Explorations of the differences between Gravity Recovery and Climate Experiment (GRACE) solutions in local regions and basins are fundamental in determining their suitability and applicability in these environments. Because of the different mathematical inversions used by the respective processing centers, individual solutions exhibit discrepancies in terms of mass increase or loss, which makes it difficult for users to select the best model for studying terrestrial water storage anomalies (TWSAs). This study compares TWSA trends, as derived from different GRACE solutions over the arid and semi-arid Botswana (2002-2019), where both storage and flux from CSR, JPL, GFZ, TUGRAZ, AIUB, and COST-G[1] were compared. The results show that the six solutions are fairly correlated with the least correlation of R=0.829 between JPL and AIUB, and a maximum of R=0.921 between CSR and TUGRAZ at a 95% confidence level. The TWSA analyses for 2002-2019 indicate that TWS is increasing in Botswana, with the least linear trend of +0.11cm/yr detected from the TUGRAZ inversion model, and the highest linear trend at +0.43cm/year from the COST-G model. On comparing TWS with rainfall, all the solutions presented the same spatio-temporal trends as the rainfall patterns, indicating that the GRACE solutions exhibit the same responses with respect to the received rainfall.  Over the 18 years investigated, the long-term rainfall trend was found to decrease, which was only detected by the TUGRAZ model in terms of the recorded equivalent water height (EWH) of -0.008cm/yr from the monthly trend observations. Overall, the AIUB inversion solution gave a better result as its signal was found to be the same as the rainfall signal.   [1] CSR = Center for Space Research;  JPL = Jet Propulsion Laboratory; GFZ = the German Research Center for Geosciences; TUGRAZ = Graz University of Techology; AIUB = the Astronomical Institute of the University of Bern; COST-G = the International Combination Service for Time-Variable Gravity Fields