|Paper title||Hydrological Regime of Sahelian Small Water Bodies from Combined Sentinel2 MSI and Sentinel3 SRAL Data|
|Form of presentation||Poster|
In semi-arid regions and especially in Sahel, water bodies such as small reservoirs, small lakes, and ponds are vital resources for people. Most studies on inland waters in Africa focus on large lakes like Lake Chad for example, but the numerous lakes and ponds, which are found near almost every village in Sahel, are poorly known. These small water bodies (SWB) are critical in terms of water resources and important for greenhouse gases and biodiversity. SWB probably increased in numbers and surface recently, due to changes in land surface properties after the big Sahelian drought of the late 20ieth century (Gal et al. 2017, doi 10.5194/hess-21-4591-2017), and to dam building, as for instance in Burkina Faso. For a more detailed assessment of changes in water resources, it is necessary to quantify water volumes variability and hydrological regime of these SWB at the regional scale.
The objectives of this work are to develop methods to monitor water quantity of SWB by combining optical and radar remote sensing. This study is carried out over 3 countries (Niger, Mali and Burkina Faso) and addresses the water regime of 40 water bodies over the 2016-2021 period.
Water surface is derived from Sentinel2 optical data. Algorithms for water detection generally face two issues in this region: i) the high number of vegetated water bodies (floating vegetation, grasses or trees), ii) the extremely high and unusual reflectance of Sahelian waters. It turned out that a threshold on the MNDWI index chosen ad hoc for each lake, implemented in Google Earth Engine, is a fast and efficient method to estimate water areas. Water levels are derived from Sentinel 3 altimetry data processed with the ALTIS software (Frappart et al. 2021, doi 10.3390/rs13112196). Careful extraction is required for water bodies in close proximity, such as the Tanvi reservoirs in Burkina Faso, since multiple signals coming from neighbouring water bodies may mix in the radar data.
Water levels and matching water areas are combined to derive surface-height curves. This allows estimating water levels from any Sentinel2 data, and densifies therefore the water levels time-serie derived from Sentinel3 altimetric data alone. Time-serie of water levels are then used to estimate water levels decrease during the dry season (generally from November to June), which is compared to the evaporation loss from each SWB estimated using Penman's method and ERA5 data.
Given that during the dry season water inflow by precipitation is null, differences between water level and evaporation are due to water losses, or uptakes from anthropogenic activities or exchanges with groundwater or river networks. For the 40 SWB studied, evaporation averages about 7 mm/d during the dry season, whereas water losses vary significantly across different water bodies. Water bodies exposed to important pumping activities exhibit significantly higher water loss rates, than the evaporation rate, thus reaching a minimum water balance value around –12.5 mm/d. Others water bodies display the opposite situation, for example for lakes in the inner Niger Delta where the flood extends into the dry season and water is supplied by groundwater or river network, with water balance around 5.7 mm/d.
The results show the potential of the water balance approach in poorly observed semi-arid regions to better understand hydrological processes, including human management of reservoirs. This is particularly relevant for the forthcoming SWOT mission, which will enable this approach to be applied at the global scale.