An explicit representation of river-floodplains relationships in the integrated hydrological - land surface model CLM-PARFLOW

Floodplains, a type of wetland regularly flooded by large rivers, are important hydrological objects to document and understand. They are places where the hydrological risk can be highly damageable, and where the high frequency of saturated soil moisture conditions due to flooding sustains an important biodiversity, provides important ecosystem services for human communities and regulates hydrological flows and exchanges between the land surface and the atmosphere.

Despite this importance, floodplain dynamics are difficult to represent in large-scale hydrologic models because of the control that small-scale topography imposes on water flow and storage. Some coarse resolution large-scale models use simplified representations of floodplain dynamics at the subgrid scale. In these cases, the relationship between water height, water storage and flooded area is parameterized. It should be noted that this approach does not always capture the complex relationships between floodplains and other hydrologic processes. On the other hand, the use of finer scale integrated hydrologic models could explicitly represent the complex relationship between rivers, aquifers and floodplains, but at an burdensome computational cost.

Here, we propose a methodology to represent floodplains in the integrated hydrological model CLM-PARFLOW, at a relatively low computational cost that allows its use in large-scale and high-resolution implementations even with a kinematic wave approach for surface flows. We prescribe an anisotropic layer near the surface in areas that are “regularly flooded” to allow up-slope flows driven by water head gradient. This anisotropic layer is defined by a depth and a tensor factor affecting horizontal permeability, and allows connecting river grids with neighboring floodplain grids when the water level is high enough to flood. The computational cost is low, as it uses the current capabilities of PARFLOW to represent horizontal subsurface flow at high resolution. We apply this representation to the Ouemé River basin in Benin (47000km²), at a resolution of 1 km, to test and optimize the parameters controlling the anisotropic layer.

First results show an improvement of horizontal flows between rivers and floodplain areas, especially during wet and high river discharge seasons, and a better representation of hydroclimate variables like ET in these areas. This methodology will further be applied to improve an existing 1 km² PARLFOW simulation over the West Africa domain (3.5 Mkm²), an area with large scale floodplain areas and intermittent endoreic ponds where the coupling between wetlands, rivers and aquifers control low-water levels in the dry seasons, and induce preferential recharge parthways, and where agriculture and pastoralism feed millions of people in West Africa.