The Use of High Resolution Mesoscale Model Fields with the CALPUFF Dispersion Modelling System in Prince George BC

1
The Use of High Resolution Mesoscale Model Fields
with the CALPUFF Dispersion Modelling System in
Prince George BC
Bryan McEwen Masters project 2000-2002 Atmospher
ic Science Group Dr. Peter L. Jackson University
of Northern British Columbia Funding supplied
by the Oil and Gas Commissions Environmental
Fund Study conducted with the High
Performance Computing Facility at UNBC with an
SGI Origin 3400.
2
CALPUFF

• A Dispersion modelling system composed of 3
  separate modules
• CALMET a meteorological model, constructs
  3-dimensional fields of wind, temperature and
  other variables from local observations.
  Deterministic in nature. Requires upper air data
  as well as data from at least one surface
  station.
• CALPUFF a Gaussian Puff dispersion model,
  predicts 3-dimensional fields of pollutant
  concentrations. Requires CALMET fields and an
  inventory of pollutant sources and emission
  rates.
• CALPOST a post processor that extracts
  pollutant concentrations from CALPUFF fields.

3
RAMS

• A prognostic (forecasting) mesoscale model, able
  to simulate circulations from hemispheric in
  scale down to 100s of metres.
• Smaller grids are nested within larger parent
  grids when higher resolution is required.
• Calculates meteorological variables each time
  step. Smaller grid sizing requires a smaller
  time step and therefore more calculations.
• Significant computer resources needed. SGI
  Origin 3400 Fraser used at the HPC facility at
  UNBC.

4
A HINDCAST instead of a FORECAST

• For regulatory modelling, a meteorological period
  in the past is used for dispersion analysis.
• RAMS is therefore used to make a hindcast of
  what the atmosphere was like during the period.
• Nearby observations are used as boundary
  conditions to guide RAMS simulations.
  Predictions are periodically corrected on the
  large scale.
• Numerical models such as RAMS are able to
  telescope down from the large scale motions to
  predict circulations on a local scale. The
  influence of topography, vegetation, bodies of
  water, radiation etc. are accounted for by the
  model.

5
Research Objectives

• (Can simulated meteorology (RAMS fields) be used
  in lieu of meteorological observations to perform
  dispersion modelling in an area of complex
  terrain??)
• To assess the use of RAMS fields in producing
  high quality CALMET meteorological fields.
• To assess the use of RAMS fields in producing
  high quality CALPUFF pollutant concentration
  estimates.

6
Methodology

• SO2 chosen to model, within a 24.5km by
• 24.5 km domain surrounding Prince
• George.
• Three 5-day periods in 1999 of moderate
• to high levels of SO2 chosen to analyze.
• 6 surface meteorological stations available.
• - 1 upper air meteorological station available.
• - 4 SO2 monitoring stations available.

Surface meteorological station
Upper air meteorological station
Significant SO2 source
SO2 monitoring station
7
Methodology contd CALPUFF FLOWCHART
8
Methodology contd MODEL CONFIGURATION
9
RAMS Initialization
The National Centers for Environmental
Prediction (NCEP) coarse gridded datasets were
used to initialize and nudge RAMS forecasts. Thes
e gridded fields at 2.5o lat. by 2.5o long, with
17 pressure levels in the vertical
10
Analysis of CALMET Meteorological Modelling
11
Analysis of Results WIND FIELDS

• - Wind data from 5 of the 6
• surface stations used to test
• each models wind fields
• (heavy arrows indicate station
• locations)
• CALMET-3 used two of these
• 5 stations in its analysis, hence
• testing for this model is
• favourably biased.
• CALMET-6 used all surface
• stations in its analysis, so no
• test of its wind fields was
• possible.

12
Analysis of Results WIND FIELDS
Average Root Mean Square Vector Error (RMSVE),
modelled surface winds compared to observed
winds at 5 station locations.
 
13
Analysis of Results WIND FIELDS
14
Analysis of Results WIND FIELDSWindrose
diagram showing RAMS-CALMET modelled wind
direction at Plaza during January Period.
15
Analysis of Results WIND FIELDSWindrose
diagram showing CALMET-3 modelled wind direction
at Plaza during January period.
16
Analysis of Results BOUNDARY LAYER
Inversions conditions, developing overnight,
exacerbate poor air quality by limiting vertical
mixing During several evenings of the
three periods examined, strong inversion condition
s existed. RAMS consistently did not develop
accurate temperature profiles within the
boundary layer. (extreme case shown)
17
Analysis of CALPUFF Dispersion Modelling
18
Analysis of Results SO2 CONCENTRATIONS
CALMET-1 CALMET-3 CALMET-6 RAMS-CALMET
JANUARY 99 98 61 67
APRIL 97 97 71 68
JUNE 79 97 83 84
Mean Relative Error (MRE) (), modelled 24-hr SO2
concentrations from observed concentrations at 4
monitoring sites.
19
Conclusions

• There is evidence that RAMS accounted for much of
  the influence of topographical heights on the
  regional winds.
• The use of RAMS fields with CALMET, although
  possessing errors, produced meteorological fields
  as good or better than CALMET using 1 or 3
  surface stations.
• CALPUFF dispersion estimates of SO2 made with
  RAMS input were more representative of observed
  concentrations than those made with 1 or 3
  station CALMET fields.