Integrated water resources management: development of data parsimonious models for reservoir planning

摘要

Data parsimonious models are developed which offer the realistic possibility for reservoir storage-yield-reliability (S-Y-R) planning analysis at ungauged sites. One of the models was based on multiple linear regression analysis and the second set of models were designed using artificial neural networks (ANNs); both models use input variables comprising runoff summary statistics, e.g. the CV of annual runoff, and reservoir systems variables such as the demand and reliability index. The models performed satisfactorily when tested using independent data sets not utilised during their development. Finally, algorithms for obtaining the runoff summary statistics at ungauged sites are presented.