Status of the assimilation of GPS RO observations: the COSMIC Mission

L. CucurullJCSDA/UCAR

J.C. Derber, R. Treadon, and R.J. Purser

Status COSMIC successful launch, 14 April, 9:40pm Eastern time.

– USAF Minotaur rocket– 6 Spacecraft deployed

Initial Orbit 500 km - final orbit 800 km, 71 degrees inclination, 6 separate planes.

All 6 satellites are healthy and functioning normally - preparing for orbit raising.

6 GPS Radio Occultation Payloads in checkout phase – Payload hardware performing well - no problems noted- – Producing reasonable occultations

Payloads are being operated with the GPS payload on half time and TIP on full time. TBB is still going through its checkout process, operating occasionally.

A new version of GPS payload software is in the process of being loaded to all 6 satellites

Expect 2,000-3,000 occultations per day from constellation at final orbit

Input

Data

CDAAC

NESDIS

JCSDABUFR Files file / sounding

COSMIC (and CHAMP) data flow to JCSDA

OSDPD

GPS Radio Occultation

Basic measurement principle:

Deduce atmospheric properties based on precise measurement of phase delay and amplitude.

GPS radio occultation measurements & processing

s1, s2, a1, a2

s1, s2

1, 2

N

T, e, P

Raw measurements of phase and amplitude of L1 and L2

Raw measurements of phase of L1 and L2

Bending angles of L1 and L2

Bending angle

Refractivity

Single path

Spherical symmetry &Satellites orbits.

Ionospheric effect cancellation

climatology

Auxiliary meteorological data

Radio holographic method, Multi path

Milestones accomplished (I) GSI/GFS code - forward operators:

– Local Refractivity:» Tested three different versions

• Non linear forward operator

• Linear forward operator

• Improved linear forward operator

» Implementation and testing completed of the forward, TL and Adjoint codes for the improved linear forward operator

– Local Bending Angle:» Tested two different versions

• Using approximations to solve the integral (ECMWF approach)

• Solving the integral with less approximations and making use of the smoothed Lagrange-polynomial interpolators (NCEP approach) - easy upgrade to 2D

» Implementation and testing completed for the forward TL and Adjoint codes for the NCEP forward operator.

Milestones accomplished (II) QC, processing, data format

– Superob vertical levels appropriate to model vertical resolution

– Implementation of QC checks in the code (i.e. after UCAR QC) based on one month comparison (July 2005) of CHAMP observations of RO and model simulations (refractivity & bending angle).

– Provide feed-back to UCAR on the comparison of observations and model simulations of profiles of refractivity & bending angle in order to improve their QC flags

– Develop the model capability to read BUFR files for radio occultation observations. Conversion from WMO BUFR to NCEP BUFR formats.

Errors– Analysis of the RO errors in BUFR format and provide feed back to UCAR

– Tune for representativeness of the model

An analysis case: 45 RO soundings (no other data)

Refractivity

Bending angle

1st 3rd

2nd

O-FG (%) O-A (%)

Data Analysis (refractivity and bending angle)– Assimilation of a single observation of refractivity/bending (CHAMP)

– Assimilation of a single profile of refractivity/bending (CHAMP)

– Assimilation of all profiles available at a given analysis time» Observations of refractivity/bending alone

» Observations of current (conventional and satellite) available observations alone

» Both refractivity/bending and current observations together

Comparison between the analysis of soundings of refractivity and bending angle (CHAMP)

– Impact no neutral

– Different impact depending on the type of GPS RO observation being assimilated.

Data Assimilation impact studies (CHAMP): 2-month cycling with/without GPS refractivity/bending angle.

Milestones accomplished (III)

GSI/GFS Impact studies: Preliminary Results2-month cycling at T62L64 JCSDA has implemented and tested

the capability of assimilating profiles of Refractivity (N) and soundings of Bending Angles (BA) in the GSI/GFS DA system.

There are some encouraging preliminary results. Before being able to assimilate the RO data in operations we need to:

– Understand the differences between the assimilation of N and BA. Why in some cases the assimilation of N results in a larger improvement than the assimilation of BA and visa versa?

– Understand the strengths and weaknesses of the GPS RO data and the model in weather analyses and forecasts. Why does the assimilation of N or BA degrade the forecasts in some cases?

Current and future work Finalize the ongoing experiments with CHAMP data in order to understand the

differences between the assimilation of refractivity and bending angle. – QC– Superobs– Representativeness error– Correlated errors

Make sure that COSMIC RO files are providing what they are supposed to. Preliminary analyses/assimilation of COSMIC observations for more tuning of GSI/GFS. Keep providing feed back to UCAR/CDAAC on the quality of the GPS RO profiles. Parallel runs with COSMIC data to select a Forward Operator (refractivity or bending

angle) for operations. In addition, this will require some updates of the GSI code:– Sigma to hybrid vertical coordinate– Log(surface pressure) to surface pressure

Transition to operations. Evaluate more sophisticated forward operators (2D).