Development of Hybrid Statistical-dynamical Model for Air Quality Prediction: Establishment and Application

Abstract

This study focuses on the development and evaluation of a new statistical model designed for forecasting next-day air quality in Hong Kong, which is a highly urbanized area with complex terrain. A new hybrid statistical-dynamical model (HSDM) contains two components: 1) Regional chemistry model (i.e., CMAQ) coupled with a regional climate model (i.e., WRF), which captures the most important physical and chemistry processes within the dynamical models, and 2) Statistical generalized additive models (GAMs) developed separately under different meteorological conditions (e.g., synoptic flow, and precipitation) from WRF and CMAQ model results. The HSDM is capable of capturing three important features of local air quality in Hong Kong. Firstly, the influence of background source contribution from the surrounding area is being captured. Secondly, the effect of pollutant emission source is taken into account for the emission strength over Pearl River Delta (PRD) area. Finally, it allows selection of the most appropriate predicted equation parameterized using local meteorological variables (e.g., surface temperature and relative humidity).
For short-term air quality forecast (i.e., 2007 and 2014), the detailed evaluation of the GAM prediction against raw CMAQ simulation showed the HSDM performed reasonably well. However, due to the discrepancy from the prediction of CMAQ results, the actual air quality prediction was inadequate. Bias-adjustment techniques (e.g., Hybrid Forecast (HF) and Kalman Filter (KF)) were applied to improve forecast accuracy, which further improve the air quality prediction to a satisfactory level.
Selection of a statistical approach is governed by its ability to provide the uncertainties for projections under different future climate scenarios. In this study, we explored the usage of HSDM on future air quality projection under IPCC AR5 RCP8.5 using the climate data from the Community Earth System Model (CESM).

Date of Award	8 Sept 2017
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Yun Fat Nicky LAM (Supervisor), Chung Leung Johnny CHAN (Supervisor) & Chi Yung Francis TAM (Supervisor)