Satellite
SSSP banner
SSSP Home
 
bullet
bullet
bullet
bullet
bullet
bullet
bullet Useful Datasets
for Satellite Data Processing

bulletEmail the Co-Chairs  
bulletSSSP Webmaster

Soundings

9. Alexander Uspensky
Group: Scientific and Research Centre on Space Hydrometeorology
(PLANETA), Russia
Instrument: ATOVS
Measurement: Vertical Temperature and Moisture Profiles (ISADP)

Level 1
Level 2
Operational Research
Global HRPT
Weather Climate

 

Regional temperature/humidity soundings of the atmosphere from NOAA-16 measurements

V.I.Soloviev, A.B.Uspensky, A.V.Kucharski

Scientific and Research Centre on Space Hydrometeorology «PLANETA» (SRC «PLANETA») 7, B.Predtechensky, 123242 Moscow, Russia

e-mail: uspensky @ imp.kiae.ru.; uspensky @ planet.iitp.ru.

The processing package has been developed to retrieve atmospheric temperature and humidity profiles from the ATOVS/NOAA-16 measurements of local area coverage. The basic algorithms of ATOVS data inversion are described. The accuracy characteristics of ATOVS based temperature profile retrievals are presented and discussed.

During 2000-2001 at the SRC Planeta, Moscow a software package has been developed and tested for ATOVS (NOAA-15, 16) data processing. It's name is ISADP for "Integrated System for the ATOVS Data Processing." "Integrated" means that the satellite data are analyzed combined with NWP outputs moreover the relevant radiosonde observations (RAOBS) are also accounted for [Soloviev et al, 2001]. The ISADP consists of two basic modules: first module performs preprocessing of ATOVS data while the second module is designed for thematic processing. The prototype for the first module is AAPP [Klaes et al, 1999], while the second module has been developed using IAPP [Li et al, 2000]. Both mentioned modules were obtained officially from Eumetsat (AAPP, end of 1998) and CIMSS Wisconsin (IAPP, 1999) and than were updated to existing computing facilities (PC, Windows 98). Along with this the framework around the ISADP has been settled to provide all necessary interfaces for input files (such as NWP, RAOBs) as well as for output sounding products. The NWP fields (Bracknell's 12h forecast for the 2.50 x 2.50 grid) and RAOBs are regularly extracted from special database supported by Hydrometeorological Center in Moscow while the ATOVS-based sounding products (vertical profiles of atmospheric temperature T(p) and humidity q(p)) are placed at the isolated section of the same database.

The main steps of the ATOVS thematic processing are the following:

  1. Forward model adjustment
  2. Detection of clouds & precipitation zones
  3. Derivation of the first guess
  4. Retrieval of geophysical parameters
  5. Quality control and editing of the retrievals (retrieval bias adjustment).

The procedures of model adjustment and cloud detection are quite similar to those from [Li et al, 2000]. Derivation of the first guess profiles T0(p), q0(p) is performed using technique of principal components analysis and linear statistical regression. This procedure is also quite similar to [Li et al, 2000]. Step 4 is carried out using iterative solution of original inverse problem, moreover unlike [Li et al, 2000] the modified procedure of smoothing parameter calculation is applied.

The new procedure has been developed for correction of biases in output retrievals. The biases are explained by the mathematical properties of inverse problem and can be generated by the model errors and the biases in the first guess used.

To reduce biases the following formulae is proposed for corrected retrievals:

Tcor(pk) = akTsat(pk) + (1-ak)Tfc(p).

Here Tsat(p), Tfc(p) are retrieved and forecasted temperature profiles; the weights ak are derived through minimization the RMS differences between Tcor and collocated RAOBs. According to results of experiments with (NOAA-16) ATOVS data the values ak are rather stable (for the matched sample of retrievals, forecasts and RAOB's during two week period) and change in the range 0.2-0.5 (for the levels between 850 and 100 hPa) and equal 1 for the levels above 100 hPa.

The algorithms and the software package ISADP were tested with ATOVS measurements; NOAA-16 ATOVS data of regional coverage (HRPT receiving station in Moscow) from the period March - June 2001 were used in a thorough comparison between ATOVS-based retrievals Tcor(p) as well as forecasts Tfc(p) and "true" profiles i.e. collocated RAOBs, or NWP analysis.

Retrieval error statistics is shown at fig 1. The plotted are RMS profiles, for corrected retrieval Tcor(p) and forecasts Tfc(p). The time period for compiled matched sample is 14 days of April-May 2001. The size of the sample is between 30 (for 10 hPa level) and 243 (for levels below 100 hPa). According to the analysis of this and other ensembles the mean RMS error of the Tcor(p) (averaged for the atmospheric layer between 1000 and 10 hPa) is in the range 1.6-1.650C while the mean RMS error for Tfc equals 1.90C. These data demonstrate rather small but positive impact of satellite data: the mean accuracy of retrievals is slightly better than the accuracy of 12h forecast. As it was expected, the positive effect is more pronounced for the levels above 100 hPa.

References

  1. Klaes D., Schraidt R. The European ATOVS and AVHRR processing package (AAPP) - Techn. Proc.10th Intern ATOVS Study Conf. Boulder USA, 27.01-02.02 1999, pp.288-294.
  2. Li J., Wolf W., Menzel P. at al. Global soundings of the atmosphere from ATOVS measurements: the algorithm and validation. - J. Appl. Meteor., 2000, v. 39, pp. 1248-1268.
  3. V.I.Soloviev,A.B.Uspensky,A.V.Kucharski. Regional temperature/humidity soundings of the atmosphere from NOAA-16 measurements. - Meteorology and Hydrology, 2002, p. 52-63.
  4. Klaes D., Schraidt R. The European ATOVS and AVHRR processing package (AAPP) - Techn. Proc.10th Intern ATOVS Study Conf. Boulder USA, 27.01-02.02 1999, pp.288-294.
  5. Lavanant L., Brunel P., Rochard G. et al. NOAA 15 sounding profiles retrieved with the ICI scheme. - Techn. Proc. 10th Intern. ATOVS Study Conf. 1999, pp 321-328.
  6. Li J., Wolf W., Menzel P. at al. International ATOVS processing package the algorithm development and its application in real data processing. - Techn. Proc. 10th Intern. ATOVS Study Conf. 1999, pp 348-359.
  7. Li J., Wolf W., Menzel P. at al. Global soundings of the atmosphere from ATOVS measurements: the algorithm and validation. - J. Appl. Meteor., 2000, v. 39, pp. 1248-1268.
  8. Reale A., Chalfant M., Wilson L. Scientific status of NOAA ATOVS sounding products. - Techn. Proc. 10th Intern. ATOVS Study Conf. 1999, pp 437-448.
  9. Smith W.L., Woolf H.M. The use of eigenvectors of statistical covariance matrices for interpreting satellite sounding radiometer observations.-J. Atm. Sci., 1976, v. 33, pp 1127-1140.
  10. Tilley F.H., Pettey M.E. et.al. Use of radiosondes in in NESDIS Advanced-TOVS sounding products system. 10th Conf. on Sat. Meteor. And oceanography, 9-14 Jan. 2000, Long Beach, Ca. pp.

 

Profile RMS Errors
Click for larger version

Fig. 1 Profile RMS errors for Tfc (), T° ( g ) and Tcor (h )