clayey barrier could reach a saturation limit within 150-200 years and 180-220 years if we consider a DT50 in soil of ~1,000 days (FAO). Consequently, it could lead to a deeper infiltration and so a pollution of groundwater. Such a scenario can represent a health risk for drinking water and for the Lake Chad, which is a major resource for this densely populated region of semi-arid Africa. Further analyses should focus on deeper layers and groundwater Paraquat contents to validate or invalidate the hypothesis of storage in this clay-rich layer.

Constraining groundwater modeling with magnetic resonance soundings.

Magnetic resonance sounding (MRS) is a noninvasive geophysical method that allows estimating the free water content and transmissivity of aquifers. In this article, the ability of MRS to improve the reliability of a numerical groundwater model is assessed. Thirty-five sites were investigated by MRS over a ∼5000 km(2) domain of the sedimentary Continental Terminal aquifer in SW Niger. Time domain electromagnetic soundings were jointly carried out to estimate the aquifer thickness. A groundwater model was previously built for this section of the aquifer and forced by the outputs from a distributed surface hydrology model, to simulate the observed long-term (1992 to 2003) rise in the water table. Uncertainty analysis had shown that independent estimates of the free water content and transmissivity values of the aquifer would facilitate cross-evaluation of the surface-water and groundwater models. MRS results indicate ranges for permeability (K = 1 × 10(-5) to 3 × 10(-4) m/s) and for free water content (w = 5% to 23% m(3) /m(3) ) narrowed by two orders of magnitude (K) and by ∼50% (w), respectively, compared to the ranges of permeability and specific yield values previously considered. These shorter parameter ranges result in a reduction in the model's equifinality (whereby multiple combinations of model's parameters are able to represent the same observed piezometric levels), allowing a better constrained estimate to be derived for net aquifer recharge (∼22 mm/year).