prepared by: ilias kavouras, vic etyemezian, dave dubois, mark green, marc pitchford, jin xu
DESCRIPTION
Preliminary assessment of the principal causes of dust-resultant haze at IMPROVE sites in the Western United States. Prepared by: Ilias Kavouras, Vic Etyemezian, Dave DuBois, Mark Green, Marc Pitchford, Jin Xu Division of Atmospheric Sciences, Desert Research Institute. - PowerPoint PPT PresentationTRANSCRIPT
Preliminary assessment of the principal causes of dust-resultant
haze at IMPROVE sites in the Western United States
Prepared by: Ilias Kavouras, Vic Etyemezian, Dave DuBois, Mark Green, Marc Pitchford, Jin Xu
Division of Atmospheric Sciences, Desert Research Institute
Prepared for: Western Regional Air Partnership, Dust Emissions Joint Forum
Tempe, 11/15/2005
i. Introduction – scope of the studyii. Development of Empirical/Heuristic Approach
(EHA)• Description of tools• Integration into GIS• Interpretation of EHA outcomes
iii. Case studiesiv. Results v. Seasonal and spatial variation of dust sourcesvi. Future studies
Outline
Scope of the studyIdentify events resulting in dust on worst dust days in WRAP
(worst dust day: A 20% worst-case visibility day when dust was the largest contributor)
1. Long-range transport (Intercontinental)
2. Windblown events
3. Upwind transport
4. Undetermined sources
(2) and (3) sources/events were further identified as Local (1 site affected) or Regional (multiple sites affected in the area)
1. Developed an empirical/heuristic approach (EHA) usinga. Elemental concentration ratiosb. Multivariate linear regression analysis c. Air mass backward trajectories;d. Land use/land cover data for US;e. USGS Wind erodibility group (soil erosion metric) for US
2. Use of the EHA to assign worst dust days into a source/eventa. Integrate all tools into a geographical information system (GIS)b. Develop a set of guideline criteriac. Generate maps for each worst dust day
Methodology
April 29,1998: 17 of the WRAP IMPROVE monitoring sites had 20% worst dust days.
Period Elemental ratiosAl/Ca Al/Si Ca/Si Fe/Si K/Fe CM/Soil
Year 2001 1.4 0.31 0.22 0.27 0.67 7.10
Year 2002 1.7 0.43 0.25 0.25 0.72 16.02
April 29, 1998 Mean 2.1 0.52 0.25 0.29 0.59 2.11
St. dev. 0.3 0.06 0.03 0.04 0.07 0.94
Asian Dust
x/y%Ux/yΖΠ1ADS
2refyxσ
2
dayyxU
refyxdayyx
x/yZ
2
100yU
2
100xU
dayyxU
Zx/y is the Z-score for the ratio X/Y;Ux/y is the uncertainty of the ratio X/Yx/yday is the ratio of X/Y during a given day;x/yref is the reference ratio of X/Y estimated based on April 29,1998 event ;σ(x/y) is the standard deviation of reference X/Y ratios and;Ux and Uy are the measurement uncertainties for elements X and Y
Asian Dust Score (ADS)
Reference Asian Dust Ratio (based on 04/29/1998 episode): Al/Ca = 2.1 + 0.3; K/Fe = 0.59 + 0.07; Al/Si = 0.52 + 0.06
AGTI (12/19/02) BIBE (03/09/02) BAND (04/25/01)
Al/Ca 1.02 (0.21) 1.50 (0.13) 1.17 (0.10)
K/Fe 0.61 (0.05) 0.61 (0.04) 0.59 (0.04)
Al/Si 0.32 (0.07) 0.39 (0.03) 0.49 (0.04)
ADS 201 1809 40943
ADS Interpretation
0-750 Small signature; Unlikely Asian dust
750-1500 Moderate signature; Asian dust influence should be considered
> 1500 Strong signature; Asian dust influence is indicated by chemical analysis, confirmation using external tools is required
ADS interpretation
Multivariate linear regression analysis to estimate Locally-
generated Windblown Dust (LWD)eakxkb1kx1kb.....2x2b1x1bepymy
ym is the measured dust mass concentration on a given site and date;yp is the dust concentration estimated by a linear combination of independent variables that describe the wind conditions;b1, b2,……., bk are the regression coefficients of the independent variables;x1, x2,……., xk are the values of independent variables that describe the wind conditions;a is the intercept which corresponds to yp when x1, x2,……., xk are equal to 0 and;ε is the residual error - the difference between the ym and yp
kGkb1kG1kb.....2G2b1G1bLWD
CASTNET AZDEQ
NPSISH
RAWS
NASA
1-h Central Meteorological
Database
•Days with precipitation for more than 10h
•Precipitation occurred after 12:00 p.m.
1-h Modified Central
Meteorological Database
A B
WD1 0o - 90o < 45o or > 315o
WD2 90o - 180o 45o - 135o
WD3 180o - 270o 135o - 225o
WD4 270o - 359o 225o - 315o
Wind speed
WS1 < 14 mph
WS2 14 – 20 mph
WS3 20 – 26 mph
WS4 > 26 mph
WS1 WS2 WS3 WS4
WD1 G1 G2 G3 G4
WD2 G5 G6 G7 G8
WD3 G9 G10 G11 G12
WD4 G13 G14 G15 G16
Σ( (24-h
Database development
24-h Modified Central
Meteorological Database
24-h “Dust” Database
IMPROVE data2001-2003
Regression analysis• Only wind conditions groups corresponding to wind speed higher than 14 miles/hour• Least-squares method• Forward, backward and stepwise variable screening methods• Regression coefficients significant at p-value < 0.10 or 0.15• Regression coefficients with Variance infiltration factor VIF > 10 were rejected• Null hypothesis (H: μ1= μ2=..... = μk= 0) for both regression coefficients and model was
investigated• Only measurements with LWD – 2 standard error 0 were considered
MLRAnalysis
IMPROVE sites where dust/wind relationship exists
Overall: 129 IMPROVE sites 71 sites with available meteorological data43 sites with statistically significant MLRA results41 sites with reliable Local Windblown Dust (LWD) results
Example: Regression coefficients by quadrant
Badlands National Park, SD Bosque del Apache, NM
WD1WS3: 0.438 0.247 WD1WS2: 39.278 8.138
WD2WS3: 2.272 0.694 WD3WS3: 7.924 1.060
WD3WS2: 1.590 0.177
Although regression coefficients can be used to predict the dependent variable using a set of independent variables, it provides no information about the relative contribution of each independent variable because independent variables means and variances were not considered.
The standardized z-score coefficients were estimated, thus all independents variables have mean value of 0 and standard deviation of 1. The standardized regression coefficients (β1, β2,......, βn) provide enough evidence of the relative contribution of the independent variables.
Badlands National Park, SD
Bosque del Apache, NM
Example: Standardized regression coefficients by quadrant
LWD vs. Total Measured Dust
#WDD with predicted LWD / total WDD (number shows the total number of WDD)
Mean LWD / total TMD (number shows the mean dust conc for WDD)
LWD and TMD for reliable MLRA results
0
10
20
30
40
50
60
70
80
90
100B
adla
nds
Nat
iona
l Par
kB
ande
lier N
atio
nal M
onum
ent
Big
Ben
d N
atio
nal P
ark
Blis
s S
tate
Par
k (T
RP
A)
Bos
que
del A
pach
eB
ryce
Can
yon
Nat
iona
l Par
kC
anyo
nlan
ds N
atio
nal P
ark
Chi
ricah
ua N
atio
nal M
onum
ent
Col
umbi
a R
iver
Gor
geC
rate
rs o
f the
Moo
n N
M(U
S D
OE
)D
eath
Val
ley
Nat
iona
l Par
kG
ila W
ilder
ness
Gre
at S
and
Dun
es N
atio
nal M
onum
ent
Gua
dalu
pe M
ount
ains
Nat
iona
l Par
kH
illsi
deH
oove
r Wild
erne
ssIk
e's
Bac
kbon
eJo
shua
Tre
eK
alm
iops
is W
ilder
ness
Lava
Bed
sLo
stw
ood
Wild
erne
ssM
edic
ine
Lake
Wild
erne
ssM
esa
Ver
de N
atio
nal P
ark
Mou
nt B
aldy
Wild
erne
ssP
asay
ten
Wild
erne
ssP
inna
cles
Nat
iona
l Mon
umen
tP
uget
Sou
ndQ
ueen
Val
ley
Roc
ky M
ount
ain
Nat
iona
l Par
kS
agua
ro N
atio
nal P
ark
Eas
tS
agua
ro N
atio
nal P
ark
Wes
t (A
ZDE
Q)
Sal
t Cre
ek W
ilder
ness
San
Ped
ro P
arks
Wild
erne
ssS
awto
oth
Nat
iona
l For
est
Sie
rra A
ncha
Sim
eono
f Wild
erne
ssS
poka
ne R
es.
Sta
rkey
Theo
dore
Roo
seve
ltTo
nto
Nat
iona
l Mon
umen
tU
L B
end
Wild
erne
ssW
emin
uche
Wild
erne
ssW
hite
Mou
ntai
n W
ilder
ness
Whi
te R
iver
Nat
iona
l For
est
Zion
Nat
iona
l Par
k
Dus
t (μg
/m3)
DustWindblown dust
Air masses backward trajectoriesNOAA HYSPLIT trajectory model
For all sites and worst dust days:
Duration: 48-h and 192-hFrequency: Every 3 hours (8:00, 14:00 and 20:00)Resolution: 1 hourStarting heights: 10, 500 and 1500 m.a.g.l.
Trajectory speed (km/h) = distance between two trajectory points• 0 – 14 miles/hour• 14 – 20 miles/hour• > 20 miles/hour
Trajectory endpoint at 8:00 a.m. (CST)"
""
!!
!
###
0.00 < speed < 14.00 mph14.00 < speed < 20.00 mphspeed > 20.00 mph
0.00 < speed < 14.00 mph14.00 < speed < 20.00 mphspeed > 20.00 mph
0.00 < speed < 14.00 mph14.00 < speed < 20.00 mphspeed > 20.00 mph
Trajectory endpoint at 8:00 p.m. (CST)
Trajectory endpoint at 2:00 p.m. (CST)
Trajectories
Air masses backward trajectories
Land use / Land coverNational Land Cover Dataset 1992 (NLCD 1992)
Landsat Thematic Mapper satellite data (U. S. Geophysical Survey and U. S. Environmental Protection Agency)
Resolution: 30 meters
21 classes of land cover (Anderson Land Cover Classification)
11. Open Water12. Perennial Ice/Snow21. Low Intensity Residential22. High Intensity Residential23.Commercial/Industrial /Transportation32. Quarries/Strip Mines/Gravel Pits33. Transitional81. Pasture/Hay82. Row Crops83. Small Grains84. Fallow85. Urban/Recreational Grasses
41. Deciduous Forest42. Evergreen Forest43. Mixed Forest91. Woody Wetlands92. Emergent Herbaceous Wetlands61. Orchards/Vineyards/
51. Shrubland71. Grasslands/Herbaceous31. Bare Rock/Sand/Clay
Data were obtained from: http://landcover.usgs.gov
Land use / Land cover
Land use / Land cover: 3 categories
Wind Erodibility Group (WEG)
.
• Indicator of susceptibility to wind erosion based on:
• soil texture,• organic matter content, • effervescence due to carbonate reaction with HCl, • rock and para-rock fragment content • minerology.• Soil moisture and the presence of frozen soil also influence soil blowing.
. The range of valid entries for wind erodibility group data is 1, 2, 3, 4, 4L, 5, 6, 7, and 8
Source: US Department of Agriculture. National Resources Conservation Services National Soil Survey Handbook: Soil Properties and Qualities (Part 618)
Data were obtained from: http://water.usgs.gov/GIS/dsdl/muid.e00.gz
Wind Erodibility Group (WEG)
Combination of Land Use / Land Cover and Wind Erodibility Group
layers
X =
Human influenced layer
Grass- and shrub-lands layer
Forest s and wetlands layer
Background layer for GIS analysis
Forests & wetlandsLow erodibility based on soil texture High erodibility based on soil texture
Shrubland and grassland areas
Low erodibility based on soil texture
High erodibility based on soil texture
Human-induced areasLow erodibility based on soil texture
High erodibility based on soil texture
Land use and wind erosion
B
Precipitation occurred at the site
! IMPROVE site with a valid sample
Ï IMPROVE site without a valid sample
IMPROVE site
Trajectory endpoint at 8:00 a.m. (CST)"
""
!!
!
###
0.00 < speed < 14.00 mph14.00 < speed < 20.00 mphspeed > 20.00 mph
0.00 < speed < 14.00 mph14.00 < speed < 20.00 mphspeed > 20.00 mph
0.00 < speed < 14.00 mph14.00 < speed < 20.00 mphspeed > 20.00 mph
Trajectory endpoint at 8:00 p.m. (CST)
Trajectory endpoint at 2:00 p.m. (CST)
TrajectoriesAsian Dust Score (only shown for worst dust days)
3 ADS < 7503 750 < ADS < 1500
3 ADS > 1500
Local windblown dust (only shown for worst dust days
§ No Met data
! LWD/TMD = 0.00
! LWD/TMD < 0.25
! 0.25 < LWD/TMD < 0.50
! 0.50 < LWD/TMD < 1.00
! LWD/TMD > 1.00
Representation of multiple linear regression of wind conditions vs. total measured dust available for this site day
Low/moderate erodible forest areas
Low/moderate erodible human-
influenced areas
Moderate/high WEG shrubland areas
IMPROVE site without a valid sample Precipitation occurred
at this IMPROVE site
IMPROVE sites with a valid sample but not a
worst dust day
The ADS is higher than 1500 and the LWD/TMD is lower
than 0.25
The ADS is higher than 1500 and no LWD was calculated
because of no meteorological data
The ADS is not calculated due to absence of reliable
chemistry data and the LWD/TMD is between 0.50
and 1.00The ADS is higher than 1500
and the LWD/TMD is 0.00
No ADS and LWD/TWD were calculated because
meteorological and chemical data were not available
High speed 14:00 trajectory for WICA over moderate/high erodible shrubland
areas that are more than 24 hours away from the site
Moderate speed 8:00 trajectory over low/moderate
erodible shrubland areas
Moderate speed 20:00 trajectory over Mexico
Low speed trajectories at and near the site
Moderate/high speed trajectories upwind of the site
Event/Confidence Level
Educated guess(+)
Somewhat confident(+++)
Very Confident(++++)
1. Asian Dust Case 1: Asian Dust Score Available for site
{Asian Dust score > 1500 at multiple sites}
OR{(Asian dust score >1500 at
one site) AND
(back trajectories suggest air mass originated over Pacific
Ocean)}
Case 2: Asian Dust Score not available for site
{(Asian dust scores at multiple sites surrounding the site of
interest >1500) AND
(back trajectories suggest air mass originated over Pacific
Ocean)}
Case 1: Asian dust score available for site
{(Asian Dust Score > 1500 at multiple sites)
AND (back trajectories suggest air mass originated over Pacific
Ocean)}
Case 2: Asian Dust Score not available for site
{(Asian dust is primary event causing dust at multiple sites
surrounding the site of interest with a confidence of +++++)
Case 1: Asian dust score available for site
{(Asian Dust Score > 1500 at multiple sites
AND (back trajectories suggest air mass originated over Pacific
Ocean)AND
(satellite or models indicate large scale transport of dust
from Asia)}
Asian dust event
Asian dust example (***** for NOAB, SIAN, YELL and ZION)
Event/Confidence Level
Educated guess(+)
Somewhat confident(+++)
Very Confident(++++)
2. Windblown Dust
Case 1: Meteorological data available and MLRA showed
significant relationship between high wind conditions and dust
measured{(LWD to total measured dust ratio
>0)AND
(Back trajectories show high wind speed (>20 mph) at or near the site
over terrain with moderate or greater wind erodibility)}
Case 2: Meteorological data not available for day of interest but
MLRA showed significant relationship between high wind
conditions and dust measured at site of interest
{(Back trajectories show high wind speed (>20 mph) at the site over terrain with moderate or greater
wind erodibility) AND
(Worst dust days at one or more sites near the site of interest are caused by windblown emission
with confidence of “+++” or higher)}
Case 1: Meteorological data available and MLRA showed
significant relationship between high wind conditions and dust
measured{(LWD to Total measured dust ratio
>0.25) AND
(back trajectories show high (>20 mph) wind speed at or near the site
over terrain with moderate or greater wind erodibility)}
Case 2: Meteorological data not available for day of interest but
MLRA showed significant relationship between high wind
conditions and dust measured at site of interest
{(Back trajectories show high wind speed (>20 mph) at the site over terrain with moderate or greater
wind erodibility) AND
(Worst dust days at one or more sites near the site of interest are caused by windblown emission with confidence of “+++++” or
higher)}
Case 1: Meteorological data available and MLRA showed significant
relationship between high wind conditions and dust
measured{(LWD to total measured dust
ratio > 0.5) AND
(back trajectories show high (>20 mph) wind speed at or
near the site over terrain with moderate or greater wind
erodibility)}
Windblown dust event
Windblown dust example (*****)
Event/Confidence Level
Educated guess(+)
Somewhat confident(+++)
Very Confident(++++)
3. Transport from windblown dust sources upwind
Case 1: Meteorological data available
ANDNo evidence of local
windblown dust (LWD=o or not calculated)
(back trajectory for site shows up to three hours of high winds (>20 mph) over terrain with moderate or greater wind erodibility within one day of transport of the site)}
OR (back trajectories for multiple
sites shows up to 3 hours of high winds (>20 mph) over terrain with moderate or greater wind
erodibility within one day of transport of the site)}
Case 2: Meteorological data not available for day of
interest but MLRA showed significant relationship
between high wind conditions and dust measured at site of
interest{(back trajectory for site shows up to three hours of high winds
(>20 mph) over terrain with moderate or greater wind
erodibility within one day of transport of the site) but absence of high winds over the site itself}
Case 1: Meteorological data available
ANDNo evidence of local
windblown dust (LWD=o or not calculated)
(back trajectory for site shows up to 8 hours of high winds (>20 mph) over terrain with moderate or greater wind erodibility within one day of transport of the site)}
OR (back trajectories for multiple
sites shows up to 3 hours of high winds (>20 mph) over terrain with moderate or greater wind
erodibility within one day of transport of the site)}
Case 2: Meteorological data not available for day of
interest but MLRA showed significant relationship
between high wind conditions and dust measured at site of
interest{(back trajectory for site shows up to eight hours of high winds
(>20 mph) over terrain with moderate or greater wind
erodibility within one day of transport of the site) but absence of high winds over the site itself}
Case 1: Meteorological data available
ANDNo evidence of local
windblown dust (LWD=o or not calculated)
(back trajectory for site shows up to 15 hours of high winds (>20 mph) over terrain with moderate or greater wind
erodibility within one day of transport of the site)}
Case 2: Meteorological data not available for day of interest but MLRA showed significant
relationship between high wind conditions and dust
measured at site of interest{(back trajectory for site shows up to 15 hours of high winds (>20 mph) over terrain with moderate or greater wind
erodibility within one day of transport of the site) but absence of high winds over the site itself}
Upwind transport example (*****)
Event/Confidence Level
Educated guess(+)
Somewhat confident(+++)
Very Confident(++++)
2-3a. Windblown Dust: Regional
Event
{(Same as for 2 or 3.) AND
{(back trajectories for multiple sites indicate a common regional flow pattern)}
{(Same as for 2 or 3.)AND
{(back trajectories for multiple sites indicate a common regional flow
pattern)}
{(Same as for 2 or 3.)AND
{(back trajectories for multiple sites indicate a common regional flow
pattern)}
Regional windblown or upwind transport events
Regional event example: R1: ZION and BRCA; R2: MEVE and WEMI