Relating the dynamics of climatological and hydrological droughts in semiarid Botswana

Dynamics of droughts have been an associated feature of climate variability particularly in semiarid regions which impact on the response of hydrological systems. This study attempts to determine drought timescale that is suitable for monitoring the effects of drought on hydrological systems which can then be used to assess the long term persistence or reversion and forecasts of the dynamics. Based on this, climatological and hydrological drought indices characterized by Standardized precipitation evapotranspiration index (SPEI) and Standardized flow index (SFI) respectively have been determined using monthly rainfall, temperature and flow data from two major river systems. The association between climatological and hydrological droughts in Botswana has been investigated using these river systems namely: Okavango that is predominantly a storage type and Limpopo which is non-storage for a period of 1975-2014. Dynamics of climatological and hydrological droughts are showing trends towards drying conditions at both river systems. It was also observed that hydrological droughts lag climatological droughts by 7 months in Limpopo and 6 months in Okavango river systems respectively. Analyses of the association between climatic and flow indices indicate that the degree of association becomes stronger with increasing timescale at the Okavango river system. However in the Limpopo river system, it was observed that high timescales of 18- and 24-months were not useful in drought monitoring. 15-months timescale was identified to best monitor drought dynamics at both locations. Therefore SPEIs and SFIs computed at 15-months timescale have been used to assess the variability and long term persistence in drought dynamics through rescaled range analysis (R/S). H-coefficients of 0.06 and 0.08 resulted for Limpopo and Okavango respectively. These H-coefficients being significantly less than 0.5 is an indication of high variability and suggests a change in dynamics from the existing conditions in these river systems. To forecast possible changes, the nonlinear autoregressive with exogenous input (NARX) artificial neural network model has been used. Results from this model agree with those of the R/S and projects generally dry conditions for the next 40 months. Results from this study are helpful not only in choosing a proper timescale but also in evaluating the futuristic drought dynamics necessary for water resources planning and management.