Geoscience seminar speaker, Dr. Sankar Arumugam, NC State UniversityThursday, March 16, 2017 -
Host: Venkat Lakshmi
Title: “Uncertainty Reduction in Climate Forecasts Using Multimodel Combination and their Relevance to Water Management”
Recent studies on multimodel combination of seasonal climate forecasts clearly show that multimodel ensembles produce better probabilistic forecasts resulting with reduced false alarms and missed targets, which could help in obtaining better estimates of reliability for the specified release from the reservoir. The talk will present a multi-combination strategy that combines climate forecasts from multiple models conditioned on the ENSO state. The performance of the strategy with existing optimal combination strategies will also be presented. The talk will also apply the forecasts from the multimodel combination and will evaluate their utility in improving allocation from two reservoirs – Falls Lake in NC and Masinga Lake in Kenya – that serve water for municipal and instream needs and for hydropower generation. As an ongoing collaboration with the State Climate Office of NC, our research group has developed the monthly updated inflow forecasting portal for various virgin basins over the Southeast US, Monthly-to-seasonal streamflow forecasts are developed and updated automatically by forcing climate forecasts with Noah Land surface model which is initialized with NLDAS2 products for specifying the initial states. The utility of this forecasting portal and the challenges in implementing it for system operation will also be discussed.