[A-HW21] Surface and subsurface hydrologic models: Technical advances and applications for water management

[AHW21-P10] Quantifying the sensitivity and stability of a multi-purpose reservoir's performance to model, inflow, and operational uncertainties

*Manvitha Molakala¹, Riddhi Singh^1,2 (1.Department of Civil Engineering, Indian Institute of Technology Bombay, Powai, Mumbai, India, 2.Interdisciplinary programme in Climate Studies, Indian Institute of Technology Bombay, Powai, Mumbai, India)

Keywords:reservoir operation, sensitivity analysis , Nagarjuna Sagar, water resources management, stability of performance measures

Performance metrics within a simulation-optimization framework are frequently used to identify and evaluate water resource management strategies. These metrics are likely to be sensitive to varying levels of input variables such as inflows, model-related choices, and strategy implementation errors. The ability to quantify the sensitivity of performance metrics to the aforementioned uncertain factors may thus be useful for decision-makers in understanding the relative importance of these factors and their interactions. Furthermore, the total variation in performance measures caused by uncertain factors may be useful in quantifying the stability of the performance measures. Using Sobol's variance-based sensitivity analysis, we quantify the performance metrics' first, second, and total order sensitivity to uncertain factors. The coefficient of variation of the metric evaluated for varying input factors is used to quantify the stability of the performance metric. The sensitivity and stability indices are then applied to four performance metrics of the multi-purpose Nagarjuna Sagar reservoir, the largest reservoir in southern India's Krishna river basin. The reservoir, which has an installed capacity of 810MW, supplies water for irrigation, industrial, and domestic use while also generating hydropower. This study's performance metrics are: (i) maximize hydropower generation, (ii) maximize the reliability of maintaining minimum environmental flows, (iii) maximize the reliability of avoiding high flow exceedance, and (iv) minimize demand deficits. We use evolutionary multi-objective direct policy search (EMODPS), which employs a state-aware operating rule based on radial basis functions, to identify the Pareto approximate set of reservoir operation strategies. In our analysis, we take the following uncertain factors into account: (i) the length of the planning horizon (ranging from 1 to 15 years), (ii) the model timestep (daily, 15-day, and monthly timesteps), (iii) imperfect operations while employing optimized strategies, and (iv) stochastic and deep uncertainties associated with inflows. Our findings show that the hydropower generation objective is the most sensitive to model choices. High flow non-exceedance reliability, demand deficits, and minimum environmental flow reliability objectives, on the other hand, are particularly sensitive to deep uncertainties in inflows. We find that the interaction effects of these factors have no effect on hydropower generation, environmental flow reliability, or demand deficits. High flow non-exceedance-related objectives, on the other hand, are sensitive to the interactions between deep uncertainties and model uncertainty. We also discover that the first-order sensitivity indices have a higher level of confidence than the total-order sensitivity indices. When subjected to uncertainty, we find that the flood reliability objective is the most stable and the demand deficits objective is the least stable. In any water management problem, our framework can be used to determine the relative importance of the uncertain factors and the stability of the performance measures.