Author : Sanjib Sharma
Publisher :
ISBN 13 :
Total Pages : pages
Book Rating : 4.:/5 (111 download)
Book Synopsis ADVANCING A REGIONAL HYDROLOGIC ENSEMBLE PREDICTION SYSTEM. by : Sanjib Sharma
Download or read book ADVANCING A REGIONAL HYDROLOGIC ENSEMBLE PREDICTION SYSTEM. written by Sanjib Sharma and published by . This book was released on 2019 with total page pages. Available in PDF, EPUB and Kindle. Book excerpt: There is great potential for using ensemble weather forecasts to improve hydrological predictions across spatial and temporal scales. To realize this potential, research is needed to formulate, assemble, and assess a full (i.e. accounting for the complete hydrometeorological forecasting chain) ensemble hydrological system and test how different system components can best contribute to improving hydrological predictions. The primary goal of this Ph.D. research is to fundamentally advance a reliable and robust Regional Hydrologic Ensemble Prediction System (RHEPS) by integrating new system components and implementing novel statistical techniques within a verifiable scientific and experimental setting. The proposed forecasting framework should facilitate understanding and quantifying the uncertainty and implications of ensemble weather forecasts on regional hydrological predictions. To meet my research goal, the following four distinct research objectives are carried out: Objective 1 (O1) - to perform a comprehensive verification analysis of ensemble precipitation forecasts from three different weather forecasting systems or guidance across the eastern U.S., including the National Oceanic and Atmospheric Administrations (NOAAs) National Centers for Environmental Prediction (NCEP) Global Ensemble Forecast System Reforecast version 2 (GEFSRv2), Short Range Ensemble Forecast (SREF) and the NCEPs Weather Prediction Center probabilistic quantitative precipitation forecasts (WPC-PQPFs). Objective 2 (O2) - to investigate the interactions between a weather preprocessor and a hydrologic postprocessor in ensemble streamflow forecasting. The terms preprocessing and postprocessing indicate the implementation of advanced statistical models and tools that rely on reforecast or hindcast information in order to improve forecast skill and reliability prior to issuing the actual forecast. Objective 3 (O3) - to determine whether the skill of hydrological multimodel forecasts is significantly larger than that of a single model, and whether the observed skill improvement is dominated by model diversity or reduction of noise associated with the ensemble size. Objective 4 (O4) - to evaluate the ability of NOAAs National Centers for Environmental Prediction (NCEP) Climate Forecast System version 2 (CFSv2) to result in skillful streamflow and water quality predictions. With O1, the weather forecasting system to use with O2 and O3 is selected. The verification result for O1 demonstrate that WPC-PQPFs tend to be superior, in terms of the forecast skill and reliability, to both the GEFSRv2 and SREF across the eastern U.S. However, GEFSRv2 is used to complete O2 and O3 since these data are available for a longer time period and expand longer lead times. With O2, two different components of the RHEPS, i.e., the weather preprocessor and hydrological postprocessor are tested. This objective is significant because it will clarify the conditions and the degree to which the combined implementation of preprocessing and postprocessing can contribute to enhance hydrological predictions. With O2, it is concluded that implementing both preprocessing and postprocessing ensures the most skill improvements, but postprocessing alone can often be a competitive alternative. With O3, it is examined if the skill improvement of hydrological multimodel forecasts is dominated by the reduction of noise associated with ensemble size, or by model diversity (i.e., additional information provided by the different models). The analysis indicates that any skill improvement of multimodel forecasts are largely dominated by model diversity and that increasing the ensemble size has only a small influence.Finally, with O4, a new dynamical-statistical approach is built and implemented to generate S2S water quantity (streamflow) and quality (nutrients and suspended sediments) predictions. This hybrid approach is more cost effective and computationally efficient than implementing a process-based water quality model, which makes it readily implementable in an operational forecasting setting. With O4, it is concluded that the dynamical CFSv2 forecasts, when combined with quantile regression, can generate skillful streamflow, nutrient load, and suspended sediment load forecasts at lead times of 1 to 3 months. Overall, the findings from this research demonstrate several strategies for enhancing hydrological forecasting and, ultimately, providing information that could be used to issue better forecasting products to the public.
