Global Estimates of Gravity Wave Parameters in the UTLS from GPS Temperature Retrievals

1
Global Estimates of Gravity Wave Parameters in
the UTLS from GPS Temperature Retrievals

• Ling Wang and M. Joan Alexander
• NorthWest Research Associates, CoRA Div.

2
Overview of Talk

• Introduction
• COSMIC and CHAMP GPS temperature data
• New method to derive gravity wave parameters and
  some preliminary results
• Uncertainties of the new GW analysis
• Summary

2/19
3
Introduction (1)

• GWs play significant roles in the dynamics and
  transport and mixing processes in the UTLS and
  they can also affect the formation of cirrus
  clouds.
• At present, observational constraints on GW
  parameters including momentum flux propagation
  direction are sorely needed to improve the
  modeling of GW effects on various processes and
  phenomena in the UTLS (and elsewhere).

3/19
4
Introduction (2)

• Hodograph analysis and Stokes parameters method
  (e.g., Eckermann and Vincent, 1989, Pure Appl.
  Geophys) have been used to derive GW propagation
  direction and other wave parameters.
• These methods require, however, both temperature
  wind information, which is NOT available for
  most satellite data.

4/19
5
Introduction (3)

• Ern et al. (2004, JGR) use phase differences
  between adjacent T profiles to get global
  estimates of GW horizontal wavelength and
  momentum flux from the CRISTA satellite data.

Horizontal wavelength and momentum flux at 25 km
altitude in August 2007 (Ern et al., 2004).
5/19
6
Introduction (4)

• Alexander et al. (2008, JGR) use a similar
  approach to derive GW horizontal structures from
  the HIRDLS satellite data.

Momentum flux and horizontal wavenumber between
20-30 km altitude in May 2006 (Alexander et al.,
2008).
6/19
7
Introduction (5)

• In most GW analyses of satellite data so far, GW
  propagation directions are not routinely derived.
• This study introduces a new method to estimate
  the complete set of GW parameters (including
  horizontal propagation direction) from the GPS RO
  temperature data alone.

7/19
8
Data (1)

• COSMIC (Constellation Observing System for
  Meteorology Ionosphere and Climate/Formosa
  Satellite 3) GPS RO data
• CHAMP (Challenging Minisatellite Payload) GPS RO
  data

• (from Anthes et al., 2008, BAMS)

• GPS RO technique phase delay of GPS signals --gt
  bending angle of radio wave --gt atmospheric
  refractivity --gt temperature profiles

8/19
9
Data (2)

• Unlike most other satellite data, GPS data do not
  have regular orbits.
• COSMIC 2006-present 1500 daily profiles
• CHAMP 2001-2008 150 daily profiles
• We use dry T profiles processed by UCAR CDAAC.
• Vertical resolution 1 km
• Accuracy sub-Kelvin
• Temperature data cover troposphere and
  stratosphere.

Map showing a typical daily COSMIC/CHAMP data
coverage.
9/19
10
Gravity Wave Analysis (1)
GW Temperature Perturbation

• Since large-scale wave motions (e.g., Kelvin
  waves) can have similar vertical wavelength as
  GWs, we extract GW perturbations by removing
  zonal modes 0-6.

One example of GPS temperature, background T
determined from our procedure and the resulting
GW temperature perturbation profile.
10/19
11
Gravity Wave Analysis (2)
T Amplitude Vertical Wavelength

• We derive the dominant GW T amp vertical
  wavelength at each height from wavelet analysis.
• GW amplitudes generally decrease with increasing
  latitudes.
• Vertical wavelengths also display strong
  latitudinal variability.

Dominant T amp vertical wavelength averaged
between 17.5-22.5 km during December
2006-February 2007.
11/19
12
Gravity Wave Analysis (3)
Horizontal Wave Vector
If we neglect time variation, at the same z for
the same vertical wavelength, phase differences
among adjacent soundings are related with
horizontal wave vector (k, l) and distances among
the soundings as follows
We can determine phase differences from
cross-wavelet analysis. If there are at least 3
soundings in one space-time cell (e.g., 15 deg by
15 deg by 6 hours), we can solve for an
over-determined linear problem for (k, l) using
LSF.
n soundings --gt n!/(n-2)!/2 equations, e.g., 4
soundings --gt 6 equations to solve for (k, l)
12/19
13
Gravity Wave Analysis (4)
Horizontal Wavelength
We use (15 deg X 15 deg X 6 hours) cells across
the globe. GWs are generally longer at lower
latitudes, being consistent with many previous
studies.
averaged between 17.5-22.5 km during December
2006-February 2007
13/19
14
Gravity Wave Analysis (5)
lt-- GW dispersion relation
Most of the waves captured in our analysis are
low intrinsic frequency inertia-GWs. The
latitudinal variation of intrinsic frequency is
consistent with radiosonde analysis (Wang,
Geller, and Alexander, 2005, JAS).
averaged between 17.5-22.5 km during December
2006-February 2007
14/19
15
Gravity Wave Analysis (6)
With T, k, l, m, intrinsic frequency derived,
the momentum flux components are evaluated using
the left formula. The magnitudes of MF are
reasonable comparing with previous studies. The
effects of both convection and topography on GW
excitations are clearly seen.
averaged between 17.5-22.5 km during December
2006-February 2007
15/19
16
Gravity Wave Analysis (7)
We compare GPS results with radiosonde analysis
using the Stokes parameters method. GWs
generally propagate eastward in the tropics and
westward in mid- and high-latitudes, being
opposite to the prevailing background winds. The
two analyses are not dealing with exactly the
same part of GW spectrum due to the details of
the two analyses.
18-24.9 km, December 2006-February 2007
16/19
17
Gravity Wave Analysis (8)
Topography is one of the major GW sources and
mountain waves are expected to propagate upwind
and orthogonal to the orientation of
topography. Results largely consistent with
what is expected from the orientation of the
southern Andes and the prevailing westerlies
17.5-22.5 km, June-August 2007
17/19
18
Uncertainties of Analysis

• We assume the existence of a single dominant GW
  mode in each (lon, lat, time) cell at each height
• - Can reduce uncertainties by reducing
  the size of each cell and setting amplitude
  thresholds for dominant modes
• - by including treatment for secondary
  modes (if there are any)
• We neglect time variation of GW modes
• - Can solve this by using the
  following complete formula to solve for k, l,
  ground-based frequency simultaneously (which puts
  additional requirement for data coverage though)

18/19
19
Summary

• We introduce a new method based on the
  cross-wavelet analysis to estimate the complete
  set of GW parameters from the GPS RO temperature
  data.
• We present some preliminary results in the UTLS
  which demonstrate the effectiveness of the new
  method.
• The uncertainties of this new GW analysis and
  possible ways to deal with them are discussed.

19/19