Advanced Extreme Rainfall Simulation Using a Numerical Weather Prediction Model

Abstract

Extreme rainfall simulations using numerical weather models provide high spatial and temporal resolution data on rainfall. These simulations are useful tools in hydrometeorology that help with risk management for rainfall-derived disasters. Yet we still lack a thorough understanding of how to generate accurate extreme rainfall simulations. This thesis is centred around three main impact factors of extreme rainfall simulations: (1) model spatial and physical configurations, (2) model initialization and spin-up time, and (3) aerosol inputs by coupled module. First, we develop a progressive multimeric configuration optimisation (PMCO) method to comprehensively evaluate impacts from spatial and physical configurations and systematically screen the optimal configuration set for the study area. Second, we present a novel optimal spin-up time identifying (OSTI) framework to help determine the appropriate spin-up time for different rainfall events without too many trial-and-error tests. It is also proved that the model-required initialization time depends on the initial and boundary conditions, which is not fixed and vary with different events and situations. Third, we show the impacts of aerosol properties, transport and seasonality on simulations of different rainfall events. We also performed an attribution analysis of changes in simulation accuracy and summarized under what circumstances the coupled module enhances or weakens simulations. Finally, we discuss what is the most influential factor and how to make advanced extreme rainfall simulations based on the above knowledge, followed by future research that needs to be addressed. Overall, this thesis advances our knowledge in relation to all three above-mentioned aspects and thereby helps us conduct more accurate extreme rainfall simulations.

Date of Award	18 Jun 2024
Original language	English
Awarding Institution	University of Bristol
Supervisor	Dawei Han (Supervisor) & Miguel A Rico-Ramirez (Supervisor)