Characterizing Influence of Hydrologic Data Correlations on Climate Change Decision Variables: Evidence from Diyala River Basin in Iraq

Abstract

Water resources decision-making processes involving impacts of climate change require coherent climate datasets, which may exhibit high spatial and temporal variability. These datasets are used in models with forcing data provided by statistical weather generators that mimic observed system behavior. The datasets must conserve historical correlations, or they will lead to wrong decisions about future climate change influences. The objectives of this study are to (1) evaluate the impact of the cross, spatial, and temporal correlations in climatic datasets on the climate change decision variables and (2) examine the contributions of variability in each subcorrelation on system performance outcomes. A predeveloped nonstationary bottom-up approach was used to assess the operational rules of a multipurpose reservoir on the Diyala River basin in Iraq. The study utilizes a statistical weather generator to synthesize 10 trajectories, of 405 different climate scenarios, by varying the preserved accuracy (100%, 66%, 33%, and ≈ 0%) of cross, spatial, and temporal correlations. The results indicated that the system performance is influenced significantly when correlations were varied, with the most sensitivity to spatial correlations, followed by the cross and temporal correlations. Ignoring the spatial correlation caused a 92.2% error in system performance indicators, and cross and temporal correlations caused errors of 17.9% and 9.3%, respectively. The results also revealed that the precipitation spatial correlation was the most sensitive component of the subcorrelation effects with a 68.6% error, but the cross correlation between precipitation and wind speed only accounted for a 2.5% error. The study demonstrated that the nature of basin datasets is of paramount importance in hydrologic modeling and climate change impact assessment.