Eco-hydrology modelling in arid areas : Study of root density impact on water fluxes in the Sahelian region

The Sahelian region experienced intense droughts between 1970 and 1990 and despite a precipitation « recovering », soils remain degraded and a decrease in the soil ability to infiltrate - an essential characteristic for vegetation recover – is observed, together with an increase of desertification and of eroding floods frequency. To tackle with this phenomenon, coordinated agricultural strategies, like the Great Green Wall project, have been encouraged and spread over large areas through NGOs. This consists in applying agro-ecological practices, like micro dams, to harvest water, favor infiltration, and hence vegetation growth. These strategies still require critical assessments and optimisation. To study such agroecological practices we are developing a modelling framework based on ParFlow-CLM to simulate the interactions between surface hydrology and vegetation in the context of crusted Sahelian soils where water transfers are highly dependent on both surface hydrodynamical properties and root distribution below the surface. Indeed, the very thin eolian crust acts as a hydraulic discontinuity that slows down soil evaporation transfers but not transpiration, which benefits from the roots below the crust and from the stems which bridge to the atmosphere. However, since the CLM family models were designed for large scale with a relatively thick mesh at the surface, the root density function proposed as a decreasing exponential function distribute the majority of roots just below the surface. This disposition is irrelevant for finer millimeter underground meshes modelling, particularly in areas with a hot and dry climate such as the Sahelian one, that dries very rapidly the first centimetres of soil. Thus, In the CLM land surface model framework and according to literature, we propose a new root distribution in the soil, using a parameterised function which is zero at the surface and at infinity, and adapt the maximum root density depth and the root concentration around this maximum. We compare the impact of both initial and proposed root functions on a Sahelian case study in Niger where all necessary data are available thanks to the AMMA-CATCH observatory. Studying this function highlighted simple causal relations between root density function parameterisation and evapotranspiration flux. By modifying the root density function, we can find a set of parameters corresponding to a better representation of transpiration, global evapotranspiration and soil moisture, and accordingly, a better representation of the studied ecosystem. Once this representation is relevant, a dynamic LAI based on allocation laws, available in the latest version of CLM, will then complete this modification of the vegetation scheme. We will then introduce the changes on surface due to agricultural practices and study the impact of their sizing.