Date post: | 11-Apr-2019 |
Category: |
Documents |
Upload: | duonghuong |
View: | 214 times |
Download: | 0 times |
Analysis and Validation of Winds from High Altitude Meteorological Analyses during Northern Hemisphere winters of 2009-2010 and 2012-2013
John McCormackSpace Science Division, Washington DC
SPARC Data Assimilation Workshop17 October 2016
Victoria, BC, Canada
SPARC DA Workshop 2016U.S. Naval Research Laboratory
Recent Advances in High-altitude DA
Introduction of hybrid ensemble DA methods provide more realistic background error covariance estimates than earlier static estimates
Significant reduction in forecast model upper-level cold bias;; temperature distribution key to getting upper level winds right.
Increase in temporal resolution: from 6-hourly analyses à 3 hourly analysis+forecast important for resolving tides!
Validation of NAVGEM winds with independent ground-based obs.
SPARC DA Workshop 2016U.S. Naval Research Laboratory
Why is stratospheric DA important?
Ocean Large Heat ReservoirLong Time Scales, Long Memory (ENSO/PDO)
TroposphereShort Time Scales
0-15 km Short Memory
Middle Atmosphere = stratosphere + mesosphere + lower thermosphere
Large Angular Momentum and Drag15-100 km Long Radiative/Dynamical Time Scale
Long Memory (NAM/AO/QBO)
wave energy
heat
Vertical Coupling
NAVGEM with AR- New SL/SI
dynamical core- New 4DVAR DA with
adjoint- New upper-level radiance channels
NOGAPS-ALPHA - Eulerian dynamical
core- 3DVAR DA- New GWD
- New SW/LW htg- H2O & O3
photochemistry
High Altitude NWP Research at NRL
U.S. Naval Research Laboratory | 4
• Collaboration between Marine Meteorology, Remote Sensing, & Space Science Divisions.
• Initially based on high-altitude version of the Navy Operational Global Atmospheric Prediction System (NOGAPS)
•ALPHA: Advanced Level Physics- High Altitude
• AR: Accelerated Representer
• NAVGEM: Navy Global Environmental Model 100 km
50 km
OperationalNOGAPS
2008 2010 2012 2016
L74 NAVGEM with Hybrid DA
- Ensemble-based DA combined with 4DVAR- Improved SW/LW htg at
upper levels- Diurnal O3 photo-
chemistry- Solar variability
Presentation Title | 5U.S. Naval Research Laboratory
NAVGEM High Altitude NWP System:DA component
Ensemble Error Covariance
(B)
E
9-hr Ensemble
Forecast (xbi)
Background(xb)
Observations(y)
Analysis(xa)
Long forecast:10 days
Short forecast: 9 hours
Data Assimilation(DA) System
E4D-Var
ForecastModel
NAVGEMObservation
Error (R)
Physics
Investigating E4D-Var
Ensemble Generation Testing
E4D-Var
Over 1M obs every 6 hoursRadiosondes, Pibals, Dropsondes, DriftsondeLand and Ship Surface Obs, Fixed and Drifting Buoys, Aircraft Obs Surface Winds: WindSat, ASCAT, SSMIS Feature Tracked Winds: AVHRR, MODIS, VIIRS Total Water Vapor: SSMI/SSMIS GPS Bending AngleIR Sounding: IASI and AIRS (NPP CrIS)MW Sounding: AMSU-A (Ch 4-14), SSMIS (Ch 2-7, 22-24), SSMIS/MHS, NPP ATMSMiddle Atmosphere: SABER T, MLS T, MLS O3, MLS H2O, SSMIS/UAS
NAVGEM T119 L74Model Grid Spacing
SPARC DA Workshop 2016U.S. Naval Research Laboratory
nav_L74 NOGAPS-ALPHA: lm = 74, nprlev = 38
-180 -90 0 90 180longitude (oE)
0
20
40
60
80
100
120
pres
sure
alt
itud
e (k
m)
1000
100
10
1
0.1
0.01
0.001
0.0001
pres
sure
(hP
a)
L74 NAVGEMSSMIS UAS Channels
Sponge layer
Forecast Model Setup• Semi-Lagrangian/semi-implicit dynamical core;; T119 spectral truncation (grid spacing ~110 km) • 74 hybrid σ-p levels, top at 6 x 10-5 hPa or ~116 km (for now);; Time step = 360 sec.
Middle Atmosphere Physics:• Parameterized non-orographic gravity wave drag, O3 & H2O photochemistry, exothermic chemical heating
NAVGEM High Altitude NWP System:Global forecast model component
U.S. Naval Research Laboratory
Comparisons with meteor radar windsNH winters 2009-2010 & 2012-2013
NAVGEM T119L74 AnalysesGlobal 1o lat/lon 3 hourly output
Nov 2009 – March 2010 (SSW late Jan 2010)
Nov 2012 – March 2013 (SSW early Jan 2013)
Meteor radar observation sitesLocation Periods Analyzed
Andenes 69.3oN , 16.0oE 1-18 Dec. 2009, 1-26 Jan 2010, 12-28 Feb 2010 1-20 Dec. 2012, 1-28 Jan 2012, 1-24 Feb. 2013
Trondheim 63.4oN , 10.5oE 1 Dec. 2012 – 28 Feb. 2013
Juliusruh 54.6oN , 13.4oE 1 Dec. 2009 – 28 Feb. 20101 Dec. 2012 – 28 Feb. 2013
Collm 51.3oN , 13.0oE 1 Dec. 2009 – 28 Feb. 20101 Dec. 2012 – 28 Feb. 2013
CMOR 43.3oN , 80.0oW 1 Dec. 2009 – 28 Feb. 2010 1 Dec. 2012 – 26 Feb. 2013
Bear Lake 41.9oN , 111.4oW 1 Dec. 2009 – 28 Feb. 2010 1 Dec. 2012 – 28 Feb. 2013
Ascension Is. 8.0oS , 14.4oW 1 Jan. 2010 – 31Mar. 2010Tierra del
Fuego 53.7oS , 67.7oW 1-31 Dec. 2012, 1 Feb. – 31 Mar 2013
Rothera 67.5oS, 68.0oW 15 Jan 2013 – 28 Feb 2013
U.S. Naval Research Laboratory
NAVGEM vs. Meteor radar V at 88 km
NAVGEM 3-hourly analyzed winds
Radar 1-hourly observations
U.S. Naval Research Laboratory
Wavelet analysis using S-Transform
S-Transform is based on approach described in Stockwell (IEEE Trans. Sig. Proc., 1996)
Pros: - No a priori assumptions
about periodicity- Time mean S-transform
spectrum exactly matches Fourier spectrum
- Flexibility in choosing wavelet “window” function
Con: Needs continuous time series.
Analysis of Ascension Is. V at 88km for Feb. 2010
SPARC DA Workshop 2016U.S. Naval Research Laboratory
Semi-diurnal tideMonthly mean amplitude and phase: V
Juliusruh 54oN
2009-2010 2012-2013
SPARC DA Workshop 2016U.S. Naval Research Laboratory
Diurnal tideMonthly mean amplitude and phase: V & U
Ascension Island 8oSJan-Mar 2010
SPARC DA Workshop 2016U.S. Naval Research Laboratory
2-day waveMonthly mean amplitude and phase: V
Ascension Island 8oSJan-Mar 2010
Tierra del Fuego 54oSDec 2012 –Feb 2013 -Mar 2013
SPARC DA Workshop 2016U.S. Naval Research Laboratory
Semi-diurnal tide & SSWs
Mean amplitudes of SW2 in V over 4 NH radar sites around SSWs of 2010 and 2013
SPARC DA Workshop 2016U.S. Naval Research Laboratory
Obs. Sensitivity TestsDo We Need Middle Atmosphere Obs.?
Latitude
SPARC DA Workshop 2016U.S. Naval Research Laboratory
Obs. Sensitivity TestsDo We Need Middle Atmosphere Obs.?
NAVGEM with MA OBS NAVGEM with no MA OBS
SPARC DA Workshop 2016U.S. Naval Research Laboratory
Summary
• New high-altitude NAVGEM system using hybrid ensemble/4DVAR DA provides 3-hourly global synoptic meteorological analyses from 0 to ~100 km.
• Comparison with independent meteor radar observations shows good agreement.
• NAVGEM analyses show modulation of SW2 around the time of major SSWs in Jan 2010 and Jan 2013.
• Middle atmospheric obs. (MLS/SABER/UAS) help capture tides