A GRACE-Streamflow Land Surface Model Calibration Approach for Improved Baseflow and Water Table Simulations over the Highly Managed Upper-Nile Basin of East Africa

Abstract:

Land surface models (LSMs) are useful tools for understanding behaviors of land hydrologic variables at different time and spatial scales. LSM outputs, however, are marked with great uncertainties resulting from the simplified assumptions on the parameterization and processes of the land surface and a poor representation of both the natural and anthropogenic controls on the system. The Upper-Nile basin, over Uganda, Kenya and Tanzania, is one region that is characteristic of significant human controls on streamflow, including Lake Victoria releases. The river Nile flow from Lake Victoria follows apriori rating curves that are not simulated by LSMs. Apart from management practices; the huge storage volume of Lake Victoria also modifies the seasonal characteristics of the Upper-Nile discharge, creating small seasonal variations in stream flow. In this study we calibrate several critical parameters in the Community Land Model (CLM.v4) in a multiobjective framework using total water storage anomalies (∆TWS) from GRACE, observed total runoff (Q) and estimated baseflow (BF) over the Upper-Nile basin. The goal is to improve the CLM parameters so that the model simulates the agreed curve (apriori) streamflow and baseflow with a better accuracy. We demonstrate the significance of improved parametrization by comparing model results of ∆TWS, Q and BF with a combination of insitu and estimated observations. Preliminary results based on RMSE statistics show that with calibration, simulations of ∆TWS, Q and BF achieve higher performance. Further, an improvement in the model’s capacity to simulate the water table depth is also evident with the calibration. Such results provide a basis for using CLM for other hydrologic experiments that could guide water resources management in this highly managed basin.