Application of COSMIC refractivity in Improving Tropical Analyses and Forecasts H. Liu, J. Anderson, B. Kuo, C. Snyder, and Y. Chen NCAR IMAGe/COSMIC/MMM.

Application of COSMIC refractivity in Improving Tropical Analyses and Forecasts

H. Liu, J. Anderson, B. Kuo, C. Snyder, and Y. Chen

NCAR IMAGe/COSMIC/MMM

Status of current tropical analyses

• Over tropical oceans, radiosondes are sparse.

• Current global analyses of temperature and moisture rely heavily on satellite radiances and winds.

Status of current tropical analyses (cont.)

• Significant areas of cloud-cover may exist over oceans, e.g., in case of hurricanes. Cloudy radiances are not yet used in most analysis systems.

• In cloudy situations, satellite cloud drift and scatterometer winds are the major data. These have much larger observational errors than radiosondes and do not measure T & Q. The analyses and forecasts of T and Q may have large uncertainty.

• Initialization of hurricane forecasts from such analyses may also have large uncertainty.

• Study of the weather and climate over oceans (e.g., ITCZ and MJO) also needs more reliable analyses.

Radio Occultation (RO) Refractivity

• Contain temperature and moisture information.

• Has good coverage over tropical oceans and not affected by clouds and precipitations.

• Retrieval of temperature and moisture are accurate.

Our Goal

Application of RO refractivity (bending angle) to improve:

Analyses and forecasts of temperature, moisture, and wind over tropical oceans

This may benefit hurricane forecast as well as study of the weather and climate over the tropics, such as MJO, ENSO etc.

This work

Examine impact of current COSMIC and CHAMP RO refractivity on improving analyses and forecasts of T and Q in tropical Atlantic during Aug. 16 - 30, 2006 in the presence of satellite cloud drift wind observations.

425 RO profiles are assimilated with a non-local quasi-excess phase operator (Sokolovskiy et al, 2005) and WRF/DART ensemble data assimilation system.

Location of the RO refractivity profiles(Aug 16-31, 2006)

QC: RO data with differences from the forecasts more than 8 times the observation error are rejected. Almost all of the RO data are assimilated.

The preliminary experiments

• EXP 1: Assimilate satellite cloud drift winds only.

(Motivated by cloudy situations, especially hurricanes)

EXP 2: Assimilate satellite cloud winds + RO refractivity assimilated using the non-local excess phase operator.

36 ensemble members are used.

Continuously 6 hour cycle assimilation with WRF (at 36 km resolution) for Aug 16-30.

• Analyses and 6-h forecasts are verified to the dropsondes and radiosonde observations, which are withheld from the assimilations.

Sounds used for Verification(Aug 16-31, 2006)

The sounds include most dropsondes and a few radiosondes

Using ensemble data assimilation system

• Advanced (non-local) RO observation operators can be easily implemented (requiring only forward models) and tested. This is especially important for the tropics where the horizontal gradient of refractivity is the largest.

• Flow dependent forecast error covariance of T and Q is included in the assimilation of RO data and this may significantly improve retrieval of T and Q from RO data.

Observation error for excess phase

RO excess phase analysis error

RO excess phase 6-h forecast error

Moisture analysis error

Moisture 6-h forecast error

Temperature analysis error

Temperature 6-h forecast error

Wind vector analyses error

Wind vector 6-h forecast error

Impact on analyses of Hurricane Ernesto (2006)

Green dots:

Best Track

Analyses of surface pressure and wind (Ctl + GPS)

(The hurricane vortex is well captured in the analyses)

Impact on analyses of Hurricane Ernesto (2006)

Analyses of surface pressure and wind (Ctl)

(The hurricane vortex disappears in the analyses)

Green dots:

Best Track

Conclusion

The results show that the GPS RO data significantly improve analyses and forecasts of temperature and moisture over tropical oceans.

The RO significantly improve analyses of the hurricane Ernesto.

Study to examine the impact of RO data in the presence of AIRS retrievals in cloud free condition is under way.

The ensemble data assimilation system is available on www.image.ucar.edu/DAReS