Institute: New Jersey
Year Established: 2020 Start Date: 2020-02-25 End Date: 2021-02-23
Total Federal Funds: $20,000 Total Non-Federal Funds: $40,000
Principal Investigators: Ruo-Qian Wang
Abstract: The sensor network of hydrological monitoring is facing the reliability and security challenges stemming from extreme weather, sensor glitches, telecommunication breakdown, vandalism, cyberattacks, and budget deficits. Hydrological forecasting, flood early warning, and other water resources forecasting systems rely on these important sensor networks to calibrate, validate, and correct the model state for accurate and sustainable prediction. The present project is targeted to 1) develop a numerical model to reproduce the sensor network failures in the numerical forecasting tasks, 2) create a framework to assess the forecasting vulnerability and sensitivity to the sensor node disconnection, and 3) explore strategies to mitigate the impacts to strengthen the forecasting resilience. An Ensemble Kalman Filter-based data-assimilation framework will be built combining the operational flood early warning model and the on-site stream gauge data streams. A Monte Carlo Simulation scheme will be implemented to reproduce the sensor network disruptions and examine the model forecasting resilience. Global sensitivity analysis and complex network-based theory will be applied to identify and rank the protection priority of the sensor network nodes. The feasibility of an innovative data imputation method will be explored for temporary data stream recovery in emergency sensor network disruptions. We anticipate the project will build a platform to test data-assimilation schemes with a current operational model and real data streams, a new framework to systematically assess the forecasting system resilience to sensor network service, and a new strategy about using data imputation method to mitigate sensor network disruption.