121

Australian Meteorological and Oceanographic Journal 60 (2010) 121-125

Introduction

The radio occultation (RO) technique, employing the GPS constellation of satellites and radio receivers in low earth orbit, can provide valuable atmospheric profile information (Ware et al. 1996; Kursinski et al. 1996; Rocken et al. 1997; Wickert et al. 2001). The radio occultation technique has been long used in the study of planetary atmospheres (e.g. Fjeldbo et al. 1971) where it has provided much valuable information. The radio occultation technique is based on measuring the

bending of radio waves by refractive index gradients in the atmosphere. By observing this bending, a vertical profile of bending angle or refractive index can be derived and subsequently a profile of temperature/moisture. In the data-sparse southern hemisphere, observations of this type are vital for analysis, forecasting and climate monitoring. These data have recently been employed in some analysis centres around the globe where they have been shown to improve operational global numerical weather prediction (e.g. Cucurull and Derber 2008; Cucurull et al. 2007, 2008). Here we describe the use of COSMIC/FORMOSAT-3 Mission (Anthes et al. 2008), MetOp and GRACE radio occultation data in the ACCESS-G forecast system at the Bureau of Meteorology.

The beneficial impact of radio occultation observations on Australian region forecasts

John Le Marshall 1,2, Yi Xiao1, Robert Norman2, Kefei Zhang2, Anthony Rea3,Lidia Cucurull4, Rolf Seecamp3, Peter Steinle1, K. Puri1 and Tan Le1

1Centre for Australian Weather and Climate Research- a partnership between CSIRO and the Bureau of Meterology (CAWCR)

2RMIT University, Australia3Bureau of Meteorology, Australia

4NASA, NOAA and DoD Joint Center for Satellite Data Assimilation (JCSDA), USA

(Manuscript received March 2010, revised May 2010)

The Constellation Observing System for Meteorology, Ionosphere and Climate (COSMIC) was launched in April 2006. This system provides a new observation type for operational meteorology that has been shown to provide significant information on the thermodynamic state of the atmosphere and to allow improvements in atmospheric analysis and prognosis. A month of COSMIC radio occultation observations, together with a smaller number of radio occultation observations from the Meteorological Operational satellite (MetOp) and Gravity Recovery and Climate Experiment (GRACE) spacecraft, have been assimilated into the global ACCESS (Australian Community Climate Earth System Simulator) system, which is being employed at the Australian Bureau of Meteorology to provide real-time operational forecasts. In this study, four-dimensional variational assimilation (4DVAR) has been used to assimilate the radio occultation and other data into the global ACCESS system (ACCESS-G), which has been used to provide forecasts to five days ahead. For the period studied, the accuracy of these forecasts has been compared to forecasts generated without the use of the radio occultation data. The forecasts using radio occultation data have been found to be improved in the lower, middle and upper troposphere. These results, combined with the relatively unbiased nature of the radio occultation observations indicate their use has the potential to improve operational analysis and forecasting in the Australian Region and also to assist in important tasks such as a regional reanalysis and climate monitoring.

Corresponding author address: J. LeMarshall, Centre for Australian Weather and Climate Research, Bureau of Meteorology, GPO Box 1289, Melbourne, Vic 3001, Australia.Email: j.lemarshall@bom.gov.au

122 Australian Meteorological and Oceanographic Journal 60:2 June 2010

The ACCESS system is based on the UK Meteorological Office Unified Model. We examine the utility of these data for forecasting over the southern hemisphere and in particular in the Australian region. We also note the characteristics of these data and point out the important ways they improve the database used for operational meteorology and in climate monitoring and forecasting.

Radio occultation

The US global positioning system (GPS) comprises a constellation of 24 satellites that continuously transmit radio signals at frequencies of 1.57542 GHz (L1) and 1.22760 GHz (L2). When viewing these satellites from a low earth-orbiting satellite (LEO) these GPS satellites are seen to rise or set behind the earth. As the GPS satellites rise or set, the signals measured by a receiver on the LEO satellites are directly affected by the earth’s ionosphere and atmosphere. Observations of the rising and setting of these satellites (GPS occultations) can be used to provide limb soundings of the earth’s atmosphere. These measurements have global coverage, high vertical resolution, and can provide information in almost all weather conditions. The active nature of the measurements and the inversion methodology provides observations showing little bias through the upper troposphere and a significant part of the stratosphere. Several authors have provided the detailed theory of radio occultation using a receiver in space (Melbourne et al. 1994; Kursinski et al. 1997; Rocken et al. 1997). The methodology is based on measuring the amount of bending by refractive index gradients, of radio waves emitted by a GPS satellite, when viewing the occultation process from a low earth-orbiting satellite. The process involves the bending of a radio wave from the GPS satellite around the earth when received by a LEO, as the GPS satellite rises or sets. The phase and amplitude of the transmitted GPS signal and the Doppler shift is measured by the receiver on the LEO satellite. Where the position and velocity of the satellites are known accurately, the Doppler shift can be used, assuming spherical symmetry, to yield bending angle (α) and impact parameter a = nr sinφ (which is physically the distance of closest approach with no bending) where n is the refractive index, r is the radial distance and φ is the angle between the ray vector and radius vector (see Kursinski et al. (2000), Fig. 1). If the atmosphere is spherically symmetric, x being the geocentric radial distance, the bending angle may be written as;

α(a) = -2a ∫ ∞

a(d ln n / dx)/(x2 – a2)1/2 dx ...1

where x = nr After using the two GPS satellite transmission frequencies to remove the bending due to the ionosphere (Vorob’ev and Krasil’nikova 1994), the variation of the corrected bending angle with the impact parameter can be inverted with an Abel transformation (Phinney and Anderson 1968; Fjeldbo et al. 1971) to recover the refractive index profile.

n(x) = exp( 1/p ∫ ∞

x α(a)/ ((x2 – a2)1/2) da ...2

The refractive index profile, n(x), can be recovered numerically. The refractive index n is related to the refractivity N through n = 1 + 10-6 N. In the neutral atmosphere the refractivity is related to the temperature T and the partial pressure of dry air PD and water vapour PW through;

N = k1 PD /T + k2 PW /T + k3 PW /T2 ...3

where the ki are experimentally determined constants (see, for example, Bevis et al. 1994). In a dry atmosphere (PW = 0), refractivity is proportional to density and the refractivity profile can be used to produce a temperature profile using the ideal gas law. In our case, however, the refractivity measurement contains both temperature and water vapour information and this information can be derived using a priori profile information and statistically optimal variational inversion (e.g. Healy and Eyre 2000).

The data impact study

The dataOne month of GPS radio occultation data from the COSMIC, MetOp and GRACE missions have been used in a data impact study. The period of the study is from 26 February 2009 to 26 March 2009. The radio occultation data taken from the COSMIC constellation have been processed at UCAR to provide observation data including pressure, temperature and moisture profile information, bending angle, refractivity, data quality and other important information. The data have been transmitted on the global telecommunications system (GTS) in BUFR format. The data coverage (1638 sounding positions) for 15 March 2009 is shown in Fig. 1. A smaller amount of GRACE data processed at the German Research Centre for Earth Science (GFZ) and MetOp data processed by the GRAS Satellite Application Facility (DMI 2008) has also been used. In this study we have assimilated refractivity. The expected fractional refractivity errors (Kursinski et al. 1997, 2000) are around 0.002 at altitudes between 9 km and 25 km

Fig. 1 GPS radio occultation sounding positions for 15 March 2009. (Image courtesy of UCAR.)

Le Marshall et al.: Impact of radio occultation obsrvations on forecasts 123

and increase to near 0.01 as the profile descends from 9 km to the surface. This translates to better than one degree Kelvin temperature accuracy from the upper troposphere into the lower stratosphere (9 km to 25 km). GPS/MET temperatures have been found to be consistent with ECMWF analyses at the 1 to 1.5 K level RMS (Kursinski et al. 1996; Rocken et al. 1997) while geopotential height comparisons with ECMWF have been found to be consistent at the 20 m level (Leroy 1997). GPS radio occultation data during the period studied (5 March to 20 March 2009) have been compared to radiosonde observations (within 400 km) and some results are shown in Fig. 2. The figure shows a plot of the mean temperature dif-ference (red) between COSMIC and radiosonde observations for the period studied. Little bias is seen in the upper troposphere and lower stratosphere. This characteristic of the radio occultation data make them potentially very useful for numerical modelling and climate studies, where they can remove bias from the analysis process.

The assimilation trialAs a prelude to radio occultation data being incorporated into the Bureau’s new operational NWP suite, the ACCESS system (which is now being employed at the Australian Bureau of Meteorology to provide real-time operational forecasts) assimilation trials, using COSMIC, MetOp and GRACE radio occultation observations taken over the globe, were undertaken using the global version of the ACCESS system (ACCESS-G). The system and assimilation methodology are described below. The trials examined the impact of using GPS radio occultation observations on forecast accuracy in the southern hemisphere and in particular over the Australian region. The results from the trial were consistent with earlier results over the southern hemisphere (for example Cucurull et al. (2008)) and importantly provided quantification of the beneficial impact of the radio occultation data over the Australian region.

The assimilation systemThe ACCESS system was implemented to upgrade the Bureau’s NWP and climate modelling capabilities. The system employed in this trial, ACCESS-G, is based on the atmospheric model of the UK Meteorological Office (UKMO). The system was initially implemented in the Centre for Australian Weather and Climate Research (CAWCR) and is now used operationally in real time in the Bureau’s National Meteorological and Oceanographic Centre (NMOC). The ACCESS-G system has a global domain and a horizontal resolution of around 100 km with 50 vertical levels. The ACCESS system has been configured using the UKMO’s Suite Control System, and uses the UKMO Unified Model (UKUM), the 4DVAR analysis system, the UKMO’s Observation Processing System (OPS) and the Surface Fields Processing system (SURF). The ACCESS-G suite uses daily sea-surface temperature (SST) analyses from the BLUElink (Brassington et al. 2007) ocean forecasting system.

The ACCESS-G suite runs with a six-hour cycle, centred on four times per day (0000 UTC, 0600 UTC, 1200 UTC, 1800 UTC), using observations processed by OPS, with a 4DVAR analysis. The ACCESS-G system is summarised in Table 1.

The data assimilation methodologyIn this trial the impact of using GPS radio occultation obser-vations on forecast accuracy in the southern hemisphere, and in particular over the Australian region, was examined. For the period studied, four-dimensional variational assimilation was employed to assimilate radio occultation refractivity data into the global ACCESS system (ACCESS-G), which provided forecasts to five days.

Fig. 2 A plot of the mean temperature difference (red) be-tween COSMIC and radiosonde observations for the period, 5 March to 20 March 2009. The number of samples (blue) is also shown. (Data courtesy of UCAR.)

Table 1. A summary of the characteristics of the ACCESS-G system.

Domain Global

Unified Model horizontal resolution (217 x 288) ~125km x 83km

Analysis horizontal resolution (163 x 216) ~166km x 111km

Vertical resolution L50 (50 vertical levels)

Observational data used AIRS, ATOVS, SCAT, AMV, (6 h window) SYNOP, SHIP, BUOY, AMDAR, AIREP, TEMP, PILOT

Sea-surface temperature (SST) Weekly global 1° SST analysis analysis

Model time step 15 minutes

Analysis time step 40 minutes

124 Australian Meteorological and Oceanographic Journal 60:2 June 2010

The forecasts from 0000 UTC for the control run and the experimental run were verified after five days against their own analyses. The control run contained the full operational database used at the Bureau of Meteorology and included the data listed in Table 1. The experimental run was identical to the control run in all aspects except that it added radio occultation data to the full operational database for use in the analysis step. Typically 500 to 600 occultation observations are available to the system every six hours, of which about two-thirds of the data are assimilated usually between around 580 to 20 hPa.

The results

The S1 skill score, the root mean square errors and the anomaly correlations (ACs) have been computed for the forecasts to five days. An improvement in the forecast accuracy was seen in mean sea-level pressure and also through the lower, middle and upper troposphere. The magnitude of the improvements was appreciable. Verifications for mean sea-level pressure, 500 hPa geopotential height and 200 hPa geopotential height are summarised below in terms of the root mean square errors and the anomaly correlations in Figs 3 and 4. Figure 3 shows verification over both the Australian region (65.0°S-17.125°N, 65.0°E-184.625°E) and over the southern hemisphere annulus (60°S-20°S, 0°E-360°E) for mean sea-level pressure (MSLP). The addition of the COSMIC, MetOp and GRACE radio occultation data has led to an improvement for the one to five-day period. Similar results were seen for 850 hPa, 500 hPa and 200 hPa geopotential heights both for the Australian region and over the southern hemisphere annulus. In Fig. 4 the beneficial impact of the GPS radio occultation data is shown for 500 hPa geopotential height and 200 hPa geopotential over the Australian region. The magnitude of the improvement in the Australian region was around three points in AC or eight hours at five days near the surface (MSLP); in addition, there was clear improvement at standard levels through the troposphere. As expected, the temperature fields, for example the 200 hPa temperature field (not plotted), show consistent improvements. The degree of improvement in this study, for example for a five-day forecast at 500 hPa over the southern hemisphere, from using GPS radio occultation data is consistent with that shown in Cucurull and Derber (2008), although their definition of the southern hemisphere annulus is a little different. Generally, the magnitude of the impacts from GPS radio occultation data shows their potential for improving forecasts in the Australian region, and overall their use in other forecast offices has shown them to be of significant value for NWP.

Summary and conclusions

Radio occultation data provided by the COSMIC, MetOp and GRACE satellites have been used to limb sound the atmosphere. These observations have allowed high vertical

resolution sounding of an accuracy which makes them beneficial for NWP and of particular value in the Australian region which is in the data-sparse southern hemisphere. Here we have assimilated refractivity data produced from the GPS radio occultation observations by the COSMIC Data Analysis and Archive Center at UCAR, the GRAS SAF and GFZ and this has resulted in an improvement in forecast

Fig. 3 RMS errors and anomaly correlations for ACCESS-G MSLP forecasts to five days, for the Australian region (left) and the southern hemisphere annulus: 60°S-20°S, 0°E-360°E (right). Shown are results for Control (black), and with GPS RO data (red).

Fig. 4 RMS errors and anomaly correlations for the Aus-tralian region for ACCESS-G 500 hPa geopotential height forecasts to five days (left) and for 200 hPa geopotential height forecasts (right). Control (black), with GPS RO data (red).

Le Marshall et al.: Impact of radio occultation obsrvations on forecasts 125

accuracy in the lower, middle and upper troposphere. These data are unaffected by clouds and precipitation, do not show instrumental drift and do not require calibration. Their use is well suited to NWP and climate studies for a number of reasons. The improved vertical resolution of the radio occultation data near the tropopause complements the advanced sounder observations in that region, while for example the higher horizontal resolution of the advanced sounders assists in the effective use of radio occultation data. In addition their unbiased nature between the mid-troposphere and the lower stratosphere allows them to potentially improve the bias tuning/calibration schemes used in numerical weather prediction and in climate reanalysis, and in conjunction with their direct use, to potentially reduce bias from the analysis process. In effect it is anticipated the unbiased nature of these observations will improve the absolute accuracy of reanalyses and also assist in the vertical resolution of features such as the tropopause. Both these aspects are of importance in climate studies. Currently a follow-on mission to the COSMIC/FORMOSAT-3 mission is under consideration and several other GNSS (Global Navigation Satellite System) related systems are being prepared to provide occultation data. Overall it would appear Australian meteorological, oceanographic, climate and environmental monitoring and prediction will clearly benefit from such missions. In relation to these endeavours, activity is underway in Australia which may lead to Australia making a contribution to a GNSS mission. In conclusion, GNSS radio occultation data have established themselves as an important component of a modern data assimilation system database. In the southern hemisphere, and in particular over the Australian region, their contribution to Australian meteorological oceanographic and climate monitoring will be considerable. In relation to operational use in Australia, currently the NMOC operational analysis and forecast suite is being moved to a new supercomputing and data-handling environment and use of this new data form will follow the establishment of this new system.

Acknowledgments

Many thanks are due to T. Skinner for his assistance in preparing the manuscript.

ReferencesAnthes, R.A., Bernhardt, P.A., Chen, Y., Cucurull, L., Dymond, K.F., Ec-

tor, D., Healy, S.B., Ho, S.-P., Hunt, D.C., Kuo, Y.-H., Liu, H., Manning, K., McCormick, C., Meehan, T.K., Randel, W.J., Rocken, C., Schreiner, W.S., Sokolovskiy, S.V., Syndergaard, S., Thompson, D.C., Trenberth, K.E., Wee, T.-K., Yen, N.L. and Zeng, Z. 2008. The COSMIC/FORMO-SAT-3 Mission: early results. Bull. Am. met. Soc, 89, 313-33.

Bevis, M., Businger, S., Chiswell, S., Herring, T.A., Anthes, R.A., Rocken, C. and Ware, R.H. 1994. GPS Meteorology: mapping zenith wet delays onto precipitable water. Jnl appl. Met, 33, 3379-86.

Brassington, G.B., Pugh, T., Spillman, C., Schulz, E., Beggs, H., Schiller, A. and Oke, P.R. 2007. BLUElink- development of operational oceanogra-phy and servicing in Australia. J. Res. Practice in Information Technol-ogy, 39, 151-64.

Cucurull, L., Derber, J.C., Treadon, R. and Purser, R.J. 2007. Assimilation of Global Positioning System radio occultation observations into NCEP’s Global Data Assimilation System. Mon. Weath. Rev., 135, 3174-93.

Cucurull, L. and Derber, J.C. 2008. Operational implementation of COS-MIC observations into the NCEP’s Global Data Assimilation System. Weath. Forecasting, 23, 702-11.

Cucurull, L., Derber, J.C., Treadon, R. and Purser, R. J. 2008. Preliminary impact studies using Global Positioning System radio occultation pro-files at NCEP. Mon. Weath. Rev., 136, 6, 1865-77.

Danish Meteorological Institute (DMI), European Centre for Medium-Range Weather Forecasts (ECMWF), Institut d’Estudis Espacials de Catalunya (IEEC), Met Office (MetO) 2008. GRAS Satellite Application Facility. And WMO FM94, (BUFR). Specification for GRAS SAF Pro-cessed Radio Occultation Data.

Fjeldbo, G., Kliore, A. and Eshleman, R. 1971. The neutral atmosphere of Venus as studied with the Mariner-5 radio occultation experiments. Astron. J., 76, 123–40.

Healy, S.B. and Eyre, J.R. 2000. Retrieving temperature, water vapour and surface pressure information from refractive index profiles derived by radio occultation: a simulation study, Q. Jl R. Met. Soc.,126, 1661–83.

Kursinski, E.R., Hajj, G.A., Bertiger, W.I., Leroy, S.S., Meehan, T.K., Ro-mans, L.J., Schofield, J.T., McCleese, D.J., Melbourne, W.G., Thornton, C.L., Yunck, T.P., Eyre, J.R. and Nagatani, R.N. 1996. Initial results of radio occultation observations of Earth’s atmosphere using the Global Positioning System. Science, 271, 1107–10.

Kursinski, E.R., Hajj, G.A., Schofield, J.T., Linfield, R.P. and Hardy, K.R. 1997. Observing Earth’s atmosphere with radio occultation measure-ments using the Global Positioning System. J. geophys. Res., 102, 23 429-65.

Kursinski, E.R., Healy, S. and Romans, L. 2000. Initial results of combining GPS occultations with ECMWF global analyses within a DVAR frame-work. Earth Planet. Space, 52, 885-92.

Leroy, S.S. 1977. Measurement of geopotential heights by GPS radio oc-cultation. J. geophys. Res., 102, 6971-86.

Melbourne, W.G., Davis, E.S., Duncan, C.B., Hajj, G.A., Hardy, K.R., Kur-sinski, E.R., Meehan, T.K., Young, L.E. and Yunck, T.P. 1994. The ap-plication of spaceborne GPS to atmospheric limb sounding and global change monitoring. JPL Publ. 94-18, Pasadena, California, April.

Phinney, R.A. and Anderson, D.L. 1968. On the radio occultation method for studying planetary atmospheres. J. geophys. Res., 73, 1819–27.

Rocken, C., Anthes, R., Exner, M., Hunt, D., Sokolovskiy, S., Ware, R., Gor-bunov, M., Schreiner, W., Feng, D., Herman, B., Kuo, Y. and Zou, X. 1997. Analysis and validation of GPS/MET data in the neutral atmo-sphere. J. geophys. Res., 102, 29849-66.

Vorob’ev, V.V. and Krasil’nikova, T.G. 1994. Estimation of the accuracy of the atmospheric refractive index recovery from Doppler shift mea-surements at frequencies used in the NAVSTAR system. Phys. Atmos. Oceans, 29, 602–609.

Ware, R., Exner, M., Feng, D., Gorbunov, M., Hardy, K., Herman, B., Kuo, Y., Meehan, T., Melbourne, W., Rocken, C., Schreiner, W., Sokolovskiy, S., Solheim, F., Zou, X., Anthes, R., Businger, S. and Trenberth, K. 1996. GPS sounding of the atmosphere from Low Earth Orbit: Preliminary Results. Bull. Am. met. Soc., 77, 19–40.

Wickert, J., Reigber, C., Beyerle, G., König, R., Marquardt, C., Schmidt, T., Grunwaldt, L., Galas, R., Meehan, T.K., Melbourne, W.G. and Hocke, K. 2001. Atmosphere sounding by GPS radio occultation: First results from CHAMP. Geophys. Res. Lett., 28, 3263–6.