Using HYSPLIT to Understand Source-Receptor Relationships: Some Examples

1
Using HYSPLIT to Understand Source-Receptor
Relationships Some Examples
Mark Cohen NOAA Air Resources Laboratory 1315
East West Highway, R/ARL, Room 3316 Silver
Spring, Maryland, 20910, USA mark.cohen_at_noaa.gov h
ttp//www.arl.noaa.gov/ss/transport/cohen.html
HYSPLIT Training NOAA Air Resources
Laboratory April 29 May 1, 2008 Silver Spring,
MD, USA
2

• We use HYSPLIT in many ways at the Air Resources
  Laboratory to investigate atmospheric
  source-receptor relationships.
• A few examples will be shown in this
  presentation
• Evaluation of potential monitoring site
  locations
• Back-trajectories
• Concentration (forward dispersion)
• Back-trajectory analysis of measurement
  episodes
• Atmospheric mercury measurements
• Criteria pollutant episodes

3

• We use HYSPLIT in many ways at the Air Resources
  Laboratory to investigate atmospheric
  source-receptor relationships.
• A few examples will be shown in this
  presentation
• Evaluation of potential monitoring site
  locations
• Back-trajectories
• Concentration (forward dispersion)
• Back-trajectory analysis of measurement
  episodes
• Atmospheric mercury measurements
• Criteria pollutant episodes

4

• If we put our site at Beltsville, would air
  parcels arriving at the site come from known
  source locations?
• That is, would we expect to get any plume impacts
  at the site?
• We want at least some of these plume impacts
• We ran back-trajectories from the potential
  location for every hour during a 4 week period
  during the summer of 2005
• 28 days x 24 hrs/day 672 trajectories

Potential atmospheric mercury monitoring site in
Beltsville, MD
This particular trajectory did not represent a
plume hit
yellow circles are large mercury point sources
within 100 km of potential site
5
Beltsville trajectory movie
6

• We use HYSPLIT in many ways at the Air Resources
  Laboratory to investigate atmospheric
  source-receptor relationships.
• A few examples will be shown in this
  presentation
• Evaluation of potential monitoring site
  locations
• Back-trajectories
• Concentration (forward dispersion)
• Back-trajectory analysis of measurement
  episodes
• Atmospheric mercury measurements
• Criteria pollutant episodes

7
Dispersion example for emissions from Morgantown
coal-fired power plant Will Plume Impact
Potential Beltsville Monitoring Site?

• April 1July 21, 2004 (16 weeks)
• Simplified wet/dry deposition
• no plume chemistry
• EDAS-40 km met data (3-hr res)
• results displayed are the concentration averaged
  between 0-10 meters above the ground for a 1
  gram/hr continuous emissions rate
• Each frame shows avg over 3 hours
• Morgantown RGM 120 kg/yr
• This is equivalent to 14 g/hr
• Example -- 10-11 g/m3 (dark blue) is in the range
  10-11 to 10-9 140 14,000 pg/m3
• Example -- 10-13 g/m3 (green) is in the range
  10-13 to 10-11 1.4 - 140 pg/m3

8
power plant emissions dispersion movie
9

• We use HYSPLIT in many ways at the Air Resources
  Laboratory to investigate atmospheric
  source-receptor relationships.
• A few examples will be shown in this
  presentation
• Evaluation of potential monitoring site
  locations
• Back-trajectories
• Concentration (forward dispersion)
• Back-trajectory analysis of measurement
  episodes
• Atmospheric mercury measurements
• Criteria pollutant episodes

10
Beltsville Episode January 7, 2007
11

• Back trajectories simulated with the NOAA HYSPLIT
  model using meteorological on a 12 km grid
• Trajectories started from the Beltsville site
  (marked with a star on the map) at 5 different
  heights representing different positions in the
  mixed layer (0.1, 0.3, 0.5, 0.7, 0.9)
• Note that the times are represented in UTC
  (Universal Time Coordinate) which is 5 hours
  later than Eastern Standard Time, e.g., 17 UTC
  1200 Eastern Standard Time
• Large mercury emissions sources are indicated on
  the map, including
• Dick Dickerson facilities (coal-fired power
  plant and municipal waste incinerator)
• Harf Harford County municipal waste incinerator
• PhRe Phoenix Services medical waste and
  Baltimore Resco municipal waste incinerators
• BS_W Brandon Shores and H.A. Wagner coal fired
  power plants
• CkPt Chalk Point coal fired power plant
• Mrgt Morgantown coal fired power plant
• PsPt Possum Point coal fired power plant
• Ptmc Potamac River coal fired power plant
• Arlg Arlington/Pentagon waste incinerator
• Brun Brunner Island coal fired power plant

100 km
12
display HYSPLIT-generated trajectory shapefiles
in GIS program (ArcView) along with emissions
13
(No Transcript)
14
(No Transcript)
15
(No Transcript)
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23

• We use HYSPLIT in many ways at the Air Resources
  Laboratory to investigate atmospheric
  source-receptor relationships.
• A few examples will be shown in this
  presentation
• Evaluation of potential monitoring site
  locations
• Back-trajectories
• Concentration (forward dispersion)
• Back-trajectory analysis of measurement
  episodes
• Atmospheric mercury measurements
• Criteria pollutant episodes

24
151 monitoring sites at which PM and/or O3
exceedances occurred during 2002-2006
25
Methodology

• 1112 episodes identified 832 PM 280 O3
• 296 in 2002
• 303 in 2003
• 182 in 2004
• 320 in 2005
• 8 in 2006
• 151 monitoring sites across Canada
• Multiple 72-hr back-trajectories were run with
  the NOAA HYSPLIT model for each episode, starting
  at the middle of the mixed layer
• 24-hr PM episodes 7 trajectories run for each
  episode, once every 4 hours (at 0, 4, 8, 12, 16,
  20, and 24 hours after the start of the episode)
• 8-hr O3 episodes 5 trajectories run for each
  episode, once every 2 hours (at 0, 2, 4, 6, and 8
  hours after the start of the episode)
• Following the above methodology, a total of 7224
  back-trajectories were attempted

26
Example of trajectories PM episode at site
30501, Aug 14, 2002 Note comparable
trajectories were simulated for each of the 1112
PM and O3 episodes... a total of 7224
trajectories were simulated
7 trajectories simulated for each 24-hr PM
episode, at 4-hour intervals
start of episode
4th hour of episode
8th hour of episode
20th hour of episode
24th hour of episode
16th hour of episode
12th hour of episode
27
this was an overwhelming amount of information...
we wanted to try to see if there were any
patterns
28
Sites divided in several ad-hoc groups to
investigate patterns
29
Sites divided in several ad-hoc groups to
investigate patterns
30
Sites divided in several ad-hoc groups to
investigate patterns
31
Sites divided in several ad-hoc groups to
investigate patterns
Group Name Number of Sites Number of PM events Number of O3 events
1 Toronto 14 139 66
2 S.W. Ontario 13 116 64
3 St. Lawrence 19 92 14
4 Montreal 17 151 27
5 Fraser River 7 33 0
6 Edmonton 5 22 0
7 Eastern 13 19 8
8 Calgary 3 13 0
9 Vancouver 5 5 0
Total in groups Total in groups 96 590 179
Not included in a group Not included in a group 55 242 101
Total Total 151 832 280
32
Here is a picture of all the 984 trajectories
that occurred during one groups PM events (group
1 in Toronto, 139 PM events)...
Still very jumbled...
33
Summary of Clustering results for Toronto group
(group 1) PM events
As one increases the number of clusters, one
reaches a point of diminishing returns in terms
of reducing the scattering around the group of
clusters
The TSV is a measure of the degree to which the
chosen clusters fit all the data
3 clusters
5 clusters
7 clusters
34
Results for 7 clusters
35
Results for 5 clusters
36
Results for 3 clusters
37
Another way to look at the universe of
back-trajectory results is to determine the
fraction of trajectories that pass through a
given grid square (in this case a 1o x 1o grid).
Here is an example for the overall results for
984 back trajectories run for the PM episodes at
sites in the Toronto group.
38
another example of grid-frequency results
Ozone events for the Eastern group of
monitoring sites
39

• We use HYSPLIT in many ways at the Air Resources
  Laboratory to investigate atmospheric
  source-receptor relationships.
• A few examples will be shown in this
  presentation
• Evaluation of potential monitoring site
  locations
• Back-trajectories
• Concentration (forward dispersion)
• Back-trajectory analysis of measurement
  episodes
• Atmospheric mercury measurements
• Criteria pollutant episodes

Thanks!
40
(No Transcript)
41
Because the meteorological data grids used for
the back-trajectory simulations was somewhat
geographically limited, not all
back-trajectories could be simulated.
Group Name Number of Sites Number of PM events Fraction of PM-event trajectories simulated Number of O3 events Fraction of O3-event trajectories simulated
1 Toronto 14 139 100 66 100
2 S.W. Ontario 13 116 100 64 100
3 St. Lawrence 19 92 100 14 100
4 Montreal 17 151 100 27 100
5 Fraser River 7 33 30 0 -
6 Edmonton 5 22 27 0 -
7 Eastern 13 19 100 8 100
8 Calgary 3 13 97 0 -
9 Vancouver 5 5 100 0 -
Total in groups Total in groups 96 590 94 179 100
Not included in a group Not included in a group 55 242 85 101 99
Total Total 151 832 91 280 99
42
80-km NOAA EDAS gridded meteorological data used
for episodes in 2002-2003
43
40-km NOAA EDAS gridded meteorological data used
for episodes in 2004-2006
44
40-km NOAA EDAS gridded meteorological data used
for episodes in 2004-2006
80-km NOAA EDAS gridded meteorological data used
for episodes in 2002-2003
45
Methodology (... continued)

• Due to limited geographical extent of
  meteorological data used to simulate the
  trajectories, only 6700 trajectories were
  actually simulated
• An example set of 7 trajectories for one PM
  episode has been provided in this file,
• jpgs and shapefiles for each trajectory were
  prepared and have been provided
• Ad-hoc groups of monitoring sites selected in an
  attempt to find patterns in the data (see table
  and maps on following slides)
• Preliminary cluster analysis performed for each
  group of sites, for PM and O3 episodes
• an example of the clustering results is shown in
  this file, and jpgs of the clustering results are
  provided for all groups / pollutants
• Preliminary analysis of gridded trajectory
  frequency performed for each group of sites for
  PM and O3 episodes
• maps for each group / pollutant were prepared and
  are included in this file

46
grid-square percentage maps for all the other
pollutants / groups
47
Ozone events for the Toronto group of
monitoring sites
48
PM events for the S.W. Ontario group of
monitoring sites
49
Ozone events for the S.W. Ontario group of
monitoring sites
50
PM events for the Eastern group of monitoring
sites
51
Ozone events for the St. Lawrence group of
monitoring sites
52
Ozone events for the Montreal group of
monitoring sites
53
PM events for the Montreal group of monitoring
sites
54
PM events for the St. Lawrence group of
monitoring sites
55
PM events for the Vancouver group of monitoring
sites
56
PM events for the Calgary group of monitoring
sites
57
PM events for the Edmonton group of monitoring
sites
58
PM events for the Fraser group of monitoring
sites