assimilation of supplemental observations during tropical ... · (gsi) data assimilation scheme was...
TRANSCRIPT
![Page 1: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/1.jpg)
Assimilation of Supplemental Observations during Tropical Cyclones in the NCEP Hybrid
Gridpoint Statistical Interpolation (GSI) scheme
Michael J. Brennan1, Sharanya J. Majumdar2, Daryl Kleist3, and Kate Howard3
1NOAA/NWS/NCEP National Hurricane Center, Miami, Florida 2RSMAS Division of Meteorology and Physical Oceanography, Miami, Florida 3NOAA/NWS/NCEP/Environmental Modeling Center, College Park, Maryland
31st Conference on Hurricanes and Tropical Meteorology 1 April 2014
1
![Page 2: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/2.jpg)
Motivation • Previous studies (e.g., Aberson 2010; Majumdar et al.
2013) have examined the impact of synoptic surveillance dropsonde data on GFS model forecasts of TC track
• In 2012 the NCEP Gridpoint Statistical Interpolation (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of 3D-Var and an Ensemble Kalman Filter (Wang et al. 2013)
• What is the impact of these supplemental observations in the new hybrid GSI on TC intensity and structure?
2
![Page 3: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/3.jpg)
Karen (2013) • Karen formed as a 45-kt tropical storm
early on 3 October 2013 in the Gulf of Mexico and reached a peak intensity of 55 kt later that day despite moderate vertical shear
• As the shear increased Karen steadily weakened before dissipating on 6 October
• Operational TC intensity guidance and global models showed Karen strengthening before reaching the northern Gulf Coast
• Hurricane Watch was issued from Grand Isle, Louisiana, to Indian Pass, Florida
3
GOES-E IR 3-6 October 2013
Kimberlain (2013)
![Page 4: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/4.jpg)
Karen Synoptic Evolution GFS Analysis
4
200-400 mb PV, 900-700 mb PV, 500-mb heights, and 200-400 mb layer average winds (kt)
![Page 5: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/5.jpg)
Karen G-IV Mission
• After the completion of the G-IV mission, it was recognized operationally that the 12Z GFS run trended much weaker with the cyclone
5
TROPICAL STORM KAREN DISCUSSION NUMBER 7
NWS NATIONAL HURRICANE CENTER MIAMI FL AL122013
400 PM CDT FRI OCT 04 2013
THE GLOBAL MODELS ARE NOW IN BETTER AGREEMENT ON
THIS EVOLUTION...AND SHOW THE MID-LEVEL CIRCULATION
WEAKENING OR DISSIPATING ENTIRELY IN THE NEXT DAY OR
TWO. IN PARTICULAR THE GFS IS WEAKER WITH ITS FORECAST
OF KAREN AFTER DATA FROM THE NOAA GULFSTREAM-IV
JET...WHICH SHOWED 200-MB WINDS WEST OF KAREN STRONGER
THAN PREVIOUSLY ANALYZED...WERE INCORPORATED INTO THE
12Z ANALYSIS. AFTER 24 HOURS... KAREN COULD STRENGTHEN
A LITTLE DUE TO AN INCREASE IN UPPER-LEVEL DIVERGENCE
AHEAD OF A MID-LATITUDE TROUGH...BUT SIGNIFICANT
STRENGTHENING IS NOT EXPECTED. AN ALTERNATIVE SCENARIO
IS THAT KAREN COULD BECOME COMPLETELY DECOUPLED FROM
THE DEEP CONVECTION AND WEAKEN.
G-IV dropsonde 200-mb winds (kt) and 1145 UTC GOES-E IR image
Can we quantify this impact?
![Page 6: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/6.jpg)
Karen NOAA G-IV Synoptic Surveillance Mission 0530-1300 UTC 4 October 2013
6
Sondes 1–14 Assimilated for 06Z
Sondes 15–38 Assimilated for 12Z
![Page 7: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/7.jpg)
Experiment Methodology • Quantify the impact of these observations using data
denial experiments
• Experiments run cycling GFS with the GSI hybrid EnKF data assimilation – Include all data (Control)
– Exclude G-IV dropsonde data (No Drop)
• Compare evolution of the TC and environment
• Run SHIPS statistical-dynamical TC intensity model (DeMaria et al. 2005) on output from Control and No Drop experiments
• All results shown here are from the 12Z cycle on 4 October to account for the impact of all dropsondes
7
![Page 8: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/8.jpg)
Low-Level Vortex and Shear F00 – 12Z 10/4/2013
8
Control No Drop
Shear Difference (Control – No Drop)
• 925-700 mb PV • 850-200 mb
wind shear magnitude
• 850-200 mb wind shear (kt)
![Page 9: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/9.jpg)
Low-Level Vortex and Shear F00 – 12Z 10/4/2013
9
Control Central Pressure: 1009 mb
No Drop Central Pressure: 1009 mb
925-700 mb PV (shaded), 850-200-mb vertical shear magnitude (kt), 850-200-mb vertical wind shear (kt)
![Page 10: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/10.jpg)
Vortex Structure (Analysis – 12Z 4 October) Control
10
PV (shaded), Potential Temperature, Wind (kt) Relative Humidity (shaded), PV, Wind (kt)
• W-E cross section along 25.2°N from 97°W to 83°W • Control shows more tilt in Karen’s PV tower in the 12Z analysis • Control also shows stronger upper-level winds west of Karen and more dry air
over the western part of Karen’s circulation relative to No Drop
![Page 11: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/11.jpg)
Vortex Structure (Analysis – 12Z 4 October) No Drop
11
PV (shaded), Potential Temperature, Wind (kt) Relative Humidity (shaded), PV, Wind (kt)
• W-E cross section along 25.2°N from 97°W to 83°W • Control shows more tilt in Karen’s PV tower in the 12Z analysis • Control also shows stronger upper-level winds west of Karen and more dry air
over the western part of Karen’s circulation relative to No Drop
![Page 12: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/12.jpg)
Drop 25 – 25.5°N 92.4°W 10Z 4 October
12
G-IV Drop 10 UTC 12 UTC Analysis: GFS Control, GFS No Drop
![Page 13: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/13.jpg)
Analyzed Profiles 1°W of Karen’s Center (25.2°N 90.9°W)
13 12 UTC Analysis: GFS Control, GFS No Drop
![Page 14: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/14.jpg)
Low-Level Vortex and Shear F06 – 18Z 10/4/2013
14
Control Central Pressure: 1009 mb
GFS Intensity: 39 kt
No Drop Central Pressure: 1008 mb
GFS Intensity: 43 kt
![Page 15: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/15.jpg)
Low-Level Vortex and Shear F12 – 00Z 10/5/2013
15
Control Central Pressure: 1008 mb
GFS Intensity: 41 kt
No Drop Central Pressure: 1007 mb
GFS Intensity: 42 kt
![Page 16: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/16.jpg)
Low-Level Vortex and Shear F18 – 06Z 10/5/2013
16
Control Central Pressure: 1009 mb
GFS Intensity: 38 kt
No Drop Central Pressure: 1009 mb
GFS Intensity: 36 kt
![Page 17: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/17.jpg)
Low-Level Vortex and Shear F24 – 12Z 10/5/2013
17
Control Central Pressure: 1009 mb
GFS Intensity: 32 kt
No Drop Central Pressure: 1008 mb
GFS Intensity: 32 kt
![Page 18: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/18.jpg)
Low-Level Vortex and Shear F30 – 18Z 10/5/2013
18
Control Central Pressure: 1009 mb
GFS Intensity: 28 kt
No Drop Central Pressure: 1009 mb
GFS Intensity: 35 kt
![Page 19: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/19.jpg)
Low-Level Vortex and Shear F36 – 00Z 10/6/2013
19
Control Central Pressure: 1007 mb
GFS Intensity: 27 kt
No Drop Central Pressure: 1006 mb
GFS Intensity: 41 kt
![Page 20: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/20.jpg)
Low-Level Vortex and Shear F42 – 06Z 10/6/2013
20
Control Central Pressure: 1008 mb
GFS Intensity: 28 kt
No Drop Central Pressure: 1005 mb
GFS Intensity: 40 kt
![Page 21: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/21.jpg)
Low-Level Vortex and Shear F48 – 12Z 10/6/2013
21
Control Central Pressure: 1007 mb
GFS Intensity: 26 kt
No Drop Central Pressure: 1004 mb
GFS Intensity: 38 kt
![Page 22: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/22.jpg)
Low-Level Vortex and Shear F54 – 18Z 10/6/2013
22
Control Central Pressure: 1007 mb
GFS Intensity: 28 kt
No Drop Central Pressure: 1003 mb
GFS Intensity: 42 kt
![Page 23: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/23.jpg)
Low-Level Vortex and Shear F60 – 00Z 10/7/2013
23
Control Central Pressure: 1006 mb
GFS Intensity: 27 kt
No Drop Central Pressure: 1003 mb
GFS Intensity: 48 kt
![Page 24: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/24.jpg)
Vortex Structure (F60) Control
24
PV (shaded), Potential Temperature, Wind (kt) Relative Humidity (shaded), PV, Wind (kt)
• NW-SE cross section along from 33.8°N 92.4°W to 24.7°N 83.4°W • By F60, Control has weak vortex with dry air above that does not intensify
ahead of approaching upper-level trough • Cyclone in No Drop is much deeper and appears to intensify in region of
upper-level divergence
![Page 25: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/25.jpg)
Vortex Structure (F60) No Drop
25
PV (shaded), Potential Temperature, Wind (kt) Relative Humidity (shaded), PV, Wind (kt)
• NW-SE cross section along from 34.5°N 91.2°W to 25.6°N 81.2°W • By F60, Control shows weak vortex with dry air above that does not intensify
ahead of approaching upper-level trough • Cyclone in No Drop is much deeper and appears to intensify in region of
upper-level divergence
![Page 26: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/26.jpg)
Karen Intensity – GFS Experiments
26
![Page 27: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/27.jpg)
SHIPS Model Experiments
27
• SHIPS run off Control shows a weaker cyclone after 24 h, but only by 3-4 kt
• Difference in intensity mainly due to weaker representation of the cyclone in GFS fields in Control relative to No Drop
• SHIPS shear calculation was quite similar in both runs (SHIPS a 0–500 km area average with the TC vortex removed to compute shear)
![Page 28: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/28.jpg)
Karen Track – GFS Experiments
28
12Z 10/4/2013 – Control, No Drop, Best Track
![Page 29: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/29.jpg)
Summary • G-IV data appear to result in slightly stronger shear and
more dry air aloft impinging on the circulation of Karen in the initial conditions at 12Z 4 October
• No Drop experiment shows 10-15 kt strengthening in 24-48 hours vortex as it approaches the northern Gulf Coast, perhaps through trough interaction
• Control experiment shows gradual decay after 12 hours, qualitatively similar to observations
• SHIPS experiments only show small differences, with SHIPS run from No Drop only 3-4 kt stronger than Control from 48-72 hours
• These results suggest that G-IV dropsonde data may be useful in improving forecasts of structure and intensity in some cases
29
![Page 30: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/30.jpg)
Future Work
• Examine additional cases (Isaac 2012, 2014?)
• See if any of the changes correlate with information in the EnKF-based ensemble
• Examine impacts of individual observations or groups of observations to see if symmetrical flight track of G-IV could be modified
30
![Page 31: Assimilation of Supplemental Observations during Tropical ... · (GSI) data assimilation scheme was upgraded to use a hybrid ensemble-variational approach with characteristics of](https://reader033.vdocuments.site/reader033/viewer/2022042922/5f6e73c00dc43819c2357973/html5/thumbnails/31.jpg)
Acknowledgements
• Thanks for Andrea Schumacher (CSU/CIRA) and Mark DeMaria (NHC) for running the SHIPS model experiments
31