EVALUATING THE PERFORMANCE OF LAND SURFACE MODELS USING GRACE DATA OVER AFRICA

We utilize the Gravity Recovery and Climate Experiment (GRACE) and land surface models (LSM: GLDAS and CLM4.5) in conjunction with other readily-available datasets for monitoring the spatial and temporal trends in Terrestrial Water Storage (TWS) over a time period of 10 years (01/2003–12/2012) and to investigate the nature of, and the factors controlling , these variations over Africa. Spatial and temporal correlations of the trends extracted from processed (smoothed [Gaussian: 200km] and destriped) GRACE-derived (TWS_GRACE) and LSM-derived (TWS_LSM) TWS indicate the following: (1) Large (≥ 90 % by area) sectors of Africa are undergoing statistically significant TWS_GRACE and TWS_LSM variations due to natural and anthropogenic causes; (2) a general correspondence between TWS_GRACE and TWS_LSM over areas (e.g., Niger and Mozambique NE basins in eastern and western Africa) largely controlled by natural (i.e., increase/decrease in precipitation and/or temperature) causes; (3) discrepancies are observed over areas that witnessed extensive anthropogenic effects measured by TWS_GRACE but unaccounted for by TWS_LSM. Examples include: (a) strong (compared to that observed by TWS_LSM) negative TWS_GRACE trends were observed over areas that witnessed heavy groundwater extraction (e.g., Western, Desert, Egypt); (b) strong (compared to that observed by TWS_LSM) positive TWS_GRACE over Lake Volta reservoir; and (c) strong (compared to that observed by TWS_LSM) negative trends over areas undergoing heavy deforestation (e.g., northern and NW Congo Basin); (4) additional discrepancies in other areas (e.g., Zambezi and the Okavango basins) are attributed to models being uncalibrated and not not simulating all of the TWS components (e.g., river storage and groundwater in GLDAS; lakes and reservoirs in GLDAS and CLM4.5) . Future work should focus on using TWS_GRACE to calibrate TWS_LSM.