Nonstationary Stochastic Tropical Cyclone Model for Regional Wind Hazard Assessment under Global Warming

Project: Research

View graph of relations

Researcher(s)

Description

Tropical cyclones (TCs) are the major natural disaster affecting the southeast coastal region of China including Hong Kong and Guangdong. TCs cause heavy causalities, extensive damage to infrastructure facilities and buildings, and set back regional economic development. Due to global warming, the frequency and intensity of future TC hazards are nonstationary, leading to the possibility of even worse TC-induced losses in the future. To establish a TC-resilient Hong Kong, a fundamental task is to develop TC hazard models that can predict future nonstationary TC hazards for large urban areas under global warming. These hazard models will benefit the decision-making concerning infrastructure loss and mitigation strategies of a wide range of organizations in the public and private sectors, such as governmental departments for emergency management and insurance companies, and infrastructure design practices. To this end, full-track TC models can generate the life cycles of synthetic TCs within a whole ocean basin for thousands of years. Combined with TC wind field models, they have been used for extreme wind hazard assessment over basin-wide coastal areas. However, previous full-track models are typically statistics-based, do not sufficiently incorporate environmental variables such as humidity and vorticity, and lack consideration of the essential physics of TC activities. Thus, their capacity to predict future nonstationary TC hazards is limited. In addition, although previous wind field models have advanced from 2D to 3D to simulate the spatial structure of TCs more realistically, they are not validated using in-situ measurements above the surface layer and may not satisfactorily replicate wind fields over complex terrains. Given the above research issues, this project aims to develop a new physics-guided nonstationary TC hazard model and assess future nonstationary TC wind hazards in Hong Kong and Guangdong under global warming. Specifically, the project includes four coherent tasks. (1) An environment-dependent full-track model, combining advanced statistical models and essential physics, will be developed to simulate synthetic TC events over the western North Pacific (WNP); the model will consist of a Poisson regression-based genesis model, a stochastic movement model driven by environmental wind fields, and an intensity model for TCs with unsaturated cores. (2) A 3D nonlinear TC wind field model with deep-learning enhancement will be developed to simulate TC wind fields over complex terrains. To this end, the multiplatform climate monitoring network of Hong Kong Observatory (HKO), including 8 weather buoys, over 50 meteorological stations, radar wind profilers, radiosonde balloons, Global Positioning System dropsondes, etc., provide measurements over the whole depth of the TC boundary layer, which will be used for model training, validation, and uncertainty analysis. Then, the statistical models of wind field model parameters for Hong Kong and Guangdong will be developed. (3) The synthetic TC model, combining the full-track model and the wind field model, will be validated using TC best track data and climate monitoring data from HKO. Then, three wind hazard models will be developed for each site of interest, including the probability of annual maximum wind speeds at the site, the probability of annual maximum wind speeds within a circular region centered at the site, and the joint probability of annual maximum wind speeds at the site and corresponding radius of maximum winds. The second model captures the hazard intensity within a region and the third model uses both the intensity and size of a cyclone event to indicate the event's destructiveness. Thus, both the second and third models are suitable for regional loss assessment. (4) The predictions of future large-scale environmental parameters from global climate models (GCMs) will be combined with the synthetic TC model to assess future nonstationary TC wind hazards under global warming. Firstly, the latest GCMs will be selected to construct a multi-model ensemble to predict future atmospheric and oceanic variables until 2100 under various scenarios of future greenhouse gas emission. Then, the predicted environmental variables will be fed into the synthetic TC model to simulate TCs during near-term, mid-term and long-term future 

Detail(s)

Project number9043326
Grant typeGRF
StatusNot started
Effective start/end date1/01/23 → …