BRAVO Back-Trajectory Analyses Why Consider ATAD? History of use at Big Bend - How Good is our past?...

BRAVO Back-Trajectory Analyses

Why Consider ATAD? History of use at Big Bend - How Good is our past? History of comparison to other wind fields.

ATAD vs Hysplit with EDAS & FNL Wind Fields How different are they and why

Results of preliminary qualitative back-trajectory analyses.

BRAVO Meeting, San Antonio, TX, 23 March 2001

How are ATAD and Hysplit Different?1. Input data Raw sounding data for ATAD Modeled EDAS or FNL wind fields for Hysplit

2. Length of time data is obtainable 1946 - present for ATAD 1997 - present for Hysplit

3. Model methodology Average wind in a transport layer and radius for ATAD “Data following” for Hysplit (other options also)

4. Frequency of Output 4 starts/day, endpoints 8 times/day for ATAD, 5 day max length hourly starts, hourly endpoints for Hysplit, max length > 10 days

74794

71945

78897

78970

70026

70133

70200

70219

70231

70261

70273

70308

70316

7032670350

70361

70398

70414

7220172202

72206

72210

72214

72230

72233

72235

72240

7224872249

72250

7225172261

722657227472293

7230572317

7231872327

723407235772363

72364

723657237672387

72393

724027240372426

72440

72456724697247672489

72493

72501

72518

72520

7252872558725627257272582

72597 72632

7263472645726497265972662

726727268172694

7271272747

7276472768

727767278672797

74389

7445574560

78526

89009

89664

91165

91285

71043

71081

71082

71109

71119

71203

7160071603

7172271801

7181171815

7181671823

7183671845

7186771876

7190671907

71908

71909

71913

71915

71917

71924

71925

71926

71934

71957

71964

76225

7638276394

76723 7838478526

78583 78866

78954

78970

UA Sites Used by ATAD on 8/10 0 GMT

ATAD Methodologyfrom Heffter, 1980

EDAS Meteorological Data Archive

The EDAS (Eta Data Assimilation System) data are a product of the NWS' National Centers for Environmental Prediction (NCEP). EDAS is a data assimilation system consisting of successive 3-h Eta model forecasts and Optimum Interpolation (OI) analyses.

•79 by 55 Lambert Conformal grid ~ 80 km resolution.

• 22 vertical layers on constant pressure surfaces from 1000 to 50 mbar

• 3 hour time increment

• Upper Air Data: 3-D winds, Temp, RH

• Surface Data includes: pressure, 10 meter winds, 2 meter Temp & RH, Momentum and heat flux

•Data is available from 1/97 to present. Note Oct. 1999 data is unavailable due to a fire at NCEP

FNL Meteorological Data ArchiveThe FNL data is a product of the Global Data Assimilation System (GDAS), which uses the Global spectral Medium Range Forecast model (MRF) to assimilate multiple sources of measured data and forecast meteorology.

•129 x 129 Polar Stereographic Grid with ~ 190 km resolution.

• 12 vertical layers on constant pressure surfaces from 1000 to 50 mbar

• 6 hour time increment

• Upper Air Data: 3-D winds, Temp, RH

• Surface Data includes: pressure, 10 meter winds, 2 meter Temp & RH, Momentum and heat flux

•Data is available from 1/97 to present.

http://capita.wustl.edu/Asia-FarEast/reports/DustMcarloSim/fnlnhgrd.gif

ATAD vs Hysplit…How Different Can They Be?Begins: Sep. 20, 1999 hr 6 (jd 263)

Duration: 2 Hours

Solid=EDAS, Dash=FNLRed = 2000m

Green = 1000mBlue = 100m

Magenta = 10mBlack=ATAD

0 20 40 60 80 100 120

02

46

810

hours

Hei

ght(

Km

)

Well, at times, pretty different!

Begins: Sep. 15, 1999 hr 18 (jd 258) Duration: 2 Hours




0 20 40 60 80 100 120

02

46

810

hours

Hei

ght(

Km

)

And at Other Times…Quite Similar

A common pattern in July & Aug...Begins: Jul. 19, 1999 hr 6 (jd 200)

Duration: 2 Hours




0 20 40 60 80 100 120

02

46

810

hours

Hei

ght(

Km

)

7/18-7/28, first few days in August, and end of August

Atad most southerly, EDAS most easterly with FNL in between.

Height Differences Between EDAS & FNL can be large - even on order of Km

Begins: Jul. 24, 1999 hr 12 (jd 205) Duration: 2 Hours




0 20 40 60 80 100 120

02

46

810

hours

Hei

ght(

Km

)

Directional Differences Between EDAS & FNL can be large (See 1000m traj.)

Begins: Jul. 18, 1999 hr 6 (jd 199) Duration: 2 Hours




0 20 40 60 80 100 120

02

46

810

hours

Hei

ght(

Km

)

Even When Heights match, Directions can Differ Between EDAS & FNL (See 2000 m)

Begins: Aug. 6, 1999 hr 18 (jd 218) Duration: 2 Hours




0 20 40 60 80 100 120

02

46

810

hours

Hei

ght(

Km

)

Big Bend Overall Residence Time

0 to 0.0050.005 to 0.010.01 to 0.050.05 to 0.10.1 to 0.50.5 to 11 to 5.5

Percent of Endpoints

Jun 1, 1999 - Sep 30, 1999

Includes All Months

ATAD 5-day trajectories


0 to 0.0050.005 to 0.010.01 to 0.050.05 to 0.10.1 to 0.50.5 to 11 to 5.5


Jun 1, 1999 - Sep 30, 1999

Includes All Months

EDAS 1000m 5-day trajectories


0 to 0.0050.005 to 0.010.01 to 0.050.05 to 0.10.1 to 0.50.5 to 11 to 5.5


Jun 1, 1999 - Sep 30, 1999

Includes All Months

FNL 1000m 5-day trajectories

Overall Residence TimeJune - September, 1999

ATAD, EDAS, FNL


0 to 0.0050.005 to 0.010.01 to 0.050.05 to 0.10.1 to 0.50.5 to 11 to 6


Jun 1, 1999 - Sep 30, 1999

Includes All Months



0 to 0.0050.005 to 0.010.01 to 0.050.05 to 0.10.1 to 0.50.5 to 11 to 5.5


Jun 1, 1999 - Sep 30, 1999

Includes All Months



0 to 0.0050.005 to 0.010.01 to 0.050.05 to 0.10.1 to 0.50.5 to 11 to 5.5


Jun 1, 1999 - Sep 30, 1999

Includes All Months

EDAS 2000m 5-day trajectoriesBig Bend Overall Residence Time

0 to 0.0050.005 to 0.010.01 to 0.050.05 to 0.10.1 to 0.50.5 to 11 to 7


Jun 1, 1999 - Sep 30, 1999

Includes All Months

EDAS 100m 5-day trajectories

2 Wind Fields and2 Heights

Correlations between Overall Residence Times 3-Day Trajectories June-Sep 1999

atad EDAS2000 EDAS1000 EDAS100 EDAS10 FNL2000 FNL1000 FNL100 FNL10 atad 1.000 0.732 0.769 0.648 0.662 0.790 0.806 0.653 0.644EDAS2000 0.732 1.000 0.878 0.753 0.787 0.962 0.811 0.714 0.730EDAS1000 0.769 0.878 1.000 0.942 0.945 0.903 0.956 0.933 0.927 EDAS100 0.648 0.753 0.942 1.000 0.994 0.780 0.870 0.973 0.986 EDAS10 0.662 0.787 0.945 0.994 1.000 0.812 0.867 0.955 0.977 FNL2000 0.790 0.962 0.903 0.780 0.812 1.000 0.860 0.751 0.762 FNL1000 0.806 0.811 0.956 0.870 0.867 0.860 1.000 0.893 0.866 FNL100 0.653 0.714 0.933 0.973 0.955 0.751 0.893 1.000 0.989 FNL10 0.644 0.730 0.927 0.986 0.977 0.762 0.866 0.989 1.000

Average DifferencesClosest Comparison to Atad: FNL 1000m

-541 to -22-22 to -7-7 to -2-2 to 88 to 2525 to 9292 to 522

ATAD - FNL1000

June-Sept 1999 Overall Residence Time Differences3-Day Trajectories

EDAS vs FNL 1000m

-424 to -16-16 to -4-4 to -1-1 to 33 to 77 to 2222 to 198

EDAS1000 - FNL1000


EDAS at 2 Heights

-1630 to -19-19 to -6-6 to -2-2 to 33 to 88 to 2222 to 220

EDAS1000 - EDAS100


Which is “Right”? Why? Does it Matter?

1. Transport heights matter. ATAD “transport layer” vs Hysplit vertical wind speed driven transport height. To do “standard” residence time analyses with Hysplit may require 3-D analysis.

2. Tracer concentrations should help on days with large differences.

3. MM5 Winds should help.

4. Qualitative results for some residence-time-type analyses suggest similar results no matter which is used.

180 200 220 240 260 280 300

020

0040

0060

0080

00

Julian Day 1999

ng/m

3

CSU Ion SO4Nov 1Oct 1Sep 1Aug 1Jul 1

90th pctile

50th pctile

10th pctile

10/12

9/1

9/14

9/158/21

7/810/13

10/59/16

9/28/208/22

8/22 9/3

7/187/49/79/6

10/3010/3110/8

Maybe Not Real

180 200 220 240 260 280 300

020

0040

0060

0080

00

Julian Day 1999

ng/m

3

CSU Ion SO4Nov 1Oct 1Sep 1Aug 1Jul 1

90th pctile

50th pctile

10th pctile

10/12

9/1

9/14

9/158/21

7/810/13

10/59/16

9/28/208/22

8/22 9/3

7/187/49/79/6

10/3010/3110/8

Maybe Not Real

TX/MX Border

E. Texas

TX/MX Border with ATAD saying Mexico

Begins: Sep. 1, 1999 hr 12 (jd 244) Duration: 2 Hours




0 20 40 60 80 100 120

02

46

810

hours

Hei

ght(

Km

)

Example of “TX/MX Border” Trajectories

Begins: Oct. 12, 1999 hr 12 (jd 285) Duration: 2 Hours




0 20 40 60 80 100 120

01

23

45

6

hours

TLD

(Km

)Example of “East Texas” Trajectories

Begins: Aug. 4, 1999 hr 12 (jd 216) Duration: 2 Hours




0 20 40 60 80 100 120

02

46

810

hours

Hei

ght(

Km

)Example of ATAD saying Mexico while Hysplit is along the border.

Big Bend High Concentration Source Contribution

0 to 0.630.63 to 1.711.71 to 2.222.22 to 2.92.9 to 3.553.55 to 5.375.37 to 10.7

Res. Time/Equal Prob. Sfc.

Jul 1, 1999 - Oct 31, 1999For CSU.SO4 > 5090.81 ng/m3

Includes Months: 7,8,9,10

ATAD Trajectories


0 to 0.130.13 to 0.560.56 to 1.621.62 to 2.722.72 to 4.684.68 to 6.266.26 to 10.36




EDAS & FNL Hourly 1000m Trajectories


0 to 0.080.08 to 0.230.23 to 1.121.12 to 2.462.46 to 4.034.03 to 5.75.7 to 14.74




FNL Hourly 1000m Trajectories

High Concentration Source Contribution Function for CSU Sulfate Ion, July-Oct,

ATAD, EDAS & FNL started at 1000 m.


0 to 0.080.08 to 0.230.23 to 1.121.12 to 2.462.46 to 4.034.03 to 5.75.7 to 14.74






0 to 0.640.64 to 1.341.34 to 2.462.46 to 3.523.52 to 5.495.49 to 7.167.16 to 17.83




FNL Hourly 500m Trajtories


0 to 0.470.47 to 0.940.94 to 1.831.83 to 2.882.88 to 4.954.95 to 7.097.09 to 18.31




FNL Hourly 100 m Trajectories


0 to 0.080.08 to 0.160.16 to 0.280.28 to 0.620.62 to 1.451.45 to 3.663.66 to 8.62




Hourly FNL 2000 m trajectories

Big Bend High Concentration Conditional Probability

0 to 0.060.06 to 0.1090.109 to 0.2140.214 to 0.3530.353 to 0.4650.465 to 0.5860.586 to 1

Probability of ConditionMin endpts/grid= 5



5-day ATAD trajectories


0 to 0.0260.026 to 0.0690.069 to 0.1580.158 to 0.2780.278 to 0.3910.391 to 0.5370.537 to 1




EDAS & FNL Hourly 1000m Trajectories


0 to 0.020.02 to 0.040.04 to 0.0790.079 to 0.1380.138 to 0.290.29 to 0.4390.439 to 1





High Concentration Conditional Probability


0 to 0.020.02 to 0.040.04 to 0.0790.079 to 0.1380.138 to 0.290.29 to 0.4390.439 to 1






0 to 0.0350.035 to 0.0680.068 to 0.1170.117 to 0.2060.206 to 0.3160.316 to 0.4470.447 to 1






0 to 0.0330.033 to 0.0580.058 to 0.0920.092 to 0.1360.136 to 0.2020.202 to 0.3560.356 to 1






0 to 0.0380.038 to 0.0780.078 to 0.1380.138 to 0.2120.212 to 0.3330.333 to 0.5480.548 to 1




Hourly FNL 2000 m trajectories

Mean Concentration

0 to 500500 to 10001000 to 20002000 to 30003000 to 40004000 to 50005000 to 8569

CSU.SO4 ng/m3

Jul 1, 1999 - Oct 31, 1999


EDAS & FNL Hourly 1000m Trajectories Mean Concentration

0 to 500500 to 10001000 to 20002000 to 30003000 to 40004000 to 50005000 to 6908

CSU.SO4 ng/m3

Jul 1, 1999 - Oct 31, 1999


ATAD Trajectories

Mean Concentration

0 to 500500 to 10001000 to 15001500 to 20002000 to 30003000 to 40004000 to 7776

CSU.SO4 ng/m3

Jul 1, 1999 - Oct 31, 1999



Mean Concentration

0 to 500500 to 10001000 to 15001500 to 20002000 to 30003000 to 40004000 to 5293

CSU.SO4 ng/m3

Jul 1, 1999 - Oct 31, 1999



-482 to -42-42 to -12-12 to -3-3 to 33 to 99 to 40.540.5 to 448

FNL1000 - EDAS1000

Difference in High Residence TimeCSU Sulfate Ion

-1817.669 to -112.216-112.216 to -23.811-23.811 to -4.141-4.141 to 6.8946.894 to 20.43520.435 to 3030 to 87

ATAD - FNL1000

Difference in High Residence TimeCSU Sulfate Ion

What’s Next?1. To use Hysplit in the “usual” ATAD residence time-type analyses requires some consideration of trajectory height. 3-D residence time?

2. If ATAD’s directions are wrong, choices….. Modify ATAD to use EDAS and/or FNL winds Switch to Hysplit with some height consideration. Abandon ATAD in favor of Monte Carlo model

3. Further analyses of BRAVO data, measured sfc & upper air winds, tracer, MM5 winds, spatial patterns in chem data should all help in deciding which model to use and why.

4. Continue with analyses of other species and other types of trajectory-based analyses.

BRAVO Back-Trajectory Analyses Why Consider ATAD? History of use at Big Bend - How Good is our past?...

Documents

Transcript of BRAVO Back-Trajectory Analyses Why Consider ATAD? History of use at Big Bend - How Good is our past?...