Floods and wetlands: combining a water-balance model and remote-sensing techniques to characterize hydrological processes of ecological importance in the Tana River Delta (Kenya)

Abstract

The Tana River Delta (TRD) provides a multitude of ecosystem services for the local communities including fishing, farming and livestock keeping. The hydrological regime of its river determines for a large part the environmental health of the delta. The devel- opment of upstream irrigation schemes and hydroelectric infrastructure can seriously impact the ecological status of the TRD. The Tana Inundation Model {TIM) presented here is the first known hydrological model of the TRD. Using it, we quantify essential hydrological variables of ecological importance for 2002-2011 such as flood extent and duration, flood timing and frequency, flood peaks and water height. TIM also provides an annual water balance. The model simulates river inflows and outflows, precipitation, overland flow, evapotranspiration and infiltration. The TRD is characterized by scarce hydrological data and a high cloud cover limiting the use of many remote sensing tech- niques. The methodology therefore combined a conventional water-balance analysis with the extraction of inundation extents from MODIS satellite imagery at a medium spatial and temporal resolution. In non extreme years and for the actual configura- tion of the Tana River, the flooded area exceeds 560km2. Floods over 200 km2 occur approximately every two years, with a mean duration of less than 25 days. River dis- charge from the upper catchment counts for over 96% of the total water inflow. This study provides the first known estimates of these variables for the Tana River Delta and is therefore primordial for the management of the water and other natural resources of the zone. The hydrological model based on the Generalized Likelihood Uncertainty Estimation (GLUE) is generic enough to be applied to other catchments with scarce hydrological data.