An air quality information system for cities with complex terrain based on high resolution NWP

1
An air quality information system for cities with
complex terrain based on high resolution NWP

• Viel Ødegaard, rd department

2
An air quality information system

• Daily forecasts for NO2 and dust (pm10/pm2.5)
  concentrations are produced for the five
  Norwegian cities Oslo, Drammen, Trondheim, Bergen
  and Stavanger during winter
• A modelling system is developed by Norwegian
  Meteorological Institute and Norwegian Institute
  for Air Research in cooperation
• The Norwegian Traffic Authorities fund the system
  and local health authorities in the cities are
  the end users
• Meteorological and air quality forecasts for the
  next day and air quality statistics are available
  for the end users on internet.

3
Components

• NWP model
• interface module
• air quality model
• subjective interpretation
• actions

4
Heavy air pollution episodes in northern areas
occur under the following meteorological
conditions

• Local circulation
• high vertical stability
• low wind speed
• negative radiation balance
• low surface temperature
• dry surface
• Large scale conditions
• persistent high pressure situation
• (followed by a change in weather regime)

5
Winter time inversion

• persistent low temperatures over several days
  lead to extensive wood burning and emission of
  particles (pm2.5) in addition to the traffic
  emmisions (NO2)
• in high pressure situations in winter time the
  radiation term dominates the thermodynamics. The
  amount of incoming solar radiation is very small
  compared to the outgoing long wave radiation as
  the solar angle is low and the days are short
• persistent radiative cooling of the surface
  builds up a surface inversion
• a change in the large scale circulation, with
  warmer air carried to the coast and over the
  mountains can strengthen the inversions. The cold
  air is captured between the topography and the
  warm air aloft

6
(No Transcript)
7
(No Transcript)
8
Spring dust

• road surface is torn off during the winter, but
  most of the matter is kept on the road surface in
  the snow. When the snow melts and the road
  surface becomes dry the particles stired up and
  dispersed by the traffic and the wind
• common in the spring, after several days with
  good melting conditions and no precipitaiton
• local concentrations of dust vary a lot. The wind
  forecasts (both speed and direction) and the
  description of snow cover (and snow melting) from
  the meteorological model is crucial

9
NWP model

• resolution required for Norwegian cities implies
  that a non-hydrostatic model is used
• MM5 v3 (Grell et al. 1994), two-way nesting, 3
  and 1 km resolution
• sigma vertical coordinate (17 layers 9 below
  1500m)
• physics
• non local plantetary boundary layer scheme (Hong
  and Pan, 1996)
• simple ice phase condensation scheme (Grell et
  al. 1994) no convection
• radiation scheme interacting with clouds (Grell
  et al. 1994)
• 5 layer soil model (Dudiah, 1996)
• run on local IBM pc-cluster 40 processors, 48
  hour forecast for the Oslo area takes 28 min.

10
domains and resolution
11
Input

• analysis interpolated from HIRLAM 0.1 deg
  resolution
• boundaries from HIRLAM 0.1 deg on 15 pressure
  levels
• snow cover, sea surface temperature, soil
  temperature and soil moisture in 7, 50 and 95 cm
  from HIRLAM
• MM5 topography and land use data base, 0.9 km
  resolution, 16 surface classes
• 24 48h forecast

12
Evaluation of forecasts

• yearly verification of last winters forecasts
• the observation network is extended with
  instruments supplied by the traffic authorities
  and the local governments
• comparing air quality forecasts with the driving
  parameters from NWP (temperature, wind and
  vertical temperature gradient) in order to adress
  the sources of the errors.
• observations in more than one level only in
  Oslo, where one station is measuring in two
  levels ( 7 and 25 m).

13
(No Transcript)
14
(No Transcript)
15
Main results from validation of the meteorology

• MM5 forecasts have higher stde than forecasts
  from HIRLAM
• Local wind directions are better represented in
  MM5
• MM5 developes too strong surface inversions
• Wind speed is often overestimated in calm
  situations and sometimes underestimated when
  there is strong winds outside the coast (increase
  model domain)
• 2m temperature has sometimes large errors. Key
  words are snow cover and boundary layer
  parameterisation in the stable case

16
(No Transcript)
17
Snow cover observations
18
Future

• develope initial fields with snow cover based on
  satellite and surface observations
• In FUMAPEX testing higher vertical resolution
  and boundary layer parameterisations (inversion
  problem)