- Email: [email protected]
Mesoscale initialisation using advanced sounder data
Pergamon www.elsevier.com/locate/asr
Adv. SpaceRes. Vol. 30, No. 11, pp. 2479-2484, 2002 © 2002 Published by Elsevier Science Ltd on behalf of COSPAR Printed in Great Britain 0273-1177/02 $22.00 + 0.00 PII: S0273-1177(02)00727-5
MESOSCALE INITIALISATION USING ADVANCED SOUNDER DATA L.M. Leslie x, J.F. Le Marshall 2 and W.L. Smith3
1 School o f Mathematics, University of New South Wales, Sydney, Australia, 2052 2 Bureau o f Meteorology Research Centre, Melbourne, Australia, 3000 3 NASA Langley Atmospheric Research Center, Hampton, VA, 23681, USA ABSTRACT There is increasing availability of continuous, high resolution observations from space-based platforms, for example, cloud and water vapour motion vectors, Advanced TOVS data and scatterometer wind fields, as well as increasing availability of continuous surface observations such as radar and profiler data. In addition, there will be, in the near future, an increase in the density of these data as instruments such as the Atmospheric InfraRed Sounder (AIRS) come into service. Full exploitation of this increased data base is expected to be attained through the use of modem data assimilation techniques such as 3- and 4dimensional variational assimilation. Here, we briefly review the 4-D variational (4-D Var.) data assimilation system, initially developed by Bennet et al. (1993, 1995, 1996) and recently extended to include the mesoscale assimilation of wind, temperature and moisture data over a limited area (Le Marshall et al. 1999, Leslie et al. 1998). We describe the assimilation methodology and the summarise initial results obtained by applying this high resolution system to mesoscale data from the N AST-I (NPOESS Advanced Sounder Testbed - Infrared) advanced sounding instrument. O 2002 Published by Elsevier Science Ltd on behalf of COSPAR. INTRODUCTION Early discussions of variational analysis may be seen in the papers of Thompson (1969), Sasaki (1969), Rodgers (1976) and Talagrand and Courtier (1987). In general, the idea o f Variational Analysis is to derive the model state vector _xthat minimises a (penalty) functional, J, defined as :
J(x) := 1/2 (x - xb)rB-' (x_- x_b) + 1/2 (2- 2°)r(C + D=)'1 02- 2 0)
(1)
wherex__b is a background estimate of x with error covariance B; z.is the vector of observed variables wffh instrumental error covariance C, Z~ is the predicted observed values given by
v = H(x2
(2)
and D is the error covariance in the "generalised interpolation", H which might be the radiative tr~aasfer equation to predictive satellite radiances or, in 4-D Var., a numerical weather prediction (NWP) model. In 1-D Var radiances may be used to give temperature and moisture soundings from a background profile vector. In 3-D Var. radiances are used to give temperature and moisture soundings, in conjunction with background field and other analysis data, so H is, indeed, a simple 2479
2480
L.M. Leslie et al.
interpolation. In 4-D Var. it is a full forward model. Typically, the x vector that minimises (1) is found by gradient descent, using calculation of J and 0J/0~ to find the minimum of J. Here, we have based our approach on that of Bennett et al. 1993, 1996 and 1997. Bennett et al. constructed the generalised inverse of a global numerical weather prediction (NWP) model in order to prepare initial conditions for the model at t = 0 hrs. Their inverse found a weighted least squares best fit to the dynamics for -24 < t < 0, the previous initial conditions and to data between t = -24 and t = 0. That is, the inverse is a weak constraint, four dimensional variational assimilation scheme. The best fit is found by solving the non-linear Euler-Lagrange equations which determine the local extrema of the penalty function. SYSTEM DEVELOPMENT
The works of Bennett et al. have been extended. The system has been implemented over a limited area (Le Marshall et al. 1996) and an extended penalty functional to allow it to use real time, continuous, high resolution cloud and water vapour drift wind data (Leslie et al. 1998), scatterometer data and TOVS data (Le Marshall et al. 1999, 2000). These works probably represent the first application to real situations of high resolution (up to 5 kin) 4-D Var. assimilation using emerging satellite data sources such as high spatial and temporal resolution atmospheric motion vectors (AMVs). These techniques have been applied principally to tropical cyclone (TC) track and intensity forecasting and are summarised in the next section. This paper addresses the 4-D Var. assimilation of advanced sounder data, first done at high resolution using aircraft data from the nonscanning High resolution Interferometer Sounder (HIS) instrument and described in a companion paper in this volume. Here, we have used data from the scanning NAST-I instrument which provide observations which are similar in character to those expected from advanced sounders such as the Geostationary Imaging Fourier Transform Spectrometer (GIFTS). The intent of this study and that in the companion paper is to explore the extent to which 4-D Var. assimilation using high spatial and temporal resolution data provides a complete mass and wind field specification, both in mid-latitudes and the tropics. RECENT APPLICATIONS The 4-D Var. assimilation system has been used previously in a series of experiments to assimilate continuous wind data to predict tropical cyclone motion (Le Marshall and Leslie, 1998). In these experiments, the system configuration was generally 180 x 180 gridpoints, 29 levels, 15 km resolution and was nested in the Bureau of Meteorology's Global Assimilation Prognosis System. The system has also been used at very high resolution to assess its utility in determining tropical cyclone intensity (Le Marshall and Leslie, 1999). In these experiments, the resolution was varied down to 5 km to demonstrate that resolution is a key factor in modelling tropical cyclone intensity. Improvements in intensity were also seen from the use of continuous atmospheric motion vector data. Overall results from these studies have shown the benefit of using a continuous high resolution atmospheric motion vector data base with modem continuous data assimilation techniques. In other experiments, scatterometer winds from ERS/2 (Le Marshall ctal. 2000) and TOVS data (Le Marshall et al. 1999) have been used in conjunction with these wind vectors to provide high resolution tropical cyclone track and intensity forecasts. In a recent study, advanced sounder data from the High resolution Intefferometer Sounder (HIS) has been assimilated using 4-D Var at 1 km resolution (Le Marshall et al. this volume).
Mesoscale Initialisation Using Advanced Sounder Data
ASSIMILATING
ADVANCED
2481
SOUNDER DATA
The first advanced Sounder to provide data for operational meteorology, the Advanced InfraRed Sounder (AIRS) will be launched in 2001. It will be followed by the Interferometer Advanced Sounder in the Infrared (IASI), the Cross-track Interferometer Sounder (CrlS) and sthe Geostationary Imaging Fourier Transform Spectrometer (GIFTS). To facilitate exploitation of these data, work is proceeding on the assimilation of high spatial, temporal and spectral resolution advanced sounder data, using 4-D Var. assimilation. The data used in this study comes from the NAST-I instrument, flown aboard a NASA. ER-2 aircraft during CAMEX-3. The NAST-I instrument is a scanning Michelson interferometer (Smith, 1997), providing interferograms at a ra~e sufficient to scan a 46 km swath at 2.6 km (nadir) resolution. The retrieval methodology used for these studies was the Eigenvector Retrieval method of Smith and Woolf (1976). The relationships between temperature, moisture and observed brightness temperature used in this study are derived using synthetic, high resolution brightness temperature spectra derived from radiosonde data. EXPERIMENT AND RESULTS The data on which this initial study is based was taken during several flights over Andros Is. (See Fig. 1) between 23 UTC on 13 September 1998 and 06 UTC on 14 September 1998. The study area is centred near 25E N, 77.5E W and was overflown, several times, by the NASA ER-2 aircraft. The areas covered by the flight are seen in Figs. 1 (a) and ASSIMILATION
1 (b).
271'.,I
25N g P I 0
23N. -80W
-78W
-76W
Fig. 1 (a) The observational data coverage from the NAST-I from 00 to02 UTC on 14 September 1998.
80W
78W
76W
Fig. 1 (b) As in (a) but for 02 UTC to 0430 UTC.
The assimilation system used a triply nested grid. The central grid consisted of 251 x 251 points at 1 km spacing. The surrounding, continuously variable grid increased by near 5% per grid point to 10 km resolution and then the outer variable grid increases to that of the NCEP/NCAR Reanalysis resolution. The assimilation was undertaken between T = 00 UTC and T = 04 UTC with the data being provided to the system in 15 minute bins. The initial and final analyses used to constrain the 4-D Var. assimilation system were sthe NCAR/NCEP Reanalysis, augmented by the NAST-I aircraft data. The system assimilated the temperature and moisture sounding data, initialising the model on its 1 km resolution grid. Output from the experiment is seen in Figs. 2 and 3. Figs. 2 (a) shows the NCEP/NCAR Reanalysis for 00 UTC on 24 September 1998 which was used to derive the analyses constraining the system. Fig. 2 (b) shows the resulting forecast 850 hPa
2482
L.M. Leslie et
al.
wind field after assimilation of NAST-I data while Fig. 3 shows the corresponding near real time NCEP/NCAR 850 hPa analysis. 14th i
Sept. 1998,
4th
850hPo
Sept.
1998,
850hPa
.
27N
26N -
25N •
~-~: ~,. ~-,.
~.
:'~
\
%
\i
\
\
\
"
\;%,
\
\
\
25N 24N "
\
i x .... × 23N 23N
"N
\:.\
\
;\
\
\
?
: 78W
80W
79W
78W "-E"
78W
7"/VV 8
Fig. 2 (b) The 850 hPa winds from the high resolution 4-D Var. assimilation at 04 UTC on 14 September 1998.
Fig. 2 (a) The N C E P / N C A R reanalyses for 00 UTC on 14 September 1998
Fig. 3 The near real time NCEP 850 hPa geopotential analysis for 00 UTC on 14 September 1998
80W 147£)
1500
75W 1525
70W 1550
1,575
The results shown here depart from the NCEP/NCAR Reanalysis fields but are consistent with cloud drift winds generated around around Andros Is. at 03 UTC. (Chris Velden, personal communication) and also with the near real time NCEP 850 hPa geopotential height analysis. In general, the 4-D Var. assimilated fields appear to be consistent with the data available from CAMEX-3. However, there is not sufficient wind data at an appropriate resolution for detailed verification. DISCUSSION
AND FUTURE
WORK
We have provided a brief history of the development of our 4-D Vat. system. Its ability to successfully assimilate cloud and water vapour drift winds, TOVS data and scatterometer winds has been summarised. Here, we have taken high temporal and spectral
Mesoscale Initialisation Using Advanced Sounder Data
2483
resolution advanced sounder data (2.6 km resolution and 2-hourly), generated from interfemgrams taken by the NAST-I instrument on board a NASA ER-2 aircraft over Andros Is. in September 1998 and successfully assimilated them into a 1 km 4-D Var. system. We believe that the study detailed here, along with that in the companion paper (Le Marshall et al. 2000) are probably the earliest 4-D variational assimilation work with real advanced sounder data. The work is continuing in order to characterise the extent to which near continuous advanced sounder data can eharacterise the mass and wind fields in the tropics and mid latitudes.
ACKNOWLEDGMENTS Thanks are due to T. Adair for his work in the preparation of this manuscript. REFERENCES Bennett, A.F., L.M. Leslie, C. Hagelberg and P.E. Powers, Tropical cyclone prediction using a barotropic model initialised by a generalised inverse method. Mon. Wea. Rev., 121, 1714- 1729, 1993. Bennel:t, A.F., B.S. Chua and L.M. Leslie, Generalized Inversion of a Global Numerical Weather Prediction Model. Meteor., Atmos. Phys., 60,165 -178, 1996. Bennel:t, A.F., B.S. Chua and L.M. Leslie, Generalised Inversion of a Global Numerical Weather Prediction Model II Analysis and Implementation, Meteor., Atmos. Phys., 62, 129- 140, 1997. Le Marshall, J.F., L.M. Leslie and A.F. Bennett, Tropical Cyclone Beti - an example of the benefits of assimilating hourly satellite wind data, Aust. Meteor. Mag. 46, 275 - 279, 1996. Le Marshall,J.F. and L.M. Leslie, Tropical Cyclone Track Prediction - Using High Resolution Satellite Data with a New Methodology, Aust. Meteor. Mag., 47, 261 266, 1998. Le Marshall, J. F. and L. M. Leslie, Modelling tropical cyclone intensity, Aust. Meteor. Mag. 48, 147-152, 1999. Le Marshall, J., G. Kelly, B. Choi, L. Leslie, G. Mills, D. Blank, P. Steinle and R. Seecamp, Recent Advances in the application of TOVS, ATOVS and Stretched VISSR data in Australia. Tech. Proc. Tenth International TOVS Study Conference, Boulder, Colorado, 27Jan. - 2Feb. 1999. 340 - 347, 1999. Le Marshall, J., L. Leslie, R. Morison, N. Pescod, R. Seecamp, C. Spinoso, Recent Developments in the Continuous Assimilation of Satellite Wind Data for Tropical Cyclone Track Forecasting. Adv. Space Res., 25, 1077 - 1080, 2000. Leslie L.M. and R.J. Purser, Three-dimensional mass-conserving semi-Lagrangian schemes employing forward trajectories, Mon. Weath. Rev., 123, 2551 - 2566, 1995. Leslie, L.M., J.F. Le Marshall, R. P. Morison, C. Spinoso, R.J. Purser, N. Pescod, and R. Seecamp, Improved hurricane track forecasting from the continuous assimilation of high-quality satellite wind data, Mon. Weath. Rev., 126, 1248 - 1257, 1998. Puff, K., G.S. Dietachmayer, G.A. Mills, N.E. Davidson, R.A. Bowen and L.W. Logan, The BMRC Limited Area Prediction System, LAPS, Aust. Meteor. Mag., 47, 203-224, 1998. Rodgers, C.D., Retrieval of Atmospheric Temperature and Composition from Remote Measurements of Thermal Radiation. Rev. Geophys. Sp. Phys., 14, 609 - 624, 1976. Sasaki Y., Proposed inclusion of time variation terms, observational and theoretical, in numerical variational objective analysis, J. Met. Soc. Japan 47, 115 - 124, 1969.
2484
L.M. Leslie et al.
Smith W. L. and H.M. Woolf, The use of eigenvectors of statistical covariance matrices for interpreting satellite sounding radiometer observations, J. Atmos. Sci. 33, 1127-1140, 1976. Smith W. L., H. E. Revercomb, H. B. Howell and H.M. Woolf, HIS-A satellite instrument to observe temperature and mois~tre profiles with high vertical resolution. Preprints Fifth Conference on Atmospheric Radiation AMS, Boston, 9 pp, 1983. Smith W., A. Larar, H. Howell, H. Revercomb, C. Sisko, D. Tobin, D. Cousins, D. Mooney, M. Gazarik and S. Mango, High spectral resolution sounding : latest results from aircraft and their implications for future satellite systems. Tech. Proc. The Ninth International TOVS Study Conference, lgls, Austria, 20 -26 February 1997, 1997. Talagrand, O. and P. Courtier, Variational Assimilation of meteorological observations with the adjoint vorticity equation - Part 1, Theory, Q. J. R. Meteor. Soc. 113, 1311 - 1328, 1987. Thompson P.D., Reduction of analysis error through constraints of dynamical consistency, Jnl. Appl. Met., 8, 738 - 742, 1969.
Adv. SpaceRes. Vol. 30, No. 11, pp. 2479-2484, 2002 © 2002 Published by Elsevier Science Ltd on behalf of COSPAR Printed in Great Britain 0273-1177/02 $22.00 + 0.00 PII: S0273-1177(02)00727-5
MESOSCALE INITIALISATION USING ADVANCED SOUNDER DATA L.M. Leslie x, J.F. Le Marshall 2 and W.L. Smith3
1 School o f Mathematics, University of New South Wales, Sydney, Australia, 2052 2 Bureau o f Meteorology Research Centre, Melbourne, Australia, 3000 3 NASA Langley Atmospheric Research Center, Hampton, VA, 23681, USA ABSTRACT There is increasing availability of continuous, high resolution observations from space-based platforms, for example, cloud and water vapour motion vectors, Advanced TOVS data and scatterometer wind fields, as well as increasing availability of continuous surface observations such as radar and profiler data. In addition, there will be, in the near future, an increase in the density of these data as instruments such as the Atmospheric InfraRed Sounder (AIRS) come into service. Full exploitation of this increased data base is expected to be attained through the use of modem data assimilation techniques such as 3- and 4dimensional variational assimilation. Here, we briefly review the 4-D variational (4-D Var.) data assimilation system, initially developed by Bennet et al. (1993, 1995, 1996) and recently extended to include the mesoscale assimilation of wind, temperature and moisture data over a limited area (Le Marshall et al. 1999, Leslie et al. 1998). We describe the assimilation methodology and the summarise initial results obtained by applying this high resolution system to mesoscale data from the N AST-I (NPOESS Advanced Sounder Testbed - Infrared) advanced sounding instrument. O 2002 Published by Elsevier Science Ltd on behalf of COSPAR. INTRODUCTION Early discussions of variational analysis may be seen in the papers of Thompson (1969), Sasaki (1969), Rodgers (1976) and Talagrand and Courtier (1987). In general, the idea o f Variational Analysis is to derive the model state vector _xthat minimises a (penalty) functional, J, defined as :
J(x) := 1/2 (x - xb)rB-' (x_- x_b) + 1/2 (2- 2°)r(C + D=)'1 02- 2 0)
(1)
wherex__b is a background estimate of x with error covariance B; z.is the vector of observed variables wffh instrumental error covariance C, Z~ is the predicted observed values given by
v = H(x2
(2)
and D is the error covariance in the "generalised interpolation", H which might be the radiative tr~aasfer equation to predictive satellite radiances or, in 4-D Var., a numerical weather prediction (NWP) model. In 1-D Var radiances may be used to give temperature and moisture soundings from a background profile vector. In 3-D Var. radiances are used to give temperature and moisture soundings, in conjunction with background field and other analysis data, so H is, indeed, a simple 2479
2480
L.M. Leslie et al.
interpolation. In 4-D Var. it is a full forward model. Typically, the x vector that minimises (1) is found by gradient descent, using calculation of J and 0J/0~ to find the minimum of J. Here, we have based our approach on that of Bennett et al. 1993, 1996 and 1997. Bennett et al. constructed the generalised inverse of a global numerical weather prediction (NWP) model in order to prepare initial conditions for the model at t = 0 hrs. Their inverse found a weighted least squares best fit to the dynamics for -24 < t < 0, the previous initial conditions and to data between t = -24 and t = 0. That is, the inverse is a weak constraint, four dimensional variational assimilation scheme. The best fit is found by solving the non-linear Euler-Lagrange equations which determine the local extrema of the penalty function. SYSTEM DEVELOPMENT
The works of Bennett et al. have been extended. The system has been implemented over a limited area (Le Marshall et al. 1996) and an extended penalty functional to allow it to use real time, continuous, high resolution cloud and water vapour drift wind data (Leslie et al. 1998), scatterometer data and TOVS data (Le Marshall et al. 1999, 2000). These works probably represent the first application to real situations of high resolution (up to 5 kin) 4-D Var. assimilation using emerging satellite data sources such as high spatial and temporal resolution atmospheric motion vectors (AMVs). These techniques have been applied principally to tropical cyclone (TC) track and intensity forecasting and are summarised in the next section. This paper addresses the 4-D Var. assimilation of advanced sounder data, first done at high resolution using aircraft data from the nonscanning High resolution Interferometer Sounder (HIS) instrument and described in a companion paper in this volume. Here, we have used data from the scanning NAST-I instrument which provide observations which are similar in character to those expected from advanced sounders such as the Geostationary Imaging Fourier Transform Spectrometer (GIFTS). The intent of this study and that in the companion paper is to explore the extent to which 4-D Var. assimilation using high spatial and temporal resolution data provides a complete mass and wind field specification, both in mid-latitudes and the tropics. RECENT APPLICATIONS The 4-D Var. assimilation system has been used previously in a series of experiments to assimilate continuous wind data to predict tropical cyclone motion (Le Marshall and Leslie, 1998). In these experiments, the system configuration was generally 180 x 180 gridpoints, 29 levels, 15 km resolution and was nested in the Bureau of Meteorology's Global Assimilation Prognosis System. The system has also been used at very high resolution to assess its utility in determining tropical cyclone intensity (Le Marshall and Leslie, 1999). In these experiments, the resolution was varied down to 5 km to demonstrate that resolution is a key factor in modelling tropical cyclone intensity. Improvements in intensity were also seen from the use of continuous atmospheric motion vector data. Overall results from these studies have shown the benefit of using a continuous high resolution atmospheric motion vector data base with modem continuous data assimilation techniques. In other experiments, scatterometer winds from ERS/2 (Le Marshall ctal. 2000) and TOVS data (Le Marshall et al. 1999) have been used in conjunction with these wind vectors to provide high resolution tropical cyclone track and intensity forecasts. In a recent study, advanced sounder data from the High resolution Intefferometer Sounder (HIS) has been assimilated using 4-D Var at 1 km resolution (Le Marshall et al. this volume).
Mesoscale Initialisation Using Advanced Sounder Data
ASSIMILATING
ADVANCED
2481
SOUNDER DATA
The first advanced Sounder to provide data for operational meteorology, the Advanced InfraRed Sounder (AIRS) will be launched in 2001. It will be followed by the Interferometer Advanced Sounder in the Infrared (IASI), the Cross-track Interferometer Sounder (CrlS) and sthe Geostationary Imaging Fourier Transform Spectrometer (GIFTS). To facilitate exploitation of these data, work is proceeding on the assimilation of high spatial, temporal and spectral resolution advanced sounder data, using 4-D Var. assimilation. The data used in this study comes from the NAST-I instrument, flown aboard a NASA. ER-2 aircraft during CAMEX-3. The NAST-I instrument is a scanning Michelson interferometer (Smith, 1997), providing interferograms at a ra~e sufficient to scan a 46 km swath at 2.6 km (nadir) resolution. The retrieval methodology used for these studies was the Eigenvector Retrieval method of Smith and Woolf (1976). The relationships between temperature, moisture and observed brightness temperature used in this study are derived using synthetic, high resolution brightness temperature spectra derived from radiosonde data. EXPERIMENT AND RESULTS The data on which this initial study is based was taken during several flights over Andros Is. (See Fig. 1) between 23 UTC on 13 September 1998 and 06 UTC on 14 September 1998. The study area is centred near 25E N, 77.5E W and was overflown, several times, by the NASA ER-2 aircraft. The areas covered by the flight are seen in Figs. 1 (a) and ASSIMILATION
1 (b).
271'.,I
25N g P I 0
23N. -80W
-78W
-76W
Fig. 1 (a) The observational data coverage from the NAST-I from 00 to02 UTC on 14 September 1998.
80W
78W
76W
Fig. 1 (b) As in (a) but for 02 UTC to 0430 UTC.
The assimilation system used a triply nested grid. The central grid consisted of 251 x 251 points at 1 km spacing. The surrounding, continuously variable grid increased by near 5% per grid point to 10 km resolution and then the outer variable grid increases to that of the NCEP/NCAR Reanalysis resolution. The assimilation was undertaken between T = 00 UTC and T = 04 UTC with the data being provided to the system in 15 minute bins. The initial and final analyses used to constrain the 4-D Var. assimilation system were sthe NCAR/NCEP Reanalysis, augmented by the NAST-I aircraft data. The system assimilated the temperature and moisture sounding data, initialising the model on its 1 km resolution grid. Output from the experiment is seen in Figs. 2 and 3. Figs. 2 (a) shows the NCEP/NCAR Reanalysis for 00 UTC on 24 September 1998 which was used to derive the analyses constraining the system. Fig. 2 (b) shows the resulting forecast 850 hPa
2482
L.M. Leslie et
al.
wind field after assimilation of NAST-I data while Fig. 3 shows the corresponding near real time NCEP/NCAR 850 hPa analysis. 14th i
Sept. 1998,
4th
850hPo
Sept.
1998,
850hPa
.
27N
26N -
25N •
~-~: ~,. ~-,.
~.
:'~
\
%
\i
\
\
\
"
\;%,
\
\
\
25N 24N "
\
i x .... × 23N 23N
"N
\:.\
\
;\
\
\
?
: 78W
80W
79W
78W "-E"
78W
7"/VV 8
Fig. 2 (b) The 850 hPa winds from the high resolution 4-D Var. assimilation at 04 UTC on 14 September 1998.
Fig. 2 (a) The N C E P / N C A R reanalyses for 00 UTC on 14 September 1998
Fig. 3 The near real time NCEP 850 hPa geopotential analysis for 00 UTC on 14 September 1998
80W 147£)
1500
75W 1525
70W 1550
1,575
The results shown here depart from the NCEP/NCAR Reanalysis fields but are consistent with cloud drift winds generated around around Andros Is. at 03 UTC. (Chris Velden, personal communication) and also with the near real time NCEP 850 hPa geopotential height analysis. In general, the 4-D Var. assimilated fields appear to be consistent with the data available from CAMEX-3. However, there is not sufficient wind data at an appropriate resolution for detailed verification. DISCUSSION
AND FUTURE
WORK
We have provided a brief history of the development of our 4-D Vat. system. Its ability to successfully assimilate cloud and water vapour drift winds, TOVS data and scatterometer winds has been summarised. Here, we have taken high temporal and spectral
Mesoscale Initialisation Using Advanced Sounder Data
2483
resolution advanced sounder data (2.6 km resolution and 2-hourly), generated from interfemgrams taken by the NAST-I instrument on board a NASA ER-2 aircraft over Andros Is. in September 1998 and successfully assimilated them into a 1 km 4-D Var. system. We believe that the study detailed here, along with that in the companion paper (Le Marshall et al. 2000) are probably the earliest 4-D variational assimilation work with real advanced sounder data. The work is continuing in order to characterise the extent to which near continuous advanced sounder data can eharacterise the mass and wind fields in the tropics and mid latitudes.
ACKNOWLEDGMENTS Thanks are due to T. Adair for his work in the preparation of this manuscript. REFERENCES Bennett, A.F., L.M. Leslie, C. Hagelberg and P.E. Powers, Tropical cyclone prediction using a barotropic model initialised by a generalised inverse method. Mon. Wea. Rev., 121, 1714- 1729, 1993. Bennel:t, A.F., B.S. Chua and L.M. Leslie, Generalized Inversion of a Global Numerical Weather Prediction Model. Meteor., Atmos. Phys., 60,165 -178, 1996. Bennel:t, A.F., B.S. Chua and L.M. Leslie, Generalised Inversion of a Global Numerical Weather Prediction Model II Analysis and Implementation, Meteor., Atmos. Phys., 62, 129- 140, 1997. Le Marshall, J.F., L.M. Leslie and A.F. Bennett, Tropical Cyclone Beti - an example of the benefits of assimilating hourly satellite wind data, Aust. Meteor. Mag. 46, 275 - 279, 1996. Le Marshall,J.F. and L.M. Leslie, Tropical Cyclone Track Prediction - Using High Resolution Satellite Data with a New Methodology, Aust. Meteor. Mag., 47, 261 266, 1998. Le Marshall, J. F. and L. M. Leslie, Modelling tropical cyclone intensity, Aust. Meteor. Mag. 48, 147-152, 1999. Le Marshall, J., G. Kelly, B. Choi, L. Leslie, G. Mills, D. Blank, P. Steinle and R. Seecamp, Recent Advances in the application of TOVS, ATOVS and Stretched VISSR data in Australia. Tech. Proc. Tenth International TOVS Study Conference, Boulder, Colorado, 27Jan. - 2Feb. 1999. 340 - 347, 1999. Le Marshall, J., L. Leslie, R. Morison, N. Pescod, R. Seecamp, C. Spinoso, Recent Developments in the Continuous Assimilation of Satellite Wind Data for Tropical Cyclone Track Forecasting. Adv. Space Res., 25, 1077 - 1080, 2000. Leslie L.M. and R.J. Purser, Three-dimensional mass-conserving semi-Lagrangian schemes employing forward trajectories, Mon. Weath. Rev., 123, 2551 - 2566, 1995. Leslie, L.M., J.F. Le Marshall, R. P. Morison, C. Spinoso, R.J. Purser, N. Pescod, and R. Seecamp, Improved hurricane track forecasting from the continuous assimilation of high-quality satellite wind data, Mon. Weath. Rev., 126, 1248 - 1257, 1998. Puff, K., G.S. Dietachmayer, G.A. Mills, N.E. Davidson, R.A. Bowen and L.W. Logan, The BMRC Limited Area Prediction System, LAPS, Aust. Meteor. Mag., 47, 203-224, 1998. Rodgers, C.D., Retrieval of Atmospheric Temperature and Composition from Remote Measurements of Thermal Radiation. Rev. Geophys. Sp. Phys., 14, 609 - 624, 1976. Sasaki Y., Proposed inclusion of time variation terms, observational and theoretical, in numerical variational objective analysis, J. Met. Soc. Japan 47, 115 - 124, 1969.
2484
L.M. Leslie et al.
Smith W. L. and H.M. Woolf, The use of eigenvectors of statistical covariance matrices for interpreting satellite sounding radiometer observations, J. Atmos. Sci. 33, 1127-1140, 1976. Smith W. L., H. E. Revercomb, H. B. Howell and H.M. Woolf, HIS-A satellite instrument to observe temperature and mois~tre profiles with high vertical resolution. Preprints Fifth Conference on Atmospheric Radiation AMS, Boston, 9 pp, 1983. Smith W., A. Larar, H. Howell, H. Revercomb, C. Sisko, D. Tobin, D. Cousins, D. Mooney, M. Gazarik and S. Mango, High spectral resolution sounding : latest results from aircraft and their implications for future satellite systems. Tech. Proc. The Ninth International TOVS Study Conference, lgls, Austria, 20 -26 February 1997, 1997. Talagrand, O. and P. Courtier, Variational Assimilation of meteorological observations with the adjoint vorticity equation - Part 1, Theory, Q. J. R. Meteor. Soc. 113, 1311 - 1328, 1987. Thompson P.D., Reduction of analysis error through constraints of dynamical consistency, Jnl. Appl. Met., 8, 738 - 742, 1969.