A RealTime Air Quality Modelling System for Hong Kong and its Vicinity

1
A Real-Time Air Quality Modelling System for
Hong Kong and its Vicinity

• Christopher Fung and Linda Yu
• Air Modelling Section, EPD
• Better Air Quality Workshop 2006,
• December 2006, Yogyakarta, Indonesia

2
Issues to Address to Simulate Air Quality in HK
and Vicinity

• Different temporal and spatial scales of
  interactions (wind, emissions, chemistry)
• Irregular terrain and land use
• Land-sea interaction in synoptic features
• Photo-chemistry involving many species
• Significant trans-boundary transport
• Scarcity of some input and validation data

3
PATH (Pollutants in the Atmosphere and their
Transport over Hongkong) in historical
perspective

• Hong Kong EPD focused on local scale modeling
  in the 80s and early 90s with the use of
  Gaussian type models
• Construction of PATH, a comprehensive regional
  scale numerical air quality modeling system,
  commenced in 1996
• PATH completed in 2001, focusing on a few
  episodes
• PATH used extensively in assessment since 1998,
  e.g. the Pearl River Delta air quality study
• Real-time simulation and forecasting capability
  since 2003

4
Early Episodes Studied by PATH
Photochemical episode 1 - 11 Oct 94
Photochemical episode 2 - 20 Aug 96
5
R-T PATHs Modelling Framework
Met. AQ Observations
Met. Modelling (MM5)
Met. Analysis (ORSM)
AQ Modelling (SAQM)
Verification analysis
AQ forecast
Emission Modelling (EMS95)
Emission Datasets
6
(No Transcript)
7
Modelling Domain (1.5, 0.5km resolutions) (49X49)
URBAN-SCALE DOMAIN (0.5 KM RESOLUTION)
URBAN-SCALE DOMAIN (1.5 KM RESOLUTION)
8
Time Taken for a 72-Hour Simulation(AMD, 2.4GHz,
dual-cpu)

• 3-domains (40.5-, 13.5- 4.5-km)
• Met. Modelling 9 hours (two-way nesting)
• AQ Modelling 18 hours (one-way nesting)
• (run in parallel with the met. calculation)
• Total elapsed time 23 hours
• (including 5 hours for ORSM output)

9

Tephigram(blue lines - temperature red lines -
dew point)
10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
Jan Dec 2004
16
Jan Dec 2004
17
Results Summary

• Model meteorological outputs compare very well
  with observations
• Model reproduces the general observed AQ trends
  (seasonal, weekly and daily) but tend to
  under-estimate peak pollution values
• Results suggest that refinement in emission
  estimates in PRD (and beyond) is needed to
  improve prediction of short-term averages
  (hourly) and high pollution episodes
• Limitations in handling secondary particles an
  upgrade in the chemistry / particulate model is
  needed

18
Air Quality Science