An Investigation of GNSS Radio Occultation
Atmospheric Sounding Technique for
Australian Meteorology
A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy
Erjiang Fu B.S.
School of Mathematical and Geospatial Sciences College of Science, Engineering and Health RMIT University August 2011
I (This page intentionally left blank)
II Declaration
I certify that except where due acknowledgement has been made, the work is that of the author alone; the work has not been submitted previously, in whole or in part, to qualify for any other academic award; the content of the thesis is the result of work which has been carried out since the official commencement date of the approved research program; any editorial work, paid or unpaid, carried out by a third party is acknowledged; and, ethics procedures and guidelines have been followed.
Erjiang Fu
I (This page intentionally left blank)
II Acknowledgements
First of all, I would like to thank my supervisors: Professor Kefei Zhang, Dr David Silcock and Dr Yuriy Kuleshov. This thesis would not have been possible without their inspiration, guidance, support and encouragement in the past four years. Especially, my main supervisor Professor Zhang, also one of my close friends, who opened the door of a scientific career for me.
During the research journey, it was my greatest honour to work with many respectable scientists and researchers. Sincere thanks go to Professor Yuei-An Liou, Professor Alexander Pavelyev, Dr Xiaolan Wang, Dr Kaye Marion, Dr Suqin Wu, Dr Chuan-Sheng Wang, Dr Robert Norman, Dr Gang-Jun Liu, Dr Falin Wu, Dr Xiaohua Xu, Dr Sue Choy, Dr Xinjia Zhou, Dr Yan Wang, Dr Dongju Peng, Dr Ming Zhu, Mr Bobby Wong, Mr Toby Manning, Miss Ying Li, Ms Jing Peng, Mr Stuart Whitman, Mr Alex Chen, Miss Gap Vimvipa, and others.
I would also like to thank the institutions that have given me access to their data and software systems, which are critical to this research. These institutions include the Australian Bureau of Meteorology, Constellation Observing System for Meteorology Ionosphere and Climate Office at UCAR (University Corporation for Atmospheric Research USA), Global Environmental and Earth Science Information System at NASA (National Aeronautics and Space Administration USA), University of Graz, Reference Antarctic Data for Environmental Research, and European GRAS Meteorology.
This PhD program is financially supported by RMIT Platform Research Institute. The research is also partially supported by a number of funded research projects endorsed to my supervisors Prof Kefei Zhang and Dr Yuriy Kuleshov:
• Australian Space Innovation and Research Project: Platform technologies for space, atmosphere and climate ($2.8 million, 2010 - 2013) • Australia International Science Linkage Program (China Special Fund: CH080253): GPS radio occultation data processing and assimilation system for weather forecast (2009)
III • Australia International Science Linkage Program Competitive Grant Scheme with USA (CG130127): Assimilation of GPS radio occultation data with numerical weather prediction system for climate monitoring (2008 - 2010) • Australia Research Council Linkage Program (LP0883288): Satellite-based radio occultation for atmospheric sounding, weather forecasting and climate monitoring in the Australian region ($1.6 million, 2008 - 2010) • Australian Bureau of Meteorology Strategic Investment Fund: Investigation of atmosphere and climate based on GPS, Galileo and LEO-LEO radio occultation in Australia (2007 - 2008)
This research work is dedicated to my beloved parents (Weiguo Fu and Qiaolin Wang), my lovely wife Yanxi Zhou, and my relatives and friends for their understanding and full support that have kept me moving forward.
IV Abstract
The Earth’s atmosphere is critical to the environment and to life on Earth. Studying the atmosphere plays an important role in many scientific areas and in particular meteorology. However, observing the atmosphere is a challenging task due to its dynamic nature and complexity of its processes. Current station-based observations and satellite remote sensing techniques have different limitations (e.g. limited coverage, and/or limited resolution). Better observation techniques are desirable to capture the detailed status and processes of the atmosphere properly.
Recent developments of the Global Navigation Satellite Systems (GNSS) radio occultation (RO) technique have offered an exciting potential for meteorological research. This emerging technique, using radio signals between the Low Earth Orbit (LEO) and GNSS satellites, probes the Earth's atmosphere and ionosphere from space. Through a number of experimental missions and related studies since 1995, GNSS RO has demonstrated many advantages over other atmospheric observation methods. The information delivered by GNSS RO missions has advanced many meteorological and climate related studies as well as applications, such as numerical weather prediction analysis, tropopause studies, ionosphere and space weather conditions.
The main aim of this thesis is to investigate the key theoretical and practical interests of the Australian meteorological community using GNSS RO. A comprehensive literature review is conducted to explore the fundamentals, the developments and challenges of the GNSS RO. A review of past and present GNSS RO missions confirms the potential for Australian meteorology with better observations at a lower cost. The next generation GNSS RO missions, with the advantage of the multiple systems, will have significant improvement in both quantity and quality of observations. A mission simulation study is undertaken to examine Australia’s opportunities from these future missions. In order to address the challenges of real-time applications, an investigation is also conducted into the system architecture of the satellite data centre for the mega data storage and processing. These studies into the GNSS RO missions and the data processing provide a good benchmark for the planning of future international and Australian missions.
V Detailed statistic analyses are carried out to assess the atmospheric profiles derived from the Global Positioning System (GPS) RO against other atmospheric measurements and modelled data. Four studies apply suitable statistics and data warehouse techniques on the large volume meteorological database to extract the error characteristics of the new atmospheric observation data sources. The results, spatial and temporal error characteristics of the new datasets, can assist meteorological researchers incorporating the RO technique into the existing meteorological systems. The impact of the co-location criteria between the resulting datasets is also examined.
Climatology will benefit significantly from the high accuracy, high resolution and consistent information obtained from the GNSS RO technique. This thesis presents two studies in climatological applications using GPS RO data. One study is undertaken to examine the temperature trends over the Antarctic using both radiosonde and CHAMP RO data. Significant lower-stratospheric cooling and tropospheric warming have been found over the Antarctic for the past five decades (1956 - 2010) using Australian radiosonde data. Temperature trends are also derived using seven years of CHAMP RO data and compared with the data from collocated radiosonde stations. Examining the tropopause structure requires detailed vertical atmospheric profile information and this is difficult in some situations. The GPS RO technique can retrieve high vertical resolution atmospheric profiles with a uniformed global coverage. The other study is designed to investigate the tropopause over the Australasian region using GPS FORMOSAT-3/COSMIC RO temperature profiles. The FORMOSAT-3/COSMIC RO data provide the best vertical resolution of upper air temperature profiles. The study results confirm that the FORMOSAT-3/COSMIC RO temperature profiles show more details of the atmosphere and the tropopause characteristics than other observation methods.
VI Table of Contents
Chapter 1 Introduction...... 1
1.1 Background ...... 1
1.2 Research motivation and objectives...... 4
1.3 Research methodology and contributions...... 6
1.4 Thesis outline ...... 8
Chapter 2 GNSS and Earth Atmospheric Observations...... 11
2.1 Introduction...... 11
2.2 Earth atmospheric observations ...... 11
2.2.1 Radiosonde observations...... 12
2.2.2 Space-based observations...... 14
2.3 GNSS and atmosphere ...... 17
2.3.1 Fundamentals of satellite positioning technique ...... 17
2.3.2 GNSS signals and the atmospheric effects...... 19
2.4 GNSS-based atmospheric observations ...... 23
2.4.1 Ground-based GPS water vapour observations...... 23
2.4.2 Space-based GNSS RO observations...... 24
2.5 Summary ...... 24
Chapter 3 The GNSS RO Technique...... 25
3.1 Introduction...... 25
3.2 GNSS RO concept and atmospheric retrieval theory...... 25
3.2.1 GNSS RO concept...... 25
VII 3.2.2 GNSS RO retrieving theory and error sources ...... 27
3.3 Meteorological and climatic applications...... 31
3.3.1 Advantages of the GNSS RO technique...... 31
3.3.2 Contributions of the GNSS RO to meteorology and climate...... 34
3.4 Summary...... 37
Chapter 4 GNSS RO Missions – The Past, Present and Future...... 39
4.1 Introduction ...... 39
4.2 Review of the GNSS RO missions...... 40
4.3 Simulation study of future RO missions in the Australasian region ...... 43
4.3.1 Next generation GNSS...... 43
4.3.2 RO event simulation study with next generation multiple GNSS ...... 46
4.4 Study of the RO data processing ...... 49
4.4.1 Introduction of the RO data processing ...... 50
4.4.2 RO data processing centre ...... 51
4.5 Summary...... 53
Chapter 5 Evaluation of the GNSS RO Retrievals for the Australasian and the Antarctic Regions ...... 55
5.1 Introduction ...... 55
5.2 Datasets used ...... 56
5.3 Evaluation studies – results and discussions ...... 60
5.3.1 Evaluating CHAMP retrievals by using NCEP reanalysis ...... 60
5.3.2 Evaluation of FORMOSAT-3/COSMIC retrievals by using radiosonde ...... 67
VIII 5.3.3 Spatial analysis of FORMOSAT-3/COSMIC evaluation results ...... 71
5.3.4 Study of the co-location criteria ...... 76
5.4 Summary ...... 81
Chapter 6 Antarctic Temperature Trend and Australasian Tropopause Studies...... 83
6.1 Introduction...... 83
6.2 Antarctic temperature change analysis...... 84
6.2.1 The Antarctic...... 84
6.2.2 Temperature change study using radiosonde data...... 85
6.2.3 Study datasets...... 87
6.2.4 Temperature trends analysis...... 87
6.2.5 Homogeneity study of temperature time series...... 90
6.2.6 Temperature trends using CHAMP RO data...... 99
6.3 Tropopause study using FORMOSAT-3/COSMIC RO data...... 107
6.3.1 The tropopause ...... 107
6.3.2 Tropopause structure determination using RO data ...... 108
6.4 Summary ...... 111
Chapter 7 Conclusions and Recommendations ...... 113
7.1 Summary ...... 113
7.2 Conclusions...... 114
7.3 Suggestions and recommendations ...... 116
References ...... 119
IX (This page intentionally left blank)
X List of Figures
Figure 2-1 Distribution of WMO radiosonde stations...... 13
Figure 2-2 Chronological development of key meteorological satellites...... 15
Figure 2-3 Earth's atmosphere structure, layers and parameters (i.e. temperature, pressure and electron density in ionosphere)...... 22
Figure 3-1 Concept map of GNSS RO technique ...... 26
Figure 3-2 A typical GNSS RO atmospheric retrieval process...... 26
Figure 3-3 GNSS RO retrieval processes...... 27
Figure 3-4 GNSS RO geometric representation...... 28
Figure 4-1 Timeline and life span of GNSS RO missions ...... 41
Figure 4-2 QZSS orbit (Japan Aerospace Exploration Agency, 2009)...... 45
Figure 4-3 Distribution of FORMOSAT-3/COSMIC RO events in a period of six hours (3pm - 9pm) on 23rd August 2010 (retrieved from FORMOSAT-3/COSMIC UCAR website http://cosmic-io.cosmic.ucar.edu/cdaac/index.html on 23rd August 2010)...... 46
Figure 4-4 Distribution of simulated daily FORMOSAT-3/COSMIC RO events in the Australasian region with GPS only and the 38 Australian radiosonde stations...... 47
Figure 4-5 Distribution of simulated daily FORMOSAT-3/COSMIC RO events in the Australasian region with GPS, Galileo and GLONASS ...... 48
Figure 4-6 Distribution of simulated daily FORMOSAT-7/COSMIC-2 RO events in the Australasian region with GPS, Galileo, GLONASS and QZS-1 ...... 48
Figure 4-7 Simulated number of daily RO events from FORMOSAT-3/COSMIC and FORMOSAT-7/COSMIC-2 with future GNSS in the Australasian region at different latitude zones ...... 49
Figure 4-8 RO data processing system structure...... 53
Figure 5-1 Locations of the 38 Australian radiosonde stations with their WMO ID.....57
XI Figure 5-2 Database system developed for the GNSS RO data management and processing ...... 59
Figure 5-3 Distribution, event number and date of the selected radio occultation events ...... 61
Figure 5-4 Refractivity profiles from the CHAMP RO events #4 and #9 against altitudes (compared with their collocated NCEP profiles) are shown in (a) and (c) respectively; and the differences between the two datasets are shown in (b) and (d)...... 62
Figure 5-5 Temperature profiles from the CHAMP RO events #8 and #9 (with their collocated NCEP profiles) are shown in (a) and (c) against altitude respectively; and the differences between the two datasets are shown in (b) and (d) respectively...... 63
Figure 5-6 Pressure profiles from the CHAMP RO events #3 and #9 against altitude (with their collocated NCEP profiles) are shown in (a) and (c) respectively; and the differences between the two datasets are shown in (b) and (d) respectively...... 64
Figure 5-7 Water vapour pressure profiles from the CHAMP RO events #4 and #5 against altitudes (with their collocated NCEP profiles) are shown in (a) and (c) respectively; and the differences between the two datasets are shown in (b) and (d) ...... 64
Figure 5-8 The mean differences of the four atmospheric parameters (i.e. refractivity, temperature, pressure and water vapour pressure) between CHAMP and NCEP. The error bars present the maximum and minimum differences between the two datasets...... 66
Figure 5-9 Distributions of the radiosonde stations (red triangles) and FORMOSAT-3/COSMIC RO events (green dots) for a period of three months from January 1 to March 31 2007...... 67
Figure 5-10 Mean differences between FORMOSAT-3/COSMIC and radiosonde in temperature (upper plot) and pressure (lower plot) of the 42 study pairs against their latitudes ...... 68
XII Figure 5-11 Differences in temperature (upper plot) and pressure (lower plot) between FORMOSAT-3/COSMIC and radiosonde against their altitudes and in the two plots, the green lines in the middle of each plot represent the mean differences and the blue and red lines are the 95% confidence intervals...... 69
Figure 5-12 Temperature differences between radiosonde measurements and FORMOSAT-3/COSMIC derived values ...... 70
Figure 5-13 Pressure differences between radiosonde measurements and FORMOSAT-3/COSMIC derived values ...... 71
Figure 5-14 Distributions of the 35 radiosonde stations (in red dots), 64,599 FORMOSAT-3/COSMIC derived atmospheric profiles (in dark dots) and 204 collocated pairs (in brown stars) for the period of 1st July 2006 to 31st July 2007...... 72
Figure 5-15 Mean differences in temperature (and standard deviations) between FORMOSAT-3/COSMIC and radiosonde and the standard deviations of the differences are represented as density maps...... 74
Figure 5-16 Means of the differences in pressure (and standard deviations) between FORMOSAT-3/COSMIC and radiosonde and the standard deviations of the differences are represented as density maps...... 74
Figure 5-17 Mean differences in temperature (and standard deviations) against the altitude from surface to 30 km with their standard deviations...... 75
Figure 5-18 Mean differences in pressure (and standard deviations) against the altitude from surface to 30 km with their standard deviations...... 76
Figure 5-19 CHAMP temperature profile comparison results (i.e. means, 95% confident levels and counts of comparison pairs divided by 1000) at standard pressure levels ...... 80
Figure 5-20 FORMOSAT-3/COSMIC temperature profile comparison results (i.e. means, 95% confident levels and counts of comparison pairs divided by 10000) at standard pressure levels...... 80
Figure 6-1 Radiosonde temperature trend analysis over three Australian stations in Antarctic ...... 88
XIII Figure 6-2 Radiosonde temperature trend analysis by using 18 Antarctic radiosonde observations at the 100 hPa pressure level...... 89
Figure 6-3 Radiosonde temperature trend analysis by using 18 Antarctic radiosonde observations at the 500 hPa pressure level...... 89
Figure 6-4 De-seasonalised series (i.e. anomalies from the annual cycle) of monthly mean atmospheric temperatures at the 500 hPa pressure level, recorded at the stations, and the corresponding multi-phase regression fits (red lines, which include the annual trends) ...... 92
Figure 6-5 De-seasonalised series (i.e. anomalies from the annual cycle) of monthly mean atmospheric temperatures at the 100 hPa pressure level, recorded at the stations, and the corresponding multi-phase regression fits (red lines, which include the annual trends) ...... 93
Figure 6-6 The mean vertical profile of the annual temperature trends (°C per decade) and their standard deviations (STD) at nine atmospheric pressure levels. These are derived from the HadAT2 (homogenised) radiosonde data from the six longer-term Antarctic stations ...... 95
Figure 6-7 The mean vertical profile of seasonal temperature trends (°C per decade) and their standard deviations (STD) at nine atmospheric pressure levels derived from the HadAT2 (homogenised) radiosonde data from the six longer-term Antarctic stations...... 96
Figure 6-8 Location of seven Antarctic stations with the 800 km Equal Area Scalable Earth Grids superimposed over a map of the continent (45 grids in total) and 500 km radius circles centred on each station ...... 100
Figure 6-9 Comparisons of i n-situ 100 hPa radiosonde data (black) and the corresponding CHAMP data (blue) at the stations indicated ...... 101
Figure 6-10 Comparison of i n-situ 500 hPa radiosonde data (black) and the corresponding CHAMP data (blue) at the indicated stations ...... 102
Figure 6-11 Comparison of linear trends estimated from in-situ radiosonde (RS) data and the corresponding CHAMP data for the period from September 2001 to August 2008...... 103
XIV Figure 6-12 Temperature trends estimated from CHAMP RO data for the period from September 2001 to August 2008 at pressure levels 400, 500, 600 and 750 hPa...... 105
Figure 6-13 Temperature trends estimated from CHAMP RO data for the period from September 2001 to August 2008 at pressure levels 50, 100, 200 and 300 hPa...... 106
Figure 6-14 Sample of temperature profiles obtained from radiosonde and FORMOSAT-3/COSMIC RO respectively...... 109
Figure 6-15 Average tropopause heights at different latitude zones and time periods (day: 1-30; 90-120, 180-210 and 270-300) derived from FORMOSAT-3/COSMIC RO...... 109
Figure 6-16 Annual average tropopause heights derived from FORMOSAT-3/COSMIC RO data in the Australasian region - 2009...... 110
Figure 6-17 Annual average tropopause heights derived from FORMOSAT-3/COSMIC RO data in the Australasian region - 2008...... 110
Figure 6-18 Annual average tropopause height derived from FORMOSAT-3/COSMIC RO data in the Australasian region - 2007...... 111
XV (This page intentionally left blank)
XVI List of Tables
Table 2-1 Indicative accuracy of the radiosonde measurements...... 13
Table 2-2 Meteorological satellites and their countries, orbits and observed information ...... 16
Table 2-3 GPS carrier signals, and code signals and their descriptions (Hofmann-Wellenhof et al., 2008) ...... 20
Table 3-1 Error sources, their characteristics and key properties ...... 30
Table 3-2 A comparison of three atmospheric observation techniques ...... 34
Table 4-1 A summary table of the four major GNSS (i.e. GPS, GLONASS, Galileo and Compass) and regional system (i.e. QZSS) ...... 45
Table 5-1 A list of 38 Australian radiosonde stations with their WMO station IDs, names and elevations...... 58
Table 5-2 Statistical differences of atmospheric parameters between NCEP and CHAMP profiles...... 65
Table 5-3 The means and standard deviations of the differences (in temperature and pressure) between FORMOSAT-3/COSMIC and radiosonde...... 73
Table 5-4 Means, standard deviations (STD) and the number of pairs (# Sample) of the temperature (ºC) discrepancies between radiosonde and RO data using different co-location criteria...... 77
Table 5-5 Analysis of variance results (i.e. sum of squares (SS), degrees of freedom (DF), mean squares (MSS), F and P-value) for temperature (ºC) discrepancies between radiosonde and CHAMP, and between radiosonde and FORMOSAT-3/COSMIC data...... 78
Table 6-1 Times of discontinuities (yyyymm) detected in the monthly mean radiosonde temperature series (H indicates that the time series was found to be homogeneous at 5% significance level) ...... 91
XVII (This page intentionally left blank)
XVIII Acronyms
3/4D: 3/4 Dimensions ATS: Applications Technology Satellites ASCII: American Standard Code for Information Interchange BoM: Australian Bureau of Meteorology C/A code: Coarse/Acquisition code CHAMP: CHAllenging Minisatellite Payload CORS: Continuously Operating Reference Station COSMIC: Constellation Observing System for Meteorology, Ionosphere & Climate CSV: Comma-Separated Values DJF: December, January and February ECMWF: European Centre for Medium -Range Weather Forecasts EGOPS: End-to-end GNSS Occultation Performance Simulator EUMETSAT: European Organisation for the Exploitation of Meteorological Satellites ESA: European Space Agency ESSA: Environmental Science Services Administration Satellite Program FORMOSAT: Formosa Satellite Mission GCOS: Global Climate Observing System GENESIS: Global Environmental & Earth Science Information System GFZ: GeoForschungsZentrum (German Research Centre for Geosciences) GLONASS: GLObal Navigation Satellite System GMS: Geosychronous Meteorological Satellite GNSS: Global Navigation Satellite System GOME: Global Ozone Monitoring Experiment GPS: Global Positioning System GPS/MET: Global Positioning System/Meteorology GRACE: Gravity Recovery and Climate Experiment GUAN: GCOS Upper-Air Network ICT: Information and Communications Technology ID: Identification IDL: Interactive Data Language JJA: June, July and August JPL: Jet Propulsion Laboratory
XIX KOMPSAT: Korea Multi-Purpose Satellites L2C: Civil code on L2 L2CL: Long-length code on L2C L2CM: Moderate-length code on L2C L5C: Civil code on L5 L5I: In-phase code on L5 L5Q: Quadraphase code on L5 LEO: Low Earth Orbit LS: Lower Stratosphere M code: Military code MAM: March, April and May MSU: Microwave Sounding Unit NCEP: National Centre for Environmental Prediction NetCDF: Network Common Data Form NNSS: Navy Navigation Satellite System NOAA: National Oceanic and Atmospheric Administration NWP: Numerical Weather Prediction POD: Precise Orbit Determination P Code: Precision Code PRN: Pseudorandom Noise QZS-1: Quasi-Zenith Satellite - 1 QZSS: Quasi-Zenith Satellite System READER: Reference Antarctic Data for Environmental Research RO: Radio Occultation ROE: Radio Occultation Event ROSA: Radio Occultation Sounder for Atmosphere SMS: Synchronous Meteorological Satellite SAC: Satellite de Applicaciones Cientifico SON: September, October and November SQL: Structured Query Language STD: Standard Deviation SUNSAT: Stellenbosch University SATellite TanDEM-X: TerraSAR-X Add-on for Digital Elevation Measurement TEC: Total Electron Content TIROS: Television Infrared Observation Satellite
XX UCAR: University Corporation for Atmospheric Research UK: United Kingdom USSR: Union of Soviet Socialist Republics USA: United States of America UT: Upper Troposphere VB: Visual Basic WMO: Weather Meteorology Organization
XXI (This page intentionally left blank)
XXII Contributions to knowledge
Journal papers
Fu, E., Zhang, K., Wu, F., Xu, X., Marion, K., Rea, A., Kuleshov, Y., and Weymouth, G.: An Evaluation of GNSS Radio Occultation Technology for Australian Meteorology, Journal of Global Positioning Systems, 6, 74-79, 2007 www.gmat.unsw.edu.au/wang/jgps/v6n1/v6n1p08.pdf
Fu, E., Zhang, K., Kuleshov, Y., Silcock, D., Zhou, X., and Zhou, Y.: Studies on simulation of radio occultation (RO) events with multiple GNSS and climatology of tropopause using COSMIC RO data for Australia, Asia-Pacific Journal of Atmospheric Sciences, under review, 2011
Kuleshov, Y., Fu, E., Wang, X., Zhang, K., Marion, K., Zhou, X., Liou, Y.-A., and Pavelyev, A.: Spatial Patterns of Recent Atmospheric Temperature Trends over Antarctica Derived from GPS Radio Occultation Observation Validated with in-situ Radiosonde Data, Geophysical Research Letters, under review, 2011
Kuleshov, Y., Wang, X., Fu, E., Zhang, K., Feng, Y., and Zwiers, F. W.: Significant Lower-Stratospheric Cooling and Tropospheric Warming Observed over Antarctica since 1956, Natural Atmosphere, under review, 2011
Xu, X., Zhang, K., Luo, J., and Fu, E.: Research on the Collocation Criteria in the Statistical Comparisons of GPS Radio Occultation and Radiosonde Observations, Wuhan University Journal of Information Science, 34, 1332-1335, doi:CNKI:SUN:WHCH.0.2009-11-019, 2009 http://www.cnki.net/kcms/detail/detail.aspx?dbCode=cjfd&QueryID=16&CurRec=19&filenam e=WHCH200911019&dbname=CJFD0911 .
Xu, X., Zhang, K., Fu, E., and Luo, J.: Inversion of Atmospheric Profiles with COSMIC Radio Occupation Data over Australia, Wuhan Daxue Xuebao (Xinxi Kexue Ban)/ Geomatics and Information Science of Wuhan University, 33, 800-804, doi:CNKI:SUN:WHCH.0.2008-08-007 2008 http://www.cnki.net/kcms/detail/detail.aspx?dbCode=cjfd&QueryID=5&CurRec=8&filename= WHCH200808007&dbname=CJFD0608 .
XXIII Zhang, K., Fu, E., Silcock, D., Wang, Y., and Kuleshov, Y.: An investigation of the co-location criteria in the evaluation of GPS radio occultation profiles using Australian radiosonde data, Atmospheric Measurement Techniques, 4, 2087-2092, 2011
Reviewed conference papers
Fu, E., Wu, F., Zhang, K., Xu, X., Rea, A., Kuleshov, Y., and Biadeglgne, B.: Validation of GNSS Radio Occultations' Performance Using NCEP Data in Australia, Proceedings of International Global Navigation Satellite Systems 2007, Sydney, Australia, 4-6 December 2007, 10 http://www.ignss.org/Conferences/PastPapers/2007ConferencePastPapers/2007PeerReviewedP apers/tabid/90/Default.aspx
Fu, E., Zhang, K., and Wu, F.: A Preliminary Investigation of GNSS Radio Occultation and its Applications in Australian Meteorology, Proceedings of 20th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2007), Fort Worth, Texas, United States, 25-28 September 2007, 1777-1781 http://www.ion.org/search/view_abstract.cfm?jp=p&idno=7387
Fu, E.: Assessing Space-based GNSS Technology for Meteorological Studies in Australia, Proceedings of 21st International Technical Meeting of the Satellite Division of the Institute of Navigation, Savannah, Georgia, USA, 19-23 September 2008, 271-276 http://www.ion.org/search/view_abstract.cfm?jp=p&idno=7957
Fu, E.: GNSS Radio Occultation Meteorology Technique and Detailed Evaluation Study Using COSMIC Data for Australia, Proceedings of International Association of Chinese Professional in Global Positioning Systems Forum, Guangzhou, China, 4-7 January 2008, 12
Fu, E., Zhang, K., Marion, K., Xu, X., Marshall, J. L., Rea, A., Weymouth, G., and Kuleshov, Y.: Assessing COSMIC GPS Radio Occultation Derived Atmospheric Parameters using Australian Radiosonde Network Data, Procedia Earth and Planetary Science, 1, 1054-1059, doi:10.1016/j.proeps.2009.09.162, 2009 http://linkinghub.elsevier.com/retrieve/pii/S1878522009001635
Xu, X., Zhang, K., Wu, F., and Fu, E.: The Impact of LEO Satellites' Orbital Parameters on the Pattern of the GPS-LEO Radio Occultation Occurrence, Proceedings of International Global Navigation Satellite Systems 2007, Sydney, Australia, 4-7 December 2007, 13 http://www.ignss.org/Conferences/PastPapers/2007ConferencePastPapers/2007PeerReviewedP apers/tabid/90/Default.aspx
XXIV Zhang, K., Biadeglgne, B., Wu, F., Kuleshov, Y., Rea, A., Hodet, G. D., and Fu, E.: A Comparison of Atmospheric Temperature and Moisture Profiles Derived from GPS Radio Occultation and Radiosonde in Australia, Proceedings of The 3rd Workshop for Space, Aeronautical and Navigational Electronics, Perth, Australia, 15-18 April 2007
Zhang, K., Fu, E., Wu, F., Xu, X., Rea, A., Kuleshov, Y., and Biadeglgne, B.: GNSS Radio Occultation for Weather and Climate Research - A Case Study in Australia, Proceedings of International Global Navigation Satellite Systems 2007, Sydney, Australia, 4-6 November 2007, 9 http://www.ignss.org/Conferences/PastPapers/2007ConferencePastPapers/2007PeerReviewedP apers/tabid/90/Default.aspx
Zhang, K., Wu, F., Fu, E., Xu, X., Rea, A., Weymouth, G., Biadeglgne, B., and Kuleshov, Y.: Validations of GNSS Radio Occultation Technique over Australia, Proceedings of International Symposium on GPS/GNSS 2008, Tokyo, Japan, 563-571 http://www.gnss-pnt.org/symposium2008/abstract/oral/D14p/7-396-a.pdf
Zhang, K., Fu, E., Liou, Y.-A., Marshall, J. L., and Kuleshov, Y.: GNSS Radio Occultation Technique and Detailed Evaluation Study Using FORMOSAT-3 Data for Australia, Proceedings of The 28th Surveying and Geoinformatics Symposium, Taipei, Taiwan, 27-28 August 2009, 15
Zhang, K., Fu, E., Xu, X., Liou, Y.-A., Marshall, J. L., and Kuleshov, Y.: A Study of Collocation Criteria of GNSS COSMIC Radio Occultation and Radiosonde Comparisons, Proceedings of IGNSS Symposium 2009, Surfers Paradise, Australia, 1-3 December 2009, 9 http://www.ignss.org/Conferences/PastPapers/2009ConferencePastPapers/2009PeerReviewedP apers/tabid/86/Default.aspx
Zhang, K., Marshall, J. L., Norman, R., Wang, C.-S., Fu, E., Li, Y., and Kuleshov, Y.: A Study on the Relationship Between Ionospheric Correction and Data Control for GPS Radio Occultation in Australia, Proceedings of International Geoscience and Remote Sensing Symposium (IGARSS), Honolulu, Hawaii, USA, 25-28 July 2010, 2940-2943, doi:10.1109/IGARSS.2010.5650216 http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=5650216
XXV
Chapter 1 Introduction
1.1 Background
Climate change is one of the most critical issues faced by the international communities. Studies have shown that global warming has been a concerning trend over past decades and is likely to continue in the future (Cohen and Waddell, 2009; IPCC, 2007; Karoly and Braganza, 2005; Steig et al., 2009). The increased temperature has impacted the Earth’s environment and many aspects of human society, such as ice melting observed in the Polar and mountain regions, and frequent occurrence of extreme weather and disasters around the world (European Environment Agency, 2010; Garnaut, 2008; Government Accountability Office., 2008; Stern, 2007). The 2000s has been reported as the warmest decade in Australian history (Australian Bureau of Meteorology, 2010a). Studies have also shown an increase in the Australian sea level and a decrease in annual rainfall over the same period (Australian Bureau of Meteorology, 2010b). A change in rainfall patterns, a higher frequency of extreme weather events and a more variable climate in Australia are expected in the future. In January and February 2011, Queensland was seriously damaged by severe floods and tropical cyclone Yasi; Victoria also had floods across many parts of the state; Sydney has experienced a long period of heat waves; and Western Australia experienced both bushfires and floods. Australia, as a dry country, experiences more damages and economic loss because of climate change than other countries (Garnaut, 2008). Therefore, studies of extreme weather, climate change and their relationships become more and more important to Australia (Beeton et al., 2006).
Providing an insight into climate change and predicting the weather is a challenging task due to the complexity and dynamics of the Earth’s atmospheric system and the limitations of current observational techniques (Ware et al., 2001; World Meteorological Organization, 2008). Many weather and climate related studies require high quality and high resolution atmospheric information (i.e. temperature, pressure, humidity and wind). Radiosonde observations have been a dominant conventional method for the acquisition of upper air atmospheric information for over 70 years. A
1 global radiosonde network with about 1500 stations has been developed (Kuo et al., 2005). However, radiosonde observational method has a limited coverage and a low spatio-temporal resolution since it is highly constrained by accessibility as well as high operational and maintenance costs (Zhang et al., 2011). For example, the limited number and the poor geographical distribution (mainly along the coast) of the 18 Antarctic weather stations are far from sufficient for many meteorological and climatological studies. This situation is difficult to improve because of the station maintenance issues in the Antarctic. Many regions (e.g. oceans, deserts, mountains and unpopulated areas) create similar problems of limited availability of the ground-based observational network. Satellite remote sensing technologies are also being used more frequently for atmospheric observations. They generally have higher horizontal resolutions and better global coverage when compared with the ground-based methods (Chuvieco, 2008). The major drawback of these remote sensing techniques is the low vertical resolution. Therefore, there is an increasing demand from the meteorological communities for better observational techniques that can capture the atmospheric structure and processes in greater details.
Recent developments of the Global Navigation Satellite Systems (GNSS), such as the Global Positioning System (GPS), have offered exciting potentials for meteorological research (Mannucci et al., 2008). GPS observations from ground-based GPS reference networks have been utilized to generate useful atmospheric water vapour information (Liou et al., 2000; Rocken et al., 2005; Ware et al., 1997). This technique can continuously and automatically monitor the water vapour content in the atmosphere over the network. The GPS radio occultation (RO) technique probes the Earth's atmosphere and ionosphere using GPS receivers onboard Low Earth Orbit (LEO) satellites (Anthes et al., 2000; Gobiet, 2005; Hajj et al., 2000; Hocke and Pavelyev, 2001; Kursinski et al., 1995; Melbourne et al., 1994; Yuan et al., 1993). The first experimental GPS RO mission (i.e. GPS/MET) was launched by UCAR (University Corporation for Atmospheric Research, U.S.) in 1996 and accurate atmospheric profiles were successfully retrieved. Follow on satellite missions such as CHAMP (CHAllenging Minisatellite Payload) and FORMOSAT-3/COSMIC (Constellation Observing System for Meteorology Ionosphere and Climate) also generated high quality and stable information of the atmosphere and ionosphere. GPS-only signals are predominately used by the past and present GNSS RO missions. The next generation
2 RO receivers have been developed to receive multi-GNSS signals such as GPS, GLONASS and Galileo (Meehan, 2009). It should be noted that in this thesis, GNSS RO is used as a general term to address the RO technique even though GPS is the dominant technology used.
A GNSS radio occultation event (ROE) occurs when a GNSS satellite is setting or rising behind the Earth's limb as viewed by a GNSS receiver aboard a LEO satellite. During an ROE the transmitted signals from the GNSS satellite travelling to the LEO satellite are delayed because of the impacts of the atmosphere and ionosphere (Liou et al., 2007). This delay can be precisely measured and converted to a vertical profile of bending angles and consequently processed to derive a refractivity profile by applying a series of transformation steps (Hajj et al., 1994; Kursinski et al., 1995; Rocken et al., 1993; Yuan et al., 1993). The refractivity is a function of the electron density in the ionosphere and the temperature, pressure, and water vapour in the natural atmosphere (Healy et al., 2002). Hence, the GNSS RO technique can provide useful information about the structure and dynamics of the ionosphere, stratosphere and troposphere.
The GNSS RO technique has a number of outstanding advantages such as its global coverage, high vertical resolution, 24 hours availability, high accuracy, all weather capability, calibration-free and low operational cost (Kursinski et al., 2000; Ware et al., 1997; Yunck et al., 2000). Many countries, such as the USA, Germany, Austria, the Russian Federation, Finland, Italy, Denmark, Argentina, Brazil and South Africa, have established experimental GNSS RO meteorology projects and launched a number of LEO experimental satellites with GPS RO receivers (GRAS Meteorology SAF, 2001; Hoeg and Kirchengast, 2002; Kirchengast and Hoeg, 2004; Sgrigna et al., 2005; Vespe et al., 2009; Ware et al., 1996). New developments of the GNSS RO technique, such as the design of GNSS receivers and the LEO-LEO technique, will offer unprecedented opportunities to obtain a large amount of high-quality and high-resolution observations of the Earth’s atmosphere (Hoeg and Kirchengast, 2002; Kursinski et al., 2002; Kursinski et al., 2009). The continuous and accurate measurements of the atmospheric profiles with good spatial and temporal resolutions from GNSS RO will be available and new opportunities for better understanding of the Earth’s atmosphere and ionosphere will follow (Anthes et al., 2000). Many meteorological and climate related applications, such as numerical weather prediction analysis, climate studies, ionosphere
3 and space weather conditions, will be advanced with the development of new generation GNSS and LEO satellite programs.
There are 38 operating radiosonde stations for the entire Australian continent and the surrounding ocean areas (Fu, 2008). These stations provide benchmark information of the upper air atmosphere for the Australian region. This supports the majority of meteorological studies and applications in Australia and the world. On the other hand, there are over 4000 atmospheric profiles generated annually over the Australasian region by the CHAMP LEO satellite and six times more profiles from the FORMOSAT-3/COSMIC constellation (Fu et al., 2009b). A simulation study has shown that the COSMIC follow-on mission will deliver over 3000 daily RO events for the Australasian region (Fu et al., 2011). Australian meteorological studies will be significantly enhanced by such a large volume of accurate and detailed four-dimensional information. The economical benefits of this GNSS-based technology can be demonstrated since Australia has a large continent, sparse population and long coastal areas (Kuleshov et al., 2011a; Zhang et al., 2009). The key aim of the PhD study is to investigate the GNSS RO atmospheric observation technique and its challenges and opportunities for the Australian region.
1.2 Research motivation and objectives
Research Motivation
RMIT, in collaboration with the Australian Bureau of Meteorology (BoM), has pioneered the GNSS RO research in Australia since 2006. Early collaboration (Biadeglgne et al., 2007; Xu et al., 2007) has demonstrated the potential of GNSS RO technology for Australia and its abilities to overcome the limitations of the existing atmospheric observational methods. A need for a comprehensive investigation into the new technique was identified so that the emerging technology could be exploited for research in Australia. This is the first dedicated PhD research program on this topic in Australia.
The GNSS RO technique relies on the L-band radio signals as it propagates through the Earth's atmosphere. The fundamental theory behind this technology is straightforward but the physical and mathematical algorithms involved in the retrieval processes are
4 highly complicated (Aragon-Angel et al., 2010; COSMIC Data Analysis and Archive Center, 2008; Engeln and Nedoluha, 2003). A number of steps are involved in order to retrieve useful ionospheric and atmospheric information from the raw GNSS radio signals captured by the space-based RO receivers. Firstly, the ray path between the radio transmitters (i.e. GNSS satellites) and receivers (i.e. GNSS RO receivers onboard LEO) needs to be calculated accurately. It is necessary to precisely determine the orbit and velocity of the transmitter GNSS satellites and the receiver LEO satellites. Noise to signals ratio must also be mitigated especially when the signal propagates through the dense lower troposphere. Derivation of the atmospheric profiles (i.e. temperature, pressure and water vapour) from atmosphere refractivity in the lower troposphere presents a major challenge due to the difficulty in separating the wet and dry atmospheric impacts on the GNSS signals. The retrieval process needs to be carefully investigated to ensure the quality of the retrievals and to provide a better understanding of the data products.
Many countries and regions have launched a series of RO satellites and processed the data at their data processing centres. Scientists from these countries have dominated studies in this field and as such these countries have benefited most from this new technique. Unfortunately, Australia, one of the key players in Southern Hemisphere and the world meteorological research, has not been officially involved in any of these RO satellite programs. Studies of the RO technique are important for Australian researchers so that they can utilise the technique and contribute to the global meteorological community. A dedicated GNSS RO observational system, through future Australian owned or international collaborative missions, can enhance meteorological research and operations for Australia using the high quality and stable RO atmospheric observations.
Research objectives
A number of research objectives were defined as a result of various team discussions between GNSS researchers from RMIT, meteorological scientists from the Australian BoM and international experts in the field of GNSS RO. The objectives of this research are to:
1. Investigate key components of the GNSS RO atmospheric observation technique including the theory and data retrieval processes.
5 2. Study the development of GNSS RO missions and impact of multiple GNSS for meteorological research in the Australian region.
3. Evaluate the error characteristics of GNSS RO retrievals by comparing with meteorological information from both conventional atmospheric observations and numerical weather models.
4. Investigate applications of GNSS RO derived meteorological information for climatological studies (i.e. temperature trend analysis and tropopause structure studies) in the Australasian and the Antarctic regions.
1.3 Research methodology and contributions
This research commenced with a comprehensive literature review that explores the fundamentals and developments of the GNSS RO technology. From the beginning of the project, a reference database covering all aspects of the GNSS RO technique has been developed and maintained. The current database contains over 800 publications including papers, theses, reports and presentations. The literature review has provided a solid base for the research and has helped to refine the objectives.
The availability of data is critical for this project. Data were obtained from various research institutes including UCAR, JPL, the Australian BoM, the UK Met Office, GFZ and EUMETSAT. The datasets used in this study include GNSS RO orbit data, derived atmospheric profiles, radiosonde measurements and numerical weather model data. These datasets are very large in sizes approximately 100 G-Byte, and are in various formats (e.g. ASCII, NetCDF and Binary).
Databases have been developed using both the enterprise database system “Oracle” and the open source system “MySQL”. Models and functions have also been developed in a number of programming languages (i.e. SQL, VB, C++ and IDL) in order to process and analyse these datasets. Through the establishment of the database and software development, the author has gained much understanding of satellite and meteorological data handling and processing. Studies of the GNSS RO data processing systems, the related information and communication technologies (ICT), and the infrastructure are carried out.
6 An orbit simulation study has been carried out to examine the number and distribution of RO events for future RO missions in the Australasian region. This simulation utilised the EGOPS software (developed by the University of Graz) with satellite orbit data. The coverage and horizontal resolution of the GNSS RO technique is dependent on the orbit design of both the GNSS and the LEO satellites. In this study, GPS and other GNSS (i.e. Galileo and GLONASS) and the Japanese regional QZSS (Quasi-Zenith Satellite System), have been examined. The study shows that over 3000 daily ROE can be expected from FORMOSAT-7/COSMIC II with multiple GNSS. The use of these new GNSS will increase not only the number of ROE significantly but also the quality of the retrievals due to increased signals. The result of this study provides reference information for future missions which are expected to benefit Australian atmospheric and ionospheric studies.
Four detailed statistical evaluation studies were carried out to assess the atmospheric retrievals from two GNSS RO missions (i.e. CHAMP and FORMOSAT-3/COSMIC). These were done by comparing the RO missions with other atmospheric records from conventional observations (i.e. radiosonde) plus modelled data. For example, the mean differences in temperature between the radiosonde and RO data is about 0.5 K with a standard deviation of about 1.1 K. The error characteristics of the statistical differences between the GNSS RO data and the compared datasets were determined. The resulting spatial and temporal error characteristics of the new datasets may assist meteorological and weather forecasting researchers implement the new technique in existing operations.
The Antarctic region plays an important role in global weather and climate systems. Changes in the atmospheric temperature over the area have been estimated in a number of studies (Johanson and Fu, 2007; Steig et al., 2009; Turner et al., 2005). CHAMP RO data and in-situ radiosonde observations were used to examine the recent temperature trends in the troposphere and the lower stratosphere over the Antarctic. The CHAMP RO and collocated Antarctic radiosonde data for 2001-2008 were found to be in a good agreement with respect to monthly mean atmospheric temperature values and trends, especially from the lower stratosphere to the upper troposphere (from 50 hPa to 300 hPa pressure levels). The CHAMP RO data provided detailed vertical structure of the temperature trends and demonstrated an overall cooling in the Antarctic lower stratosphere and a warming in the troposphere. These results were in agreement with
7 radiosonde analyses. For example, the lower stratospheric cooling during the 2001-2008 period was strongest at about 50-100 hPa over a large area of the Antarctic (i.e. longitude 60°W-to-0°-to-120°E, including about two thirds of East Antarctica and about half of West Antarctica). Small positive temperature trends were observed over West Antarctica and adjacent to the Amundsen and Bellinshausen Seas at all pressure levels from 100 to 700 hPa. While the suitability of the CHAMP RO data for climate change detection is demonstrated for the lower stratosphere and the upper troposphere, the CHAMP RO and radiosonde data are less consistent in terms of temperature trends in the lower to mid-troposphere. A study of temperature trends over Antarctica using long-term (about 50 years) radiosonde data was also conducted. This research addressed some statistical issues identified from other studies and a new method is applied to the temperature trend analysis. The study result shows that the mean cooling rate in the lower stratosphere is about -0.60°C per decade, and the mean warming rate in the troposphere is about 0.18°C per decade.
The tropopause is a layer of atmosphere which separates the troposphere and the stratosphere and is important for many weather and climatological phenomena. High vertical resolution atmospheric profiles are desired for advanced and detailed tropopause studies. The Australian BoM maintains a network of 38 radiosonde stations which provides the major upper-air information for the Australian continent, surrounding oceans and also the Antarctic (Fu et al. 2009). Most of the stations are located in the coastal regions and there are only a few inland. The vertical resolution of radiosonde data is typically from a few hundreds meters to one kilometre which imposes certain limitations on atmospheric studies, such as the tropopause. FORMOSAT-3/COSMIC RO data, with a much higher vertical resolution and better coverage compared with radiosonde, was used to examine the tropopause structure in the Australasian region for a period of three years (from 2007 to 2009). The study shows that the high resolution and stability of RO data has the capability to serve as a benchmark measurements for many climate studies.
1.4 Thesis outline
This thesis consists of seven chapters. Each of these chapters is briefly described below.
8 Chapter One: introduces the research background, motivation, research objectives, methodology, and the main contributions.
Chapter Two: presents a detailed review of GNSS, Earth atmosphere/meteorology observation techniques, and the GNSS based atmospheric observation techniques.
Chapter Three: examines the fundamentals of the GNSS RO technique, especially the data retrieval processes and the key error sources. The advantages and applications of the technique for weather and climate are also discussed.
Chapter Four: studies the GNSS RO missions, data processing centres, the development of future GNSS RO missions, and discussions on the ICT infrastructure for RO data archive and processing.
Chapter Five: presents four detailed evaluation studies of the CHAMP and FORMOSAT-3/COSMIC RO retrievals using radiosonde and modelled data.
Chapter Six: evaluates GNSS RO data for Antarctic climatological and temperature trend analysis, and tropopause studies for the Australasian region.
Chapter Seven: summaries the contributions of this research and provides recommendations for future studies.
9 (This page intentionally left blank)
10 Chapter 2 GNSS and Earth Atmospheric Observations
2.1 Introduction
Observing meteorological conditions and processes is a challenging task due to the high-uncertainty and high-dynamics nature of the Earth's atmosphere (King, 2007). The GNSS RO atmospheric observation technique has been introduced for acquisition of numerous and comprehensive Earth’s atmospheric information (Liou, 2010; Mannucci et al., 2008; Melbourne et al., 1994). This chapter reviews atmospheric observations including ground-based atmosphere observations and meteorological remote sensing satellites. An overview of the GNSS technology, in particular GPS, is presented. This includes the fundamentals of the GNSS and satellite positioning technique, its signals and atmospheric corrections, and then the future development of GNSS modernization. Finally, an introduction of the new GNSS based meteorological observation technique will be provided.
2.2 Earth atmospheric observations
The Earth's atmosphere is critical to all life systems and plays a unique role in the weather and climate systems. Energy, water and atmospheric elements are exchanged and transferred dynamically in the atmosphere, and also between land/ocean and the atmosphere. All these processes contribute to the weather phenomena and climate change. To study the atmosphere and its processes, meteorologists characterise the atmosphere by temperature, pressure, water vapour, wind, precipitation and atmospheric compositions (World Meteorological Organization, 2009). Observing and monitoring these parameters are fundamental to all meteorological and climatological studies.
Atmospheric information, which supports the study of climate and weather applications, is mainly obtained from observing systems mounted on balloons, aircrafts and satellites (Kramer, 2002). Due to the many advantages of satellite atmospheric observing systems, there is a decrease in conventional meteorological observations (i.e. the radiosonde
11 observations) and an increase in satellite based observations (National Research Council, 2007).
2.2.1 Radiosonde observations
Radiosonde is an observational technique that measures atmospheric profiles of pressure, temperature, and relative humidity by using a set of sensors attached to balloons (Ware et al., 2001). Data obtained by these sensors are sent to the ground receiving stations by a radio transmitter. The atmospheric profiles measured by radiosonde range from the Earth’s surface to around 35 km altitude. However, many measurements are limited to 25 km because of the high operational costs and the poorer accuracies at higher altitudes (Australian Bureau of Meteorology, 2002).
Figure 2-1 shows a network of radiosonde stations that has been developed internationally since the 1930s (World Meteorological Organization, 2008). Station distribution and the sampling frequency of the network are important for meteorological applications, especially for meteorological phenomena identification and prediction (National Oceanic and Atmospheric Administration, 1997). The World Meteorology Organization (WMO) recommends a radiosonde network with a spacing of about 250 km over large areas and 1000 km for unpopulated areas, and with an observation frequency of one to four times (at hour of 00, 06 , 12 and 18 local time) per day (National Oceanic and Atmospheric Administration, 1997). This standard is difficult to satisfy for unpopulated areas such as the central Australian region and the Antarctic.
Table 2-1 presents the atmospheric indicators measured by radiosonde and their general accuracies (National Oceanic and Atmospheric Administration, 1997). The radiosonde technique has been well developed and can provide highly accurate measurements, e.g., ±0.5 °C accuracy for temperature and ±1 hPa accuracy for pressure from surface up to 100 hPa. A GPS receiver is used with the radiosonde sensors for positioning. The positional data can provide wind information with an accuracy of about 1 m/s for speed and 5° for direction (National Oceanic and Atmospheric Administration, 1997).
12
Figure 2-1 Distribution of WMO radiosonde stations
Table 2-1 Indicative accuracy of the radiosonde measurements
Variable Range Accuracy
Pressure From surface to 100 hPa 1 hPa to 2 hPa near 100 hPa
100 to 10 hPa 2 %
Temperature From surface to 100 hPa 0.5 °C
100 to 10 hPa 1 °C
Relative humidity Troposphere 5 %
Wind direction From surface to 100 hPa 5° for wind speed less than 1.5 m/s, 2.5° for wind speed greater
100 to 10 hPa 5°
Wind speed From surface to 100 hPa 1 m/s
100 to 10 hPa 2 m/s
Geopotential height of From surface to 100 hPa 1 % near the surface significant level decreasing to 0.5 % at 100 hPa
It can be seen from Figure 2-1 that the distribution of the radiosonde network is uneven i.e. stations are concentrated on land (continents and islands) and data from radiosonde
13 network are limited in areas over the ocean. In the Southern Hemisphere, the density of the radiosonde network is significantly lower than that in the Northern Hemisphere. The insufficient atmospheric information in the Southern Hemisphere and over the ocean has a negative impact on many scientific and operational applications (National Research Council, 2007). Radiosonde equipment can also be dropped from aircraft or rocket. This is useful for special investigations such as tropical cyclones. Nevertheless, this technique is not cost-effective.
2.2.2 Space-based observations
Meteorological satellites and other related Earth observation techniques have developed rapidly since the successful TIROS-1 (Television Infrared Observation Satellite -1) mission in 1960 (Mohr and Schmetz, 2006). The USA navigation satellite was small (about 120 kg) and in a LEO (about 700 km altitude). Although it operated for only 78 days, it is a land mark development in history as it sent back the first imagery of the Earth from space.
During the 1960s and 1970s, after TIROS, a large number of meteorological satellites were sent into orbit (Chuvieco, 2008; Kidd et al., 2009; Schwarz, 2004). Figure 2-2 indicates the key milestones of these space meteorological programs by different countries (from 1060s to the recent). These important programs included the first meteorological satellite Nimbus-1 in 1964, the first geostationary meteorological satellite ATS-1 (Applications Technology Satellites) in 1966, the first operational weather satellite ESSA-1 (Environmental Science Services Administration Satellite Program) in 1966, the first temperature sounder Nimbus-3 in 1969, the first operational temperature sounder NOAA-2 n 1972, the early operational geostationary satellites (SMS-1 (Synchronous Meteorological Satellite) from the USA, the GMS (Geosynchronous Meteorological Satellite) from Japan and Meteosat-1 from Europe, and polar-orbiting satellite TIROS-N in 1978.
14 1960 1964 TIROS-1 Nimbus-1
1966 1966 ESSA-1 ATS-1
1972 1969 NOAA-2 Nimbus-3
1975 1974 1975 SMS-2 SMS-1 GOES-1 1977 GOES-2 1977 GMS 1980 1978 GOES-4 GOES-3 1981 1981 GOES-5 GMS-2 1983 1984 GOES-5 GMS-3 1987 GOES-7 1989 GMS-4
1994 1994 ELEKTRO N1 GOES-8 1994 FY-2 1995 1995 GMS-5 GOES-9 1997 1997 FY-2A GOES-10 1999 2000 TERRA 2000 GOES-11 2001 FY-2B GOES-12 2002 2002 AQUA 2004 Meteosat-8 2004 2005 AURA FY-2C 2005 MTSAT-1R 2006 Meteosat-9 MTSAT-2 2006 2008 2006 FY-2D FY-2E GOES-13 2008 2010 FY-3A GOES-15 2009 2010 GOES-14 2010 FY-3B 2011 COMS-1 2011 NPP ELEKTRO L1 Figure 2-2 Chronological development of key meteorological satellites
These satellite-based meteorological observational programs have been developed over the last three decades. Depending on their purposes, these satellites operated in different orbits such as geostationary orbit, polar orbit and sun-synchronous orbit (Kidd et al., 2009). The satellites typically carried a number of observational instruments. Visible spectrum remote sensing instruments can capture information about the cloud system and wind. Infrared spectrum instruments use different spectrum bands to determine atmospheric information including temperature, pressure, water, ozone, carbon dioxide, etc. Table 2-2 shows the key meteorological satellite programs that provide important atmospheric information for current meteorological research and operations (World Meteorological Organization, 2008).
15 Table 2-2 Meteorological satellites and their countries, orbits and observed information
Satellite Country Orbit Observed atmospheric information
TIROS-N/NOAA-A USA LEO Imagery and temperature sounding
DMSP USA LEO Image and microwave imagery
METEOR Russia LEO Image
Feng Yun -1 P.R. LEO Image China
SPOT French LEO High-resolution image
ERS-1 Europe LEO Sea-surface temperature and wind
GOES USA Geostationary Image
GMS Japan Geostationary Image
METEOSAT Europe Geostationary Image
INSAT Indian Geostationary Image and Wind
Satellite-based meteorological observational techniques have many advantages when compared to conventional ground-based observations (World Meteorological Organization, 2008). Firstly, the space sounding observations have a global coverage. Therefore, many remote areas such as mountains, deserts, Polar Regions and the oceans can be continuously observed. Atmospheric conditions over these areas play a significant role in climate and weather events. Secondly, satellites fly at a high speed, and as such the payload instruments can observe large areas with a high temporal resolution. A ground system that covers a large area at the similar temporal resolution will result in extremely high costs of station establishment and operation. Thirdly, the satellite observational system is available 24 hours and 7 days a week, and is not affected by the weather conditions. These advantages provide meteorological satellites with unique capability for atmospheric observation. For example, the generation, development and transformation of tropical cyclones, across a large area from the ocean to land, can only be observed by satellite-based instruments (Chevallier and Engelen, 2007; Drusch et al., 2007).
16 However, most existing satellite systems do have some limitations:
1. Low vertical resolution – lower vertical resolution than radiosonde measured profiles. 2. Low point accuracy – the accuracy in each point (pixel) is relatively low from these early satellite based techniques. 3. Poor observational capability for the lower atmosphere – space-based observation techniques view the atmosphere from orbits of hundreds and even thousands of kilometres. 4. Poor flexibility for instrument updates – software updates on the space instruments is difficult and hardware updates almost impossible. 5. A complicated information retrieval process.
Due to these limitations, ground-based observations still play an important role in meteorological applications. The new GPS based technique has all the advantages of the space-based systems but overcomes the serious disadvantages.
2.3 GNSS and atmosphere
2.3.1 Fundamentals of satellite positioning technique
Satellite positioning technique utilises man-made satellites to determine locations on and near the Earth's surface. The space-based positioning technology can provide accurate positional services for a large area and in all weather conditions. The Navy Navigation Satellite System (NNSS), developed by the USA Department of Defence in the 1950s, was the first satellite positioning system (Prasad and Ruggieri, 2005). Another early satellite positioning system, named Tsikada, was developed by the USSR (Union of Soviet Socialist Republics) in the 1970s. The performance of these early systems was poor in both accuracy and usability (Hofmann-Wellenhof et al., 2001). GPS was developed by the USA military from the 1970s to provide a satellite-based positioning system with higher accuracy, better global coverage and a quicker positioning system by employing a constellation of 24 satellites (Gleason and Gebre-Egziabher, 2009).
17 Over the last three decades, GPS has developed and improved continually. The success of GPS has encouraged the development of other satellite positioning systems. Many countries, such as China, Japan, Russia, as well as the European Union have proposed their own satellite positioning systems. The terminology of GNSS was introduced to label all of these systems. GPS is currently the most successful satellite positioning system by means of technology developments and applications. Many other satellites, such as meteorological and Earth observing satellites, also use GPS to determine their orbit information.
GPS components
The GPS has three segments, namely the space segment, the control segment and the user segment (Kaplan and Hegarty, 2006). The GPS satellite constellation (i.e. the space segment) currently consists of 30 satellites (2010). These satellites travel at an altitude around 20,200 km above the earth surface. The constellation orbit is designed to ensure clear observations of at least four satellites above 15º elevation at any point on the Earth at all times. This is the minimum required for the global positioning services.
The control segment includes a master control station, six monitor stations and four ground antennas. The six monitor stations are located at Colorado Springs, Cape Canaveral, Ascension Island, Diego Garcia, Hawaii and Kwajalein (Hofmann-Wellenhof et al., 2008). The monitor station at Colorado Springs is the only one that is located in the central USA and is also located close to the master control station. The Cape Canaveral station is located on the coastline of Florida. The other four stations are all within ocean areas. These monitoring stations track all GPS satellites in view continuously. All information collected at each monitor stations is sent to the master control station. Based on the satellite tracking data, the orbit and clock parameters are calculated and then uploaded to the GPS satellites via the ground antennas located at the four monitor stations (i.e. Cape Canaveral, Ascension Island, Diego Garcia and Kwajalein).
The GPS user segment refers to the numerous application receivers of various types. A typical receiver has one or an array of GPS antennas, precise clocks, and a processor.
18 For the GPS RO technique, a space-based high accuracy GPS receiver is specially designed for occultation purposes and mounted on LEO satellites.
GPS positioning
The GPS receiver captures time and orbit information transmitted from the GPS satellites. Basically, the distance between GPS receiver and GPS satellite can be calculated using the transit time (i.e. the time taken for the GPS signal to travel from the GPS satellite to the receiver). In order to solve a three dimensional position, knowledge of the distances to at least three GPS satellites are required. In most situations, the receiver is able to view more than four satellites and the additional satellites help to deliver highly accurate positions. The determination of the exact distance between receiving satellite and transmitting satellite requires a number of corrections to remove biases such as clock errors, atmospheric effects and multipath. The time delay due to the atmosphere (including the ionosphere, stratosphere and troposphere) is one of the most important factors contributing to the GPS error budget.
2.3.2 GNSS signals and the atmospheric effects
GPS signals
GPS positioning is based on the measurements of time difference from precise atomic clocks. GPS carrier signals from GPS satellites are a multiple of atomic clock frequency of 10.23 MHz (Xu, 2007). Currently, the two main GPS carrier signals, L1 and L2, have frequencies of 1575.42 MHz (i.e. 154 times of 10.23 MHz) and 1227.60 MHz (i.e. 120 times of 10.23 MHz) respectively. The latest GPS satellite, Block IIF launched on 27th May 2010, has an additional frequency (L5) with a frequency of 1176.45 MHz (i.e. 115 times of 10.23 MHz) (Schmid, 2009).
Both the L1 and L2 carrier signals carry a pseudorandom noise (PRN) code, namely the coarse/acquisition (C/A code) and precision (P) code (Hofmann-Wellenhof et al., 2008). The C/A code is for civilian use and P code is for non civilian use (i.e. military use). The new signal mounted on L5 is an additional enhanced civilian-use signal. There are a number of other codes reserved for GPS, which can be used in the future system (summarised in Table 2-3). The L1 and L2 carrier signals also contain navigation
19 messages which includes information about satellite orbits, the satellite status and correction data.
Table 2-3 GPS carrier signals, and code signals and their descriptions (Hofmann-Wellenhof et al., 2008)
Link Code Description
L1 C/A Coarse/acquisition code
P(Y) Precision code; Y-code replaces the P-code in anti-spoofing mode
M Military code
L1C Civil code on L1
L2 L2C Civil code on L2; general reference to the code signal on L2, which consists of some combinations of C/A, L2CM, and L2CL
P(Y) Precision code; Y-code replaces the P-code in anti-spoofing mode
M Military code
L2CM Moderate-length code on L2C
L2CL Long-length code on L2C
L5 L5C Civil code on L5; general reference to the civil code on L5, which consist of some combinations of L5I and L5Q
L5I In-phase code on L5
L5Q Quadraphase code on L5
20 Error sources of the GPS measurements
In order to obtain high accuracy GPS measurements, the bias errors in GPS observations need to be eliminated. The main errors result from clocks in the receiver and onboard the satellite, the ionospheric effects, the atmospheric (stratospheric and tropospheric) effects, Earth and ocean tide effects, multipath effects, and the relativistic effects.
These errors can be modelled in an observation equation below (Xu, 2007).
s s Rr (tr ,te ) = ρ r (tr ,te ) − (δtr −δts )c + δ ion + δ tro + δ tide + δ mul + δ rel + ε where
s Rr : the measured pseudorange between receiver (r) and GPS satellite (s),
tr ,te : GPS signal reception and emission time of the satellite and receiver respectively
s ρ r : the geometric distance between receiver (r) and GPS satellite (s),
δt r ,δt s : the clock errors from receiver and satellite respectively,
c : the velocity of light,
δ ion : the ionospheric effects,
δ tro : the atmospheric effects,
δ tide : the Earth tide and ocean loading tide effects,
δ mul : the multipath effects,
δ rel : the relativistic effects and,
ε : all the remaining errors.
Atmospheric effects: ionosphere and natural atmosphere
The Earth’s atmosphere extends from the Earth’s surface up to an altitude of about 10,000 km. Based on the characteristics of different zones of the atmosphere, five
21 atmospheric layers are defined (Piccinini et al., 2000). These layers are the troposphere, stratosphere, mesosphere, thermosphere and exosphere in order of increasing altitude (see Figure 2-3). In electromagnetic physics, the atmosphere is categorized into two parts: 1) the ionosphere (including thermosphere and exosphere) where the atoms and molecules are charged electrically by the solar radiation from the Sun, and 2) the natural atmosphere which is lower part of the atmosphere (Moran and Morgan, 1986). Figure 2-3 shows the general profiles of key atmospheric parameters, namely temperature and pressure in the natural atmosphere, plus electron density in the ionosphere.
Altitude (km) 10,000 Exosphere
690 Thermosphere Ionosphere
0.001 85 Mesosphere Mesosphere 10 4 10 5 10 6 45 1 Stratosphere Electron Density Stratosphere 100 (cm -3) 12 Troposphere 1000
300 600 900 1200 Pressure (hPa) Temperature ( ºººC) Figure 2-3 Earth's atmosphere structure, layers and parameters (i.e. temperature, pressure and electron density in ionosphere)
GPS signals are L-band electromagnetic waves (1.023 GHz ) (Hofmann-Wellenhof et al., 2008). The Earth's atmosphere has major effects on the signal propagation. As the signal propagates through the atmosphere, the signal is reflected, refracted and absorbed by the atmosphere (Misra and Enge, 2006). The ionospheric effect on GPS signals is frequency-dependent and studies of this effect on more than one frequency (L1 and L2) can distinguish the ionospheric effect from other errors. On the other hand, the effect of the natural atmosphere on GPS signals is frequency-independent. Details of the free atmospheric effect on GPS signals and the GPS RO atmospheric information retrieval process will be discussed in Chapter Three.
22 2.4 GNSS-based atmospheric observations
GNSS meteorology is a technique that can attain atmospheric information from GNSS observations. The GNSS receivers can be located on the ground or be mounted on aircrafts and satellites. With the ground-based GNSS observations, integrated perceptible water vapour between the GNSS satellites and receivers can be computed. The space-based technique utilises GNSS receiver on-board aircrafts and other satellites to probe atmospheric profile information.
2.4.1 Ground-based GPS water vapour observations
Water vapour is one of the key factors that cause GPS signal delays. Development in the high-precision GPS positioning technique has opened a new opportunity to extract column water vapour information along the GPS signal path. Since the early 1990s, many studies have proved that high quality slant-path water vapour information can be obtained by ground-based GPS stations (Bevis et al., 1992; Rocken et al., 1993; Rocken et al., 1997b; Ware et al., 1997).
Compared to other observational techniques, the GPS water vapour monitoring method has three key advantages:
• Continuously monitor,
• Measure multiple lines between the station and the GPS satellites in view,
• All weather condition capability
With a network of CORS (continuously operating reference station) GPS stations and the tomography technique, the slant-path water vapour information can be mapped into a three-dimensional moisture field over the network (Braun and Hove, 2005; Nilsson et al., 2007; Troller et al., 2006). Assimilation of the slant-path water vapour from GPS into a Numerical Weather Prediction (NWP) model has been done in near real-time (Gendt et al., 2001; Haan et al., 2009).
23 2.4.2 Space-based GNSS RO observations
The method of radio occultation for atmospheric observation was first applied in interplanetary discovery. A joint effort made by Stanford University and the Jet Propulsion Laboratory (JPL) developed a radio observing system for the Mariner 3 and 4 missions to Mars (Yunck et al., 2000) in the 1960s. For many decades, this method has been used in most solar system discovery missions to study both the ionosphere and atmosphere of the planets and their moons (Fjeldbo et al., 1971; Pavelyev and Kucherjavenkov, 1978).
It was not surprising that the radio occultation technique had been proposed for Earth's atmospheric observations when GPS became a concept in the 1970s. However, the first experimental mission GPS/MET occurred in 1995 and the reason for the delay was mainly due to the low accuracy of GPS and various technological barriers of GPS signal retrievals (Hajj et al., 2002b; Hocke, 1997). The pioneering GPS/MET mission in 1995 was a very successful mission that delivered encouraging results (Beyerle et al., 2001). In the following decade, many GPS RO missions were proposed and developed by the international community. Many studies have proved that the GNSS RO Earth atmospheric observational technique can overcome the limitations of current observing techniques, such as a poor sampling distribution, a low resolution of radiosonde (from ground stations, ships and airplanes) and a low vertical resolution of down-looking satellites (Anthes et al., 2000; Chao et al., 2000; Foelsche et al., 2006).
2.5 Summary
The GNSS RO technique is one of the most important and valuable meteorological applications of GNSS. It addresses one of the most challenging aspects of Earth sciences i.e. the Earth's atmospheric observation. This chapter has reviewed the satellite positioning technology and the role of the atmosphere in GNSS positioning technique. The current Earth's atmospheric observational technologies, including both ground-based CORS stations and satellite remote sensing, have been summarized. The chapter presented a general description of the GNSS-based observation techniques (i.e. ground- and space-based techniques). The key aspects of the GNSS RO technique will be discussed in depth in the following chapters.
24 Chapter 3 The GNSS RO Technique
3.1 Introduction
In Chapter 2, a brief introduction to the GNSS atmospheric observational techniques including both ground-based GPS water vapour observations and space-based RO observation has been given. Chapter 3 discusses the fundamentals of the GNSS RO and its technical components in particular the retrieval process and the error sources related. The two key aspects of the GNSS RO retrieval technique are: 1) the precise orbit determination (POD) of both the GNSS and LEO satellites, and 2) the algorithms that extract the atmospheric information from the raw GNSS RO signal phase measurements. A study of the fundamental theory will show how the atmospheric parameters (i.e. bending angle, refractivity, temperature, pressure and moisture) are derived from the GNSS signals and phase measurements. The retrieval process includes the algorithms and methods to eliminate the clock errors (the satellite and receiver clock error), the correction of atmospheric errors, and also the consideration of the uncertainty of the atmosphere. The remainder of this chapter will then discuss the advantages of the RO technique and its wider applications in meteorological and climate studies.
3.2 GNSS RO concept and atmospheric retrieval theory
3.2.1 GNSS RO concept
The GNSS RO technique utilises a high precision GNSS receiver mounted on an LEO satellite to track the navigation signals transmitted by the GNSS satellites. As the LEO satellite rises or sets behind the Earth, the view of the GNSS satellites changes, The signals pass through a large amount of atmosphere (including both the ionosphere and natural atmosphere) and are affected significantly (Figure 3-1). The propagation of each signal is influenced by the atmosphere and the received GNSS signals will show a delay due to the atmospheric effects. However, the atmospheric information is embedded in the received signals as well as other information and errors. With careful
25 examination of the errors and atmospheric effects in the radio signals, atmospheric bending angle and refractivity profile can be derived from the GNSS measurements.
Figure 3-1 Concept map of GNSS RO technique
Figure 3-2 A typical GNSS RO atmospheric retrieval process
The atmospheric bending angle and refractivity are therefore related to different atmospheric properties (i.e. temperature, pressure, humidity and electron density) at different zones of the atmosphere (Figure 3-2). Therefore, the direct atmospheric products of the GNSS RO retrievals are the electron density profiles in ionosphere, the temperature and pressure profiles in the stratosphere and high troposphere, and the
26 water vapour profiles in the middle and low levels of the troposphere (Hocke, 1997). This thesis is focused on the natural atmosphere and consequently the GNSS RO retrieval process refers to the derivation of this natural atmospheric information i.e. radio wave bending related to the natural atmosphere, refractivity, temperature, pressure and water vapour profiles in the natural atmosphere.
3.2.2 GNSS RO retrieving theory and error sources
The RO LEO satellite captures the signals not only from the occulting GNSS satellites but also other GNSS satellites in view. The observation of multiple GNSS satellites ensures the high precision orbit determination of the LEO satellite. The LEO’s orbit determination, together with the precise GNSS satellite orbit information, is a key element of the GNSS RO retrieval process (Hwang et al., 2008; Kursinski et al., 2000).
GNSS and LEO POD GNSS precise orbit
LEO navigation LEO orbit and clock GNSS RO table observations errors
Ground-based GNSS observations Single/double GNSS RO Data assimilation difference observations
NWP model Position and velocity Position and velocity Bending angle of satellites of satellites profiles
atmospheric profiles Excess phase delays Refractivity profiles
Amplitude Dry temperature and pressure profiles Inversion of atmospheric parameters Figure 3-3 GNSS RO retrieval processes
Figure 3-3 shows the whole GNSS RO atmospheric retrieval process and there are three key segments in the process namely GNSS and LEO satellite POD, inversion of atmospheric parameters and data assimilation to the numerical weather models. The POD is based on the GNSS observations and reference data from GNSS ground stations. With precise obit information, then the excess phase delay can be calculated. Then the
27 bending angles and refractivity are derived from the delay. The bending angles and refractivity profiles can be assimilated into numerical weather models while the further retrieved temperature and pressure profiles are valuable for climate studies.
The impact of atmospheric effects on the GNSS signals is contained in the Doppler shifts measured by the GNSS RO receiver after the satellites movement contributions are removed. From the geometry in Figure 3-4, the relationship between the angles ( ΦT and ΦR) and vectors (V T and V R) with Doppler shift (f d) and transmitter frequency (f T) can be presented as (Kursinski et al., 2000):
r θ r θ fd = −fT / c(VT cos Φ T + VT sin Φ T + VR cos Φ R − VR sin Φ R )
In order to separate the effects of the natural atmosphere and the ionosphere on the Doppler shifts, both the L1 and L2 signals are captured and then a double differencing technique is used.
Figure 3-4 GNSS RO geometric representation
The bending angle α can be calculated from ΦR, ΦT, and θ based on the equation below (Kursinski et al., 2000):
α = ΦT + ΦR +θ −π
28 Using the Abelian transformation, the refractivity index (n) at the tangent point can be calculated from the bending angle using the equation below (Fjeldbo et al., 1971):
1 ∞ α n(r) = Exp ∫ da π a 2 2 1 a − a1 where a is the asymptotic ray minimum distance and r is the tangent radius.
In the Earth’s atmosphere, there are four factors affecting the refractivity index of the radio signals (i.e. the dry neutral atmosphere, water vapour, free electrons and particles). The equation below represents their relationships to the refractivity index (Kursinski et al., 2000).
P P n N = ()n −1 ×10 6 = 77 6. + 73.3 ×10 5 w − 03.4 ×10 7 e + 4.1 W T T2 f 2
Refractivity (N) is affected by dry pressure P (hPa) and temperature T (K); wet pressure 3 Pw and temperature; electron number density n e (m ) and transmitter frequency f (Hz); and water content W (m 3).
In different layers of the atmosphere, the contributions of the above four sources are variable and sometimes are even negligible. For example, the electron effects density can be safely ignored in the troposphere and lower stratosphere. Also the water in the stratosphere is at a very low level. This variability is taken into account during the retrieval process.
Error sources in the retrieval process
The atmospheric retrieval process involves a sequence of steps to eliminate errors and to retrieve other information contained in the GPS signals received (Beyerle et al., 2006; Gorbunov and Lauritsen, 2007; Hoeg et al., 2002; Rieder and Kirchengast, 2001; Steiner and Kirchengast, 2005). It is critical to consider the sources of errors at each step of the GNSS RO retrieval process to obtain useful results. The main steps are 1) derive the atmospheric effects (both ionosphere and atmosphere) on the radio signals, 2) remove the ionospheric effects; and 3) calculate different properties of the atmosphere in different levels from the refractivity. In each step, there are different sources of
29 errors. The errors can also be categorised into three groups, i.e. errors related to satellite and receiver, errors related to ionosphere, and errors related to the atmosphere (Kursinski et al., 2000). Table 3-1 presents these error sources and their key characteristics and the impacted atmospheric properties (i.e. retrieval results).
Table 3-1 Error sources, their characteristics and key properties
Error sources Characteristics Key properties
Errors related to satellite and receiver
Signal to noise ratio Random phase error Refractivity and temperature at performance high altitudes; Loss of signals in the lower troposphere
Clock instabilities Phase error and selective Refractivity, pressure and availability (SA) temperature
Local multipath Phase error Pressure, geopotential, vertical refractivity and temperature profiles.
Orbit determination Orbit and velocity errors (both Refractivity, pressure and accuracy GNSS and RO satellites) temperature
Errors related to ionosphere
Residual ionospheric Retrieval error Pressure and temperature effects
Errors related to atmosphere
Refractivity constant Retrieval error Pressure and temperature in mid uncertainties and low altitudes
Uncertainties in Abelian Retrieval error Pressure and temperature in high method and hydrostatic altitudes integral boundary condition
Horizontal refractivity Retrieval error Refractivity in troposphere structure
The geometrical optics Retrieval error Bending angle at high altitudes approximation
Water vapour/density Retrieval error Dry and wet components in ambiguity lower atmosphere
30 3.3 Meteorological and climatic applications
This section reviews the advantages of the GNSS RO technique compared with conventional atmospheric observation techniques, i.e. radiosonde and satellite remote sensing. The applications in weather forecasting and climate studies using the GNSS RO products are also discussed.
3.3.1 Advantages of the GNSS RO technique
GNSS RO technique compared with ground-based observations
Space-based atmospheric observation technologies overcome many limitations of the ground-based methods (Feng and Herman, 1999; Ware et al., 1995; Wickert et al., 2009b). Some of these limitations are critical issues for atmospheric sciences and are difficult to improve. For example, the coverage plus the horizontal resolution of the radiosonde station network is limited because the network requires suitable physical locations and easy accessibility (World Meteorological Organization, 2008). The area covered by the radiosonde station network is relatively small when compared with those not being covered, such as oceans, deserts, and mountains. Many areas have either no suitable places for building stations or hard to access by personnel. However, the coverage and distribution of the satellite based RO technique is related to the orbit design of the LEO satellite and the GPS constellation (Fu et al., 2011). A simulation study on the locations of ROE has been conducted and the results are presented in Chapter 4. The study demonstrates that the GNSS RO technique has uniformed global coverage and the capability of focusing on particular regions with more density observations. Moreover, ground-based observational stations have high operation and maintenance costs and this is another limiting factor for the establishment of an ideal observational network. The space-based GNSS RO atmospheric observational technique requires less ground resources when compared with the ground network. With its global capability, the operational cost of a GNSS RO mission is relatively low in the long-term and can be shared by international partners.
31 GNSS RO technique compared with satellite remote sensing
When Compared with satellite remote sensing techniques, one of the main advantages of the space-based GNSS RO technique is its capability to observe all levels of the atmosphere from the atmosphere to the ionosphere, i.e. from the ground to the outer edge of the atmosphere (about 10,000 km) (Melbourne et al., 1994; Yunck and Hajj, 2005). A remote sensing mission often focuses on specific layers of the atmosphere to achieve better results. Moreover, in terms of the complexity and the operational costs of the space systems, the GNSS RO technique uses space-borne GNSS receivers, which have low power consumption and have a lower cost compared with the remote sensing satellite instrumentation. Many GNSS RO LEO satellites are compact in size, light weight and have low power consumption. The six FORMOSAT-3/COSMIC satellites were launched in one spacecraft (Fong et al., 2008; Liou et al., 2006). Often, GNSS RO LEO satellites are equipped with a variety of different instruments to support for multiple scientific purposes (Anthes et al., 2008; Wickert et al., 2001b; Wickert et al., 2009a). These unique characteristics of the LEO satellites make the GNSS RO technique more cost effective compared with other Earth observation satellite missions. For example, SUNSAT was a university initiative and was launched with a RO receiver (Mostert and Koekemoer, 1997).
GNSS RO technique compared with sensor-based observations
The core of the GNSS RO technique is radio science and the atmospheric information is retrieved from the atmospheric impacts on the GNSS radio signals (Pavelyev and Kucherjavenkov, 1978; Pavelyev et al., 2009). The RO technique has many advantages when compared with the direct sensor-based radiosonde measurements and image-based remote sensing techniques. First, the radio technique is not affected by the weather condition. The weather (for example clouds, rain, aerosols) will not influence the radio signals and the accuracies retrieval of atmosphere. On the other hand, remote sensing instruments can be affected by clouds (even particles) and weather phenomena which leads to various difficulties and issues in the image processing (Guzzi, 2003). The radiosonde technique measures atmospheric information using sensors, which can only work in specified conditions (e.g. a specified temperature range), and their performance and accuracy can be affected by the weathers conditions. The weather balloons are difficult to launch in extreme weather conditions and also may not reach
32 the designed level of the atmosphere. Another issue related to the radiosonde is that its performance has dependencies on altitude and pressure (National Oceanic and Atmospheric Administration, 1997). GNSS RO observations are generally independent to the weather conditions (Kursinski et al., 2000).
The benefits of the GNSS-based system
GNSS radio signals are generated using atomic clocks aboard. The GNSS and RO receivers are using similar clock technologies. As a result, the measurements of GNSS RO signals are stable and precise. Therefore, GNSS RO measurements have no long-term drift and mission-to-mission biases. This ensures continuous high quality observations from the GNSS RO techniques across all missions. However, the radiosonde weather sensors are designed and produced by various companies that employ different techniques and standards. Research shows that there are significant systematic bias between different types and models of radiosonde sensor systems (He et al., 2009; Ho et al., 2009). This bias is one of the major error sources in atmospheric studies (Knudsen et al., 2006). The other benefit of the GNSS RO is that the atom clocks generate radio signals continuously. The sampling rate of the current GNSS RO receivers can reach as high as 50 Hz. Therefore, a high vertical resolution of the atmospheric profiles can be achieved from these GNSS RO missions. The vertical resolution of GNSS RO is much higher than both the satellite remote sensing and the radiosonde techniques (Fu et al., 2009a; Hayashi et al., 2009). The details of atmospheric status and processes can be recorded and are valuable for many atmospheric studies and applications.
Table 3-2 compares the GNSS RO technique with the radiosonde and remote sensing methods with respect to atmospheric observations. The outstanding technique for each category is highlighted in italics. The GNSS RO demonstrates significant advantages over the other two techniques.
33 Table 3-2 A comparison of three atmospheric observation techniques
Observation Radiosonde Remote Sensing GNSS RO
Coverage Ground-based and Space-based and Space-based and limited by geographical global coverage global coverage accessibility
Vertical resolution High Low Very high
Horizontal inhomogeneous uniform Homogeneous, resolution depended on the LEO orbit design
Temporal 2-4 times per day Depend on the orbit Depend on the LEO resolution design, normally orbit design, normally higher than radiosonde higher than radiosonde
Weather condition Unavailable in extreme Significantly affected All weather capability weather conditions by cloud and particles in the atmosphere
Cost High operation and Low in a long-term Low in a long-term maintenance cost
Calibration Necessary and Calibration needed for No need, also mission important different instruments independent
Layers of From surface up to Instruments are All layers atmosphere 30km specially designed for certain layers
Accuracy High Low Very high
Stability Low High Very high
3.3.2 Contributions of the GNSS RO to meteorology and climate
The atmospheric data products delivered from past and current GNSS RO missions have shown encouraging results in many potential applications. This section discusses how the use of GNSS RO data has contributed to climatology, i.e. covering the data gap in some regions, serving as an important data source for many weather forecasting and prediction models, and providing quality atmospheric profiles for detailed and advanced atmospheric studies.
34 Covering the data gaps
GNSS RO data have provided significant additional information to weather studies in data-limited areas such as the Southern Hemisphere (SH), Polar Regions, unpopulated areas, and over the oceans. Recent research has suggested that CHAMP data can help with a number of SH studies: for example, reducing the temperature cold bias error, improving water vapour analysis in mid-latitudes, correcting the surface pressure analysis, and increasing the accuracy of forecasts (Zou et al., 2004). CHAMP and SAC-C atmospheric retrievals have been used to improve NWP analysis and forecasts for the Antarctic (Wee et al., 2008). The GNSS RO technique has brought a new impetus for scientists interested in the Antarctic, where traditionally there have been very limited weather observations available. Various studies have also been carried out to validate radiosonde measurements by using GNSS RO data (Alexander et al., 2009; Boccara et al., 2008; Ge, 2006; Juarez et al., 2009; Kuo et al., 2002; McDonald and Hertzog, 2008; Nedoluha et al., 2007; Vey and Dietrich, 2008). The marine boundary layer using the GNSS RO retrievals has been investigated (Xie, 2006). These studies have demonstrated the value of the GNSS RO technique for providing upper air profiles with a global coverage. In the near future, with more GNSS RO satellites becoming available, both the weather and climate studies in previously data-sparse areas will be significantly improved. The GNSS RO missions provide uniform global coverage observations and fill the geographical gaps from the station-based observation networks. Furthermore, the technique is “mission calibration free” and continuous quality data can be expected over a long-term from GNSS RO missions. This is an important improvement for climate studies such as trend analyses.
Quality data source for numerical models
The role of the upper air information is critical for many numerical weather and climate models. Radiosonde measurements are a key input for the global prediction models (e.g. NCEP, MetOffice and ECMWF), and also the regional models (e.g. rainfall forecasts and cyclone models). This is because they provide the best vertical resolution throughout the atmosphere. The GNSS RO technique generates higher quality and resolution atmospheric profiles with uniformed global coverage. Therefore, the GNSS RO data are an important data source for many numerical weather prediction models. Major international meteorological centres (e.g. NOAA, MetOffice and ECMWF) have
35 assimilated the GNSS RO data into their global weather predication models and improvements of these models have been demonstrated (Cucurull et al., 2008; Kuo et al., 2000; Loscher, 2004; Loscher et al., 2006; Pingel and Rhodin, 2009; Rennie, 2008).
Direct observations of cyclones are difficult to obtain. Information about the development of the cyclones is limited to observations from aircraft dropsonde sounding and data from ground-based radar stations. The accuracy of the prediction of cyclone tracks can be improved when more information is available and this could lead to a significant reduction of damages from cyclones. The GNSS RO technique is not affected by weather conditions therefore valuable atmospheric profiles can be obtained from the GNSS RO events that occur during the cyclones. Some of these RO events have been captured and applied in cyclone forecasting studies and have showed good results (Guo et al., 2009; Huang et al., 2005; Kueh et al., 2009; Kuo et al., 1997). Rainfall prediction is another example of where the GNSS RO data can contribute additional data inputs to the models. Some research groups have successfully used the CHAMP and FORMOSAT-3/COSMIC RO data to rainfall predictions (Huang et al., 2007; Seko et al., 2009).
High resolution atmospheric information
The GNSS RO atmospheric observation technique has global coverage and an all weather capability. Therefore, the Earth’s atmosphere can be observed more conveniently and efficiently than previously possible with the conventional radiosonde and remote sensing techniques. It is expected that over 3000 daily atmospheric profiles can be generated from the FORMOSAT-7/COSMIC-2 mission for the Australasian region (see Chapter 4). This number of observations is much higher than those from the 38 Australian radiosonde stations observing 2 to 4 times per day. Moreover, these observations are not limited by the same location and time constraints experienced by the radiosonde observation technique. The RO data is more uniformly distributed in both space and time. With advances in multiple GNSS, it is expected that the data will significantly improve the coverage, horizontal resolution and sampling frequency of atmospheric observations. Furthermore, the GNSS RO technique has a higher vertical resolution than the radiosonde as well as satellite remote sensing due to its high sampling rate. It is expected that the GNSS RO technique can observe the Earth’s atmosphere where traditional radiosonde and remote sensing can not.
36 The GNSS RO technique currently provides the best (vertical, horizontal and temporal) resolutions for atmospheric observations. This ongoing quality data has been invaluable for many aspects of atmospheric studies. For example, the global gravity wave signatures in the upper troposphere and stratosphere can be obtained from GNSS RO profiles (Alexander et al., 2008; Hocke and Tsuda, 2001; Liou et al., 2003; Torre et al., 2009). Studies of atmospheric waves, turbulences and tides have be carried out using GNSS RO data (Cornman et al., 2009; Gorbunov and Kirchengast, 2006; Gubenko et al., 2008; Torre et al., 2006; Zeng et al., 2008). Wang et al. (2009) captured atmospheric abnormal state during the 2008 Chile volcano eruption. Climate studies rely on high resolution atmospheric observations. Studies have demonstrated the potential of GNSS RO profiles for characterizing troposphere structures and changes (Bizzarri et al., 2006; Foelsche and Kirchengast, 2004; Schmidt et al., 2005a; Sokolovskiy et al., 2007; Staten and Reichler, 2008).
3.4 Summary
This chapter introduced the fundamentals of the GNSS RO technique and outlined the technical aspects such as the retrieval processes and error sources. Advantages of the GNSS RO techniques have been reviewed with a comprehensive comparison of the current atmospheric observational techniques such as radiosonde and satellite remote sensing. The RO technique has the potential to advance many atmospheric related studies by providing high quality, high resolution atmospheric information with a global coverage.
37 (This page intentionally left blank)
38 Chapter 4 GNSS RO Missions – The Past, Present and Future
4.1 Introduction
The radio occultation technique for atmospheric observations has been applied and developed through planetary discovery programs to Venus, Mars, Saturn and their moons since the 1960s (Fjeldbo et al., 1971; Kolosov et al., 1979). During the late 1970s and the 1980s, GPS, also based on radio technology, had been developed for global positioning, navigation and timing purposes. Atmospheric scientists realized the potentials using GPS for the Earth’s atmospheric observations (Businger et al., 1996; Yunck and Melbourne, 1995). The proof-of-concept GPS RO mission the GPS/MET was launched in 1997 by the USA. Since the success of the first mission, a number of follow-on missions have been developed by different countries and regions. A review of these RO missions is presented in this chapter.
One of the notable trends in the development of the RO technique is the adoption of multiple GNSS (Kirchengast, 2005; Meehan, 2009). GPS has commenced a process of modernization and the first GPS III satellite has been scheduled for launch in 2014 (Hofmann-Wellenhof et al., 2008). The Russian GLONASS system is also embarking for updating to its next generation system. The European Galileo and the Chinese Compass systems have been under rapid development. Other countries are also planning regional satellite navigation systems or augmented systems such as Japan’s QZSS and the Indian Regional Navigational Satellite System (IRNSS) (Gleason and Gebre-Egziabher, 2009). As part of this research, a simulation study was designed to model the RO events from the proposed FORMOSAT-7/COSMIC-2 RO mission with GNSS receivers (i.e. receiving signals from GPS, GLONASS, Galileo and QZSS) for the Australian region. This study has shown a vast increase in the number of RO events in the region using FORMOSAT-7/COSMIC-2 and a large quantity of high quality plus high resolution atmospheric profiles (Section 4.3).
FORMOSAT-3/COSMIC generates about 2,500 daily atmosphere profiles. The future RO missions, with the capability of capturing multiple GNSS signals, will deliver tens
39 of thousands of daily RO events. The is a need to reduce the system and data processing latency so as to provide real-time data for many meteorological and forecasting applications (Larsen et al., 2005a; Terzo et al., 2009). Processing the vast amount of RO data and other reference data brings critical challenges for the data retrieval process. Study of the ICT infrastructure and data centre software engine has been conducted and recommendations of the handling and processing of the RO data for future RO missions are discussed in Section 4.4.
This research has clearly shown that Australia will gain valuable knowledge in the meteorological and satellite observation fields from the implementation of the recommended RO technique and process.
4.2 Review of the GNSS RO missions
The proof of concept satellite program GPS/MET was proposed and implemented by NOAA (National Oceanic and Atmospheric Administration). The MicroLab_1 satellite was launched into an orbit of 750 km, with a 70° inclination on 3 April 1995. The first atmospheric profile retrieved from this RO mission was obtained on 16 April 1995 (Healy, 1998). It was the first satellite equipped with a GPS RO payload for meteorological purposes. Dozens of RO satellites have been launched by various countries or via international projects since the late 1990s. Figure 4-1 shows the timeline and life span of the current and past RO missions as well as the proposed future missions in the following decade years.
After GPS/MET, South Africa and Denmark launched their first satellites (i.e. SUNSAT and Ørsted, respectively) which were equipped with the GPS RO receivers (Larsen et al., 2005b; Mostert and Koekemoer, 1997). SAC-C was an international effort by the USA, French, Brazil, Denmark and Italy (Hajj et al., 2002a). CHAMP (Germany) has the longest history of quality RO data from 2001 to 2008. Such a long period of continuous RO data permitted high quality climatological studies. In 2002, NASA launched an innovative twin LEO satellites GRACE (Beyerle et al., 2005). The two GRACE satellites travelled in orbit with 220 km apart and the mission was designed for measuring gravity recovery and atmospheric observations (by its RO
40 receiver). These early 21st century missions were significant advances by the international community.
Figure 4-1 Timeline and life span of GNSS RO missions
By 2007, both CHAMP and GRACE were near to the end of their mission life. The FORMOSAT-3/COSMIC constellation and a few other new RO missions now provide continuous measurements of the atmosphere. The FORMOSAT-3/COSMIC is a constellation of six LEO satellites and has successfully generated quality RO data. Metop is a series of three RO satellites and the first one (Metop-A) was launched in 2006 (Loiselet et al., 2000). Germany launched TerraSAR-X and TranDEM-X in 2007 and 2010 respectively and both satellites provide Earth observations including RO atmospheric data. Also, by 2010, other countries such as India, France and Korea, had launched RO satellites into various orbits (Cho et al., 2009; Notarpietro et al., 2006). There is also a large number of planned RO missions scheduled post 2010. These include FORMOSAT-7/COSMIC-2, Metop-B, SAC-D, CICERO, EQUARS, Megha-Tropiques, etc (Cho et al., 2009; Fong et al., 2009a).
41 CHAMP mission
The CHAMP mission was a German/USA research satellite launched by a Russian COSMOS space vehicle on July 15, 2000 into a near polar (i = 87°) circular orbit and an initial orbital altitude of 454 km (Wickert et al., 2001a). The orbit design ensured homogeneous and global coverage of observations for various atmosphere/ionosphere applications. The objectives of this project were to investigate the Earth’s gravity, magnetic field and also provide GPS RO meteorological data. The “BlackJack” GPS receiver, developed by the Jet Propulsion Laboratory (JPL), was used by the CHAMP satellite. Compared to GPS/MET, an improved occultation rearward-facing antenna with a better gain of +5 db was used. There were about 230 daily GPS RO events recorded by CHAMP, 64% of them (approximately 160) could be successfully converted to quality atmospheric profiles (Wickert et al., 2006). High quality CHAMP RO observations were obtained, and the general mean differences with existing weather data for temperature were less than 0.4 K for a range of altitude between 10 km and 35 km. CHAMP produced seven and half years (2001 - 2008) of quality atmospheric profiles and it had the longest RO observation data set from one satellite (Foelsche et al., 2006). The CHAMP RO observation has been used in a number of climatological studies including those presented in Chapter 6.
FORMOSAT-3/COSMIC mission
The FORMOSAT-3/COSMIC mission was designed to assist with weather forecasting, climate monitoring, and atmospheric, ionospheric and geodesy research by employing a constellation of six LEO satellites (Fong et al., 2009b). The final orbits of the six satellites are between 750 and 800 km with an inclination of 72˚ in 6 planes with a spacing of 24˚. Three ground stations were established in Alaska, Sweden and Taiwan respectively and one multiple task station is located in Taiwan for communicating with the LEO satellites.
Each FORMOSAT-3/COSMIC satellite is equipped with a GPS RO payload for tracking GPS signals, a tiny ionospheric photometer payload for measuring ionospheric total electron content (TEC) from the satellite’s nadir direction, and a tri-band beacon payload for generating high resolution satellite to ground-station TEC (Anthes et al., 2008). The FORMOSAT-3/COSMIC GPS receiver is a NASA/JPL “BlackJack”
42 space-borne GPS receiver. This type of receivers was used by CHAMP and other early GPS RO programs. The integrated GPS receiver system consists of five units, namely a scientific grade GPS RO receiver, dual occultation antennae, dual precise orbit determination antennae, payload controller and a solid state recorder. With this robust space-borne GPS receiver system, the LEO satellites’ precise orbit could be accurately determined and both the rising and setting GPS RO events can be captured.
The success of the FORMOSAT-3/COSMIC constellation proved that the GNSS RO technique had a capability to not only generate high vertical resolution atmospheric profiles but also provide a good horizontal resolution with global coverage (Huang et al., 2009). It has demonstrated that GNSS RO data was a useful technique for meteorological agencies.
4.3 Simulation study of future RO missions in the Australasian region
4.3.1 Next generation GNSS
The GPS is currently the dominant GNSS and to date all RO missions have relied on receiving only GPS signals. However, in the era of GPS modernisation and significant GNSS expansion, new satellites with new signals have been developed (Gleason and Gebre-Egziabher, 2009). Within the next 10 years, about one hundred new GNSS satellites will be launched and thus new signals will be transmitted from them (Gao, 2008). These satellites will incorporate modernised GPS and GLONASS systems, plus newly designed Galileo and Compass systems. There are also some regional GNSS such as QZSS from Japan (Japan Aerospace Exploration Agency, 2009) and the Indian IRNSS. The next generation RO missions (e.g. FORMOSAT-7/COSMIC-2) will be equipped with new GNSS RO receivers (e.g. TriG) which are designed to receive signals from these new satellites (Meehan, 2009). The horizontal and temporal resolution of RO profiles will be significantly improved due to the increased number of satellites and the improved signal strength and signal spectrum. The quality of the RO technique will be also improved because of the increased number of radio frequencies available for the retrieval processing. These new GNSS will utilise different radio
43 frequencies and hence future GNSS RO missions will take advantages of the different signals to improve the retrieval process for better data quality.
The simulation study, undertaken in this research, was designed to model the number and distribution of RO events obtained from FORMOSAT-7/COSMIC-2 using various GNSS (i.e. GPS, GLONASS, Galileo and QZSS). The results show that the number of daily RO retrievals in the Australasian region will increase from a few hundreds to over three thousands due to the use of multiple GNSS. FORMOSAT-7/COSMIC-2 will also provide more RO events in the tropical regions where the atmosphere is more dynamic and consequently high vertical and horizontal resolutions, as well as the high temporal resolution will improve our knowledge about the tropical atmosphere. The spatial distribution (over various latitude zones) of these RO events is analysed. The results of this research provide a reference for Australian meteorologists and climatologists who intend to use the RO data in their research and applications.
Recent GNSS development
The success of GPS as a critical space-based positioning infrastructure has lead to the rapid development of other GNSS (Jacobson and Ebooks Corporation., 2007). GLONASS is currently the only alternative global operational satellite positioning system. The GPS has 24 nominal satellites but currently there are 30 satellites in orbit (2011). The European Union proposed its own GNSS (named Galileo) in the 1980s and it has been designed as a service-oriented positioning system. Two testing satellites Galileo In-Orbit Validation Element (GIOVE) test satellites A/B have been launched and the first operational satellite is scheduled for launch in 2012. China has launched seven Compass satellites and is on a fast track to establish its own GNSS. The Chinese system has a more complicated constellation. It will have 27 satellites in a medium earth orbit (altitude of about 21,500 km), 5 satellites in a geostationary orbit (altitude of about 36,000 km), and 3 satellites on the inclined geosynchronous orbit (altitude of about 36,000 km).
A regional augmentation system has been designed by Japan (named QZSS) and it has three satellites. It will provide enhanced navigation services by emitting compatible GPS navigation signals (L1C/A, L2C and L5C) and augmentation information for GPS corrections (Japan Aerospace Exploration Agency, 2009). The first QZS-1 was
44 launched in 2010. The QZSS satellites fly over Japan, eastern Asia and the Australian region and will provide better opportunities of utilizing GNSS RO for these areas (Figure 4-2).
Figure 4-2 QZSS orbit (Japan Aerospace Exploration Agency, 2009)
Table 4-1 A summary table of the four major GNSS (i.e. GPS, GLONASS, Galileo and Compass) and regional system (i.e. QZSS) GPS GLONASS Galileo Compass QZSS Nominal 24 24 30 35 3 Number of satellites Orbit 20,200 km 19,100 km 23,222 km 27 in medium 39,000 km altitude earth orbit, 5 in geostationary orbit, and 3 in geosynchronous orbit Carrier L1: 1,575 MHz G1: 1,602 MHz E1: 1,575 MHz E1: 1,557 MHz L1: 1,575 MHz frequencies L2: 1,227 MHz G2: 1,246 MHz E5a: 1,176 MHz E2: 1,561 MHz L2: 1,227 MHz L5: 1,176 MHz G3: 1,204 MHz E5b: 1,207 MHz E6: 1,268 MHz L5: 1,176 MHz E6: 1,268 MHz E5b: 1,207 MHz
Table 4-1 summarizes the details of four major GNSS and one regional system (i.e. QZSS) including the nominal number of satellites in their constellation, orbit altitudes, and carrier frequencies etc.
45 4.3.2 RO event simulation study with next generation multiple GNSS
FORMOSAT-3/COSMIC is the first GPS RO mission that employs a six-satellite constellation for RO observations and it provides approximately 2,500 daily GPS RO events globally. Figure 4-3 is the snapshot of six hour ROE on 23 rd August 2010. This is comparable to the number of atmospheric profiles obtained by the global radiosonde network which has about 2000 stations worldwide. On average, about 300 daily RO events can be observed within the Australasian region (see Figure 4-4). This is an order of magnitude greater than the number of atmospheric profiles obtained by the 38 Australian radiosonde stations. The EGOPS (End-to-End GNSS Occultation Performance Simulator) software developed by the University of Graz is used in this simulation study (Kirchengast et al., 2007). Two line element sets (TLE) of both GNSS and LEO satellites are obtained from the Centre for Space Standards and Innovation and used for satellite orbit information in the simulation study. L1 and L2 GNSS frequencies were simulated in the study.
Figure 4-3 Distribution of FORMOSAT-3/COSMIC RO events in a period of six hours (3pm - 9pm) on 23rd August 2010 (retrieved from FORMOSAT-3/COSMIC UCAR website http://cosmic-io.cosmic.ucar.edu/cdaac/index.html on 23rd August 2010)
46
Figure 4-4 Distribution of simulated daily FORMOSAT-3/COSMIC RO events in the Australasian region with GPS only and the 38 Australian radiosonde stations
FORMOSAT-3/COSMIC is designed with an expected life of about 5 years and the mission is expected end to around 2014/2015. The follow-on mission, FORMOSAT-7/COSMIC-2, has planned to launch 12 GNSS RO satellites in 2014. This mission will maintain a continuous quality of RO data. The FORMOSAT-7/COSMIC-2 satellites will employ a TriG RO receiver (developed by JPL) which is able to receive signals from GPS, GLONASS and Galileo and it should provide approximately 12,000 global RO events each day.
With the rapid development next generation of GNSS, new RO missions will take atmospheric observation technique and research to a new era by
• Increasing the number of transmitting and receiving satellites that will improve the coverage and horizontal resolution of the RO observations;
• Providing additional carrier frequencies (e.g. L5 and E1) that will improve the atmospheric retrieval process such as ionospheric corrections;
• Using new signals with more transmission power that will enhanced the GPS Signal-to-Noise ratio, consequently improving the RO retrieval process in the lower atmosphere; and
• Adding new regional systems (e.g. QZSS) that will improve distribution of RO events in the region of interest.
47 Figure 4-5 and Figure 4-6 illustrate the simulated distributions of daily RO events from both FORMOSAT-3/COSMIC and FORMOSAT-7/COSMIC-2 with GPS, Galileo, GLONASS and QZS-1. The expected increased density of RO profiles from these missions which employ the future GNSSS is clearly demonstrated.
Figure 4-5 Distribution of simulated daily FORMOSAT-3/COSMIC RO events in the Australasian region with GPS, Galileo and GLONASS
Figure 4-6 Distribution of simulated daily FORMOSAT-7/COSMIC-2 RO events in the Australasian region with GPS, Galileo, GLONASS and QZS-1
The daily RO events from FORMOSAT-7/COSMIC-2 using the major GNSS (i.e. GPS, Galileo, GLONASS and QZSS) will be in excess of 3,000 (Figure 4-7). The
48 FORMOSAT-7/COSMIC-2 constellation is designed to provide a uniform coverage of the global with an increased number of observations in equatorial and tropical regions compared to FORMOSAT-3/COSMIC configurations. It will be achieved with the launch of 6 satellites at an inclination of 72˚ orbit and 6 satellites at an inclination of 24˚ orbit.
# of ROE 3500
3000
2500
2000
1500
1000
500
0 0 — -10 -10 — -20 -20 — -30 -30 — -40 -40 — -50 -50 — -60 Total Latitude zones
COSMIC (GPS) COSMIC (GPS+Galileo+GLONASS) COSMIC II (GPS+Glaileo+GLONASS) COSMIC II (GPS+Glaileo+GLONASS+QZSS)
Figure 4-7 Simulated number of daily RO events from FORMOSAT-3/COSMIC and FORMOSAT-7/COSMIC-2 with future GNSS in the Australasian region at different latitude zones
4.4 Study of the RO data processing
Weather forecasting, and most meteorological research, relies on accurate real-time atmospheric information. In response to extreme weather events, real-time observations and quick data processing and analysis can lead to significant social and scientific benefits. However, real-time atmospheric observations and data processing create critical challenges for IT infrastructure and software engineering. With the rapid development of this monitoring technique, the quantity of the data available is increased dramatically and as a consequence the demand for robust data storage and processing solutions is increased. With the GNSS RO technique, the high sampling rate means that a large amount of data will be generated every hour. The limited
49 computation, storage and power consumption on the micro-satellite restricts the onboard processing. The number of ground communication stations is also a limited resource. These are often shared by hundreds of satellites. Therefore, the RO data centres will be required to store a vast quantity of data plus produce atmospheric information in near real time in order to meet the needs of many meteorological users.
4.4.1 Introduction of the RO data processing
The GNSS RO signals are captured by and stored in the LEO satellites. The LEO satellite flies over one of the ground communication stations every 100 minutes where data are downloaded. These are then sent to the central data processing centre directly. In order to retrieve the atmospheric information from the raw GPS signal measurements, various input data from other sources are necessary and then a number of steps in data processing and analyses are required.
First, precise satellite orbits need to be calculated based on the reference data from other data service centres. The input reference data required are in near real-time and of a large volume. In general, both the GPS RO raw data and input data are stored in binary files. Computations in this step are also intensive.
The bending angle of the GPS RO signals and hence the refractivity of the atmosphere requires the precise satellite orbit information as the first step. Due to the physical characteristic of the atmosphere and its impact on the GPS signals, computations are conducted at different portions of the atmosphere, i.e. ionosphere, stratosphere and troposphere. The bending angle and refractivity profiles from these different atmospheric layers are retrieved. These profiles can be used directly for meteorological research and operations. Most of the data files are in common NetCDF or BUFR binary formats.
The most important atmospheric characteristics needed for general meteorological use are the temperature, pressure and water vapour. In order to separate these profiles from refractivity and bending angle profiles, background information is required at different layers of atmosphere. The background information is obtained from other meteorological observations and models. The final products can be stored in various file formats for different purposes.
50 The UCAR COSMIC Office offers RO products in four classes (Levels 0, 1A, 1B and 2). Level 0 is raw GPS measurement data while Level 2 contains the final products including atmospheric refractivity, temperature, pressure and water vapour pressure profiles. Intermediate products (Levels 1A and 1B) consist of atmospheric phase delay, signal amplitude and atmospheric Doppler shifts and atmospheric bending.
The current processing delay is about 2 hours. The delay is mainly due to the limited number of ground communication stations. Data are downloaded to the stations every 100 minutes and additional approximately 30 minutes are needed for data processing. This is real-time processing for high priority applications such as forecasting and cyclone tracking. The post-processing, which requires higher quality of satellite orbit data and other background data, takes more time.
4.4.2 RO data processing centre
The Australian region can gain the maximum advantages from this space-based technology because 1) it is located in the Southern Hemisphere where a very limited number of weather observational stations exist compared with the Northern Hemisphere, 2) it has large unpopulated areas where the cost of establishing ground-based weather stations is high, and 3) it owns two satellite communication stations located in the Southern Hemisphere. Australia has unique geographical advantages which could permit it to play an important role in satellite Earth observations. In December 2002, Australia launched an LEO Earth observation satellite with a GPS RO unit FedSat (Federation Satellite) onboard respectively. However, no useful RO information was received. It is important for Australia to continue its research efforts and development in this area. This could advance Australian meteorological applications and provide a better response to climate change challenges and also enhance the Australian reputation in the international space industry. Establishing an Australian GNSS RO data processing centre could be a core component of the future endeavour in this scientific area.
Another challenging aspect is the complexity and cost of the Information Technology and Communication (ITC) to support meteorological information, especially in this era of explosion usage of satellite observations. As discussed earlier, the future GNSS RO
51 will utilise hundreds of satellites and generate thousands of atmospheric profiles daily. Moreover, the atmospheric observations and analyses are always complicated by the extreme weather events and the long-term needs to respond to climate change. More observations along with faster processing capacity are the continuous demand from meteorologists. It should be noted that GNSS RO profiles will not be used alone but in conjunction with other observations. All data will then be assimilated into various meteorological models. The Australian GNSS RO data processing centre, therefore, will need to access the downloaded satellite data directly, to process the data with short latency, and also integrate seamlessly with the BoM’s information infrastructure.
Figure 4-8 shows the proposed system structure for the RO data processing. The GNSS RO data centre will receive the LEO data (including LEO navigation observations and GNSS RO observations) from the ground satellite communication stations. There will also be other necessary inputs for the data retrieval process including the GNSS satellite precise orbits, GNSS ground-station observations and meteorological background data. These datasets will be stored in the database and be ready for processing. The data retrieval process algorithms produce a number of atmospheric data products (L0, L1a, L1b and L2 products) at various stages and for different purposes (i.e. POD processing, bending angle, ionospheric and atmospheric retrieval, and assimilation processes). The production interface is required to exchange information with other meteorological models and allow users to extract data in a customized format.
The main challenges for the GNSS RO data processing centre are the large amount of data involved and the critical requirement in real-time processing for meteorological applications. In order to address these challenges, comprehensive investment in the ITC infrastructure, data model and database, as well as the software engineering will be required. “Cloud” computing based infrastructure has a number of advantages that meet such ITC infrastructure needs because it provides cluster computation and mega storage capability plus the flexibility for extension. High volumes of dimensional data and Spatio-temporal information are key issues in data modelling and database management. The mega volume of the data increases the complexity of the data storage and the computational process. The data model/database and related techniques (e.g. spatial and temporal indexing) play a critical role in the efficient data maintenance and data processing. Similarly, software engineers often experience difficulty to handle the
52 multi-dimensional information in an efficient manner, especially for real-time applications. The GNSS RO data processing presents significant challenges in computer engineering and requires comprehensive and robust solutions.
• GNSS previse orbit LEO LEO LEO • GNSS ground observations • Meteorological background • LEO navigation observations • GNSS RO observations 3rd party data Ground Station Ground Station Ground Station
Data ingest and maintenance
Real-time Post POD POD Product Achieve database Banding angle retrieval database (L0, L1a, Ionosphere and atmospheric L1b and L2) profiles retrieval Assimilation process
Production interface
Figure 4-8 RO data processing system structure
The rapid development of satellite missions has generated large volumes of quality data and these data are creating new opportunities for many research communities. Because all atmospheric data obtained are valuable, there is an additional issue of utilizing the data across all GNSS missions. Often satellite missions have their own data format, data processing software, infrastructures, and their own user communities. Sharing of the infrastructure, software and data is difficult but is important for efficient research. There is a need to standardise the satellite data formats, software framework and ICT infrastructure in order to share resources and reduce the implementation and maintenance costs. This is a vital consideration for all future satellite RO missions.
4.5 Summary
This chapter presented the history, current status and future of GNSS RO missions and the increasing importance of the technique in global meteorological research
53 communities. Many meteorological agencies have started to utilise this space-based atmospheric observational technique and even develop their own GNSS RO satellites. Australia is likely to be one of the counties which will benefit greatly from such satellite technology due to its large unpopulated areas and its critical geographical position for input into world meteorology. A simulation study has been conducted to study the Australian opportunities from the GNSS RO technique using the future multiple GNSS technology. Over three thousands GNSS RO retrieved atmospheric profiles can be expected daily for the Australasian region. The amount of data will make an important contribution to Australian meteorological studies and also to the global context. This chapter also discussed the computational challenges of establishing an Australian GNSS RO data process centre a systematic skeleton of the data processing centre is investigated. This would provide Australia with direct access to the technique and data. A comprehensive computer engineering solution is required to address challenges in the ICT infrastructure, data model/database, and software. It is vital to invest in such satellite data processing centre to enhance Australian capabilities not only in the meteorological studies but also in many other space related studies and industries.
54 Chapter 5 Evaluation of the GNSS RO Retrievals for the Australasian and the Antarctic Regions
5.1 Introduction
As mentioned earlier radio occultation technology, for probing atmospheric information, has been utilised in a number of planetary discovery missions since the 1960s and for the Earth atmospheric observations since 1995. However, studies in incorporating the new atmospheric data into existing meteorological operational systems and applications are still in an early stage for meteorological communities. One reason for this delay is because of the technical difficulties involved in the retrieval processes for the lower atmosphere (i.e. from surface to the lower stratosphere). This portion of the atmosphere however, is the key region of interest for meteorologists and climatologists. The technical difficulties and the recent international research have been discussed in Chapter 3. Another challenging factor is due to the complexities of the atmospheric systems and the meteorological models. Many current meteorological systems will be influenced by the large quantities of valuable data introduced by GNSS RO technology. Therefore, comprehensive and detailed evaluations of the new RO data need to be conducted. The characteristics of the datasets must be studied more thoroughly to maximise the benefits of the new technology and to avoid/reduce problems that might be introduced into the current systems and models.
The results obtained from evaluation studies of GNSS RO observations are of particular interest to the Australian meteorological community (such as BoM) as the new technology will be introduced into practical usages for the first time. A series of evaluation studies focusing on the key research areas related to Australia (i.e. the Australian continent, the surrounding oceans and the Antarctic region) have been conducted through collaborative projects between RMIT and the Australian BoM since 2004. In these studies, atmospheric profiles obtained from CHAMP and FORMOSAT-3/COSMIC RO missions were evaluated. These profiles processed by two different data centres (i.e. JPL and UCAR) using independent data processing systems have been evaluated. Radiosonde measurements (from the Australian BoM and
55 the WMO) and NCEP data were used as a comparison with that GNSS RO derived atmospheric profiles.
Four evaluation studies were designed and conducted to assess the different aspects of the RO data compared with different datasets over various study areas and periods. The results of these studies have demonstrated good agreements between GNSS RO retrievals and radiosonde data and NWP model data (Fu et al., 2007; Fu, 2008; Fu et al., 2009b). The details of these studies are presented and the outcomes are discussed in this chapter.
5.2 Datasets used
Australian Radiosonde Observations
Australia is one of the key members of the WMO and it maintains 16 Global Climate Observing System Upper-air Network (GUAN) stations (Australian Bureau of Meteorology, 2002). These stations (including three in Antarctica) provide consistent high quality upper-air information about the sate of the atmosphere. The data from these stations are the basis for many Australian meteorological applications and are available to all WMO national members. The Australian GUAN stations utilise 800g balloons which can reach altitudes up to 30 km. Atmospheric information of temperature, pressure and wind are measured throughout the network. In addition to the 16 GUAN stations, Australia also maintains another 22 radiosonde stations that can observe the upper-air atmosphere at a comparable standard. Australia has been consistently using Vaisala radiosonde system at all the 38 stations.
Figure 5-1 shows the locations of the 38 Australian radiosonde stations. Most of these stations are located at airports within cities on the Australian mainland and a few in the central remote areas. There are also three stations in Antarctica, four at remote Australian islands and one in Tasmania.
56
Figure 5-1 Locations of the 38 Australian radiosonde stations with their WMO ID
For these evaluation studies, radiosonde data from these 38 stations were obtained from the Australian BoM. The data contained profiles of air temperature (ºC), dew point temperature (ºC), relative humidity (%), pressure (hPa), geopotential altitude (gpm to nearest 0.1 m) and their quality indicators. Relative humidity is calculated from the dew point temperature and the air temperature. All the other parameters are direct observations. The Australian radiosonde stations are routinely operated four times per day at 00:00, 07:00, 13:00 and 19:00 hour (local time) respectively. However, the reality is that the number of observations varies from day to day due to many technical or non-technical reasons, including device faults and difficult weather conditions. Table 5-1 lists the station IDs, names and their elevations above mean sea level (created based on Australian Bureau of Meteorology, 2002).
57 Table 5-1 A list of 38 Australian radiosonde stations with their WMO station IDs, names and elevations WMO Elevation WMO Elevation Station name Station name ID (meters) ID (meters) SYDNEY AIRPORT 94203 BROOME AIRPORT 7.4 94767 6 AMO
PORT HEDLAND 94312 6.4 94294 TOWNSVILLE AERO 4.3 AIRPORT
LEARMONTH WAGGA WAGGA 94302 5 94910 212 AIRPORT AMO
MEEKATHARRA 94430 517 94170 WEIPA AERO 18 AIRPORT
GERALDTON WILLIAMTOWN 94403 33 94776 9 AIRPORT RAAF
WOOMERA 94610 PERTH AIRPORT 15.4 94659 166.6 AERODROME ADELAIDE 94802 ALBANY AIRPORT 68 94672 2 AIRPORT
94638 ESPERANCE 25 94578 BRISBANE AERO 4.5
CHARLEVILLE 94647 EUCLA 93.1 94510 301.6 AERO
KALGOORLIE-BOU 94637 365.3 94711 COBAR MO 260 LDER AIRPORT
GILES 94461 METEOROLOGICAL 598 94120 DARWIN AIRPORT 30.4 OFFICE
94150 GOVE AIRPORT 51.6 89611 CASEY 40
COCOS ISLAND 94975 HOBART AIRPORT 4 96996 3 AIRPORT
MELBOURNE 94866 113.4 89571 DAVIS 18 AIRPORT LORD HOWE 95527 MOREE AERO 213 94995 5 ISLAND AERO
MOUNT GAMBIER MACQUARIE 94821 63 94998 6 AERO ISLAND
94332 MOUNT ISA AERO 340.3 89564 MAWSON 9.9
NOWRA RAN AIR NORFOLK ISLAND 94750 109 94996 111.7 STATION AWS AERO
ROCKHAMPTON 94374 10.4 94299 WILLIS ISLAND 8.1 AERO
58 Data Generated by Weather Models
Operational global analysis data from NCEP (National Centre for Environmental Prediction) was also used in the validation studies. A number of observation data are used in NCEP models as inputs, e.g. global radiosonde data, marine surface data, aircraft data, land surface data, satellite sounder data and satellite cloud drift winds (Kalnay et al., 1996). It produces at 1º × 1º grids in 26 levels from 1000 hPa to 10 hPa at six hourly intervals.
Data pre-processing
These evaluation studies employed very large quantities of data (hundreds of G-bytes) from different sources, e.g. the Australian Bureau of Meteorology, UCAR and JPL. These datasets were also in different formats, e.g. ACSII, CSV and NetCDF. In order to utilise all the data efficiently, a database system was designed and developed to manage and process the datasets as part of this research. The data system (shown in Figure 5-2) consists of 1) a database system with a number of tables for hosting the different datasets, 2) data pipe modules that can input data from different sources, and 3) data analysis modules for data processing and statistical analysis.
Database with •ASCII table structure •CSV A B •NetCDF Data Data Match Location Pipe Analysis •BUFF Match Time •Binary Modules Modules Match resolution •……
Figure 5-2 Database system developed for the GNSS RO data management and processing
The database was implemented in the open-source database management system MySQL 5.2. The data are organised in three levels, namely databases (for different data categories, i.e. RO data products and radiosonde data), tables that contain the raw data, and data views that are a logical structure of tables for operations. The designs of the tables and data views have also considered the size and usage of the datasets that are
59 needed for different datasets. The data frequently required are extracted into dedicated tables and/or data views. For example, RO title tables were created to contain the basic RO events information, e.g. latitude/longitude, time and satellite ID. Indexes are also created on tables to speed data retrieval.
The data pipe modules were written in different languages for different data formats. These languages include C++, SQL (Structured Query Language) and IDL (Interactive Data Language). These modules have been developed and updated during the research processes to handle the varying datasets. The data match modules were developed to generate comparable data pairs from the database. Since data were measured via independent instruments, they were matched using their 3D locations (position and altitude) and time. Due to the limited number of data available and the different resolutions used amongst these datasets, an interpolation mechanism is required. In most cases, this interpolation was obtained via an interpolation routine developed in IDL.
5.3 Evaluation studies – results and discussions
5.3.1 Evaluating CHAMP retrievals by using NCEP reanalysis
In this study, ten CHAMP RO profiles were found with the collocated NCEP weather prediction model generated data. These CHAMP RO atmospheric profiles (i.e. refractivity, temperature, pressure and water vapour pressure) were obtained from the Global Environmental & Earth Science Information System (GENESIS) developed by JPL.
The height ranges of these CHAMP RO events are between altitude of about several hundred of meters and about 30 km above the Earth’s surface. The vertical resolution of the CHAMP RO generated atmospheric profiles was about 300 meters. Ten CHAMP RO events were selected for this study. These RO events occurred in the Australasian regions (within the continent, around the coastal line and in the surrounding ocean areas) between April and November 2006 (see Figure 5-3).
60
Figure 5-3 Distribution, event number and date of the selected radio occultation events
Refractivity
The air density of Earth atmosphere varies with different altitudes. This results in atmospheric refraction of GNSS radio signals when they pass through the atmosphere. The study of atmospheric refractivity will provide us inside information about air density, pressure and temperature. Refractivity is the first atmospheric parameter retrieved from the GNSS RO signal measurements.
Refractivity profiles derived from the ten CHAMP RO events were compared with the collocated NCEP data. The mean difference in refractivity index between CHAMP RO Event #4 and its collocated NCEP gave the smallest difference of the ten study events, while Event #9 had the largest mean difference.
Figure 5-4 illustrates the comparison results of refractivity from CHAMP RO Event #4 and #9. Charts (a) and (c) in Figure 5-4 show the refractivity profiles (blue line with marks for CHAMP data and red line for NCEP data) at different altitudes, while Charts (b) and (d) display the differences between the two profiles. In general, CHAMP RO retrievals appear to agree very well with the NCEP data. The refractivity differences between NCEP and CHAMP RO in the troposphere are larger than those in the stratosphere. The differences in refractivity index between the two datasets can be over 15 units in the lower troposphere. The results also show that the CHAMP RO retrieved refractivity index is generally higher than those from the NCEP data.
61 (a) Refractivity profiles (c) Refractivity profiles (CHAMP RO Event 4 and collocated NCEP) (CHAMP RO Event 9 and collocated NCEP) 350 350
300 CHAMP 300 CHAMP NCEP 250 250 NCEP 200 200
150 150
100 100 Refractivity(Index) Refractivity(Index)
50 50
0 0 0.2 3.5 5.8 7.5 9.6 11.9 14.7 18.0 21.7 26.3 0.2 1.4 3.5 5.2 6.5 8.1 10.1 13.0 16.2 19.5 Altitude (km) Altitude (km)
(b) Differences of the Refractivity profiles (d) Differences of the Refractivity profiles (CHAMP RO Event 4 and collocated NCEP) (CHAMP RO Event 9 and collocated NCEP) 5 5
0 0
-5 -5
-10 -10 Refractivity Refractivity (Index) Refractivity Refractivity (Index) -15 -15
-20 -20 0.2 3.5 5.8 7.5 9.6 11.9 14.7 18.0 21.7 26.3 0.2 1.4 3.5 5.2 6.5 8.1 10.1 13.0 16.2 19.5 23.8 29.0 Altitude (km) Altitude (km) Figure 5-4 Refractivity profiles from the CHAMP RO events #4 and #9 against altitudes (compared with their collocated NCEP profiles) are shown in (a) and (c) respectively; and the differences between the two datasets are shown in (b) and (d).
Temperature
The CHAMP data description states that the CHAMP RO temperature should be replaced by the NCEP data when the temperature is warmer than -43 ºC in the troposphere. It should be noted that this study was conducted in 2007 using early CHAMP data while the data used in other studies are from other sources which do not have this limitation. Therefore, there are no differences between CHAMP and NCEP temperatures found in the lower troposphere. This is due to the general poor retrieval results of CHAMP RO in the lower troposphere. The mean difference between the Event #8 and collocated NCEP temperature profiles presented the smallest one of the ten study events, while Event #9 has the largest mean difference. In general, the temperature differences are within 2ºC and the CHAMP RO retrievals are cooler than NECP data (Figure 5-5).
62 (a) Temperature profiles (c) Temperature profiles (CHAMP RO Event 8 and collocated NCEP) (CHAMP RO Event 9 and collocated NCEP) 20 20
0 CHAMP 0 CHAMP NCEP NCEP
-20 -20
-40 -40 Temperature (C) Temperature Temperature (C) Temperature -60 -60
-80 -80 0.6 1.7 3.4 4.8 6.0 7.9 0.2 1.4 3.5 5.2 6.5 8.1 10.9 13.1 16.0 19.4 23.4 28.1 10.1 13.0 16.2 19.5 23.8 29.0 Altitude (km) Altitude (km)
(b) Differences of the temperature profiles (d) Differences of the temperature profiles (CHAMP RO Event 8 and collocated NCEP) (CHAMP RO Event 9 and collocated NCEP) 6 6
4 4
2 2
0 0
-2 -2 Temperature (C) Temperature Temperature (C) Temperature -4 -4
-6 -6 0.6 1.7 3.4 4.8 6.0 7.9 0.2 1.4 3.5 5.2 6.5 8.1 10.9 13.1 16.0 19.4 23.4 28.1 10.1 13.0 16.2 19.5 23.8 29.0 Altitude (km) Altitude (km) Figure 5-5 Temperature profiles from the CHAMP RO events #8 and #9 (with their collocated NCEP profiles) are shown in (a) and (c) against altitude respectively; and the differences between the two datasets are shown in (b) and (d) respectively.
Pressure and water vapour pressure
Figure 5-6 presents the comparison results for dry pressure over CHAMP RO events #3 and #9 and Figure 5-7 presents water vapour pressure over Event #4 and #5. Both the dry and wet pressure comparison results show a good agreement, especially in the stratosphere. Water vapour is a dominant factor for refractivity in the lower troposphere and it is difficult to separate the wet component from the pressure using the GNSS RO technique.
63 (a) Pressure profiles (c) Pressure profiles (CHAMP RO Event 3 and collocated NCEP) (CHAMP RO Event 9 and collocated NCEP) 1,000 1,000 900 900 CHAMP 800 CHAMP 800 NCEP 700 NCEP 700 600 600 500 500 400 400 Pressure (hPa) Pressure 300 (hPa) Pressure 300 200 200 100 100 0 0 0.6 2.7 3.8 5.6 7.5 9.8 0.2 1.4 3.5 5.2 6.5 8.1 12.5 15.5 18.3 21.5 25.7 30.0 10.1 13.0 16.2 19.5 23.8 29.0 Altitude (km) Altitude (km)
(b) Differences of the pressure profiles (d) Differences of the pressure profiles (CHAMP RO Event 3 and collocated NCEP) (CHAMP RO Event 9 and collocated NCEP) 6 6
4 4
2 2
0 0
-2 Pressure (hPa) Pressure -2 Pressure (hPa) Pressure
-4 -4
-6 -6 0.6 2.7 3.8 5.6 7.5 9.8 0.2 1.4 3.5 5.2 6.5 8.1 12.5 15.5 18.3 21.5 25.7 30.0 10.1 13.0 16.2 19.5 23.8 29.0 Altitude (km) Altitude (km) Figure 5-6 Pressure profiles from the CHAMP RO events #3 and #9 against altitude (with their collocated NCEP profiles) are shown in (a) and (c) respectively; and the differences between the two datasets are shown in (b) and (d) respectively
(a) Water vapour pressure profiles (c) Water vapour pressure profiles (CHAMP RO Event 4 and collocated NCEP) (CHAMP RO Event 5 and collocated NCEP) 18 18 CHAMP 15 CHAMP 15 NCEP NCEP 12 12
9 9
6 6 Pressure (hPa) Pressure Pressure (hPa) Pressure
3 3
0 0 0.2 3.5 5.8 7.5 9.6 0.3 1.9 4.8 6.8 9.5 11.9 14.7 18.0 21.7 26.3 12.4 14.7 18.2 22.2 27.0 Altitude (km) Altitude (km)
(b) Differences of the water vapour pressure profiles (d) Differences of the water vapour pressure profiles (CHAMP RO Event 4 and collocated NCEP) (CHAMP RO Event 5 and collocated NCEP) 2 2 1 1 0 0 -1 -1 -2 -2 -3 -3 -4 -4 Pressure (hPa) Pressure Pressure (hPa) Pressure -5 -5 -6 -6 -7 -7 0.2 3.5 5.8 7.5 9.6 0.3 1.9 4.8 6.8 9.5 11.9 14.7 18.0 21.7 26.3 12.4 14.7 18.2 22.2 27.0 Altitude (km) Altitude (km) Figure 5-7 Water vapour pressure profiles from the CHAMP RO events #4 and #5 against altitudes (with their collocated NCEP profiles) are shown in (a) and (c) respectively; and the differences between the two datasets are shown in (b) and (d)
64 Overall results
Atmospheric refractivity, temperature, pressure and water vapour pressure were analysed and evaluated in the Australian context for the ten selected CHAMP RO events. The results have demonstrated good agreements between CHAMP RO and NECP profiles. Table 5-2 summaries the statistical differences of the four atmospheric parameters derived from the CHAMP and NCEP data over the ten selected events. The mean difference in temperature is 0.34 ºC with a standard deviation of 1.11 ºC. The dry and wet pressures have also shown good results, i.e. for dry pressure (mean difference: 0.39 hPa and standard deviation: 0.62 hPa) and wet pressure (mean difference:-0.35 hPa and standard deviation: 0.88 hPa) respectively. Refractivity index values range between 0 and 320 units. The mean value of the differences for refractivity is only -1.62 units with a standard deviation of 4.06 units.
Table 5-2 Statistical differences of atmospheric parameters between NCEP and CHAMP profiles
Mean differences (Standard Deviations) RO Events Refractivity Water Vapour Temperature (ºC) Pressure (hPa) (index) Pressure (hPa)
#1 -1.98 (6.35) 0.167 (0.84) 0.85 (0.95) -0.47 (1.42)
#2 -1.32 (3.13) 0.487 (0.87) 0.63 (0.45) -0.27 (0.63)
#3 -0.71 (1.99) 0.520 (0.92) 0.01 (0.22) -0.14 (0.44)
#4 -0.50 (1.70) 0.320 (0.75) 0.41 (0.35) -0.10 (0.36)
#5 -2.35 (5.67) 0.497 (0.88) 0.65 (0.50) -0.53 (1.25)
#6 -2.07 (3.26) 0.812 (1.37) 0.44 (0.17) -0.45 (0.72)
#7 -1.71 (5.10) 0.644 (1.06) 0.58 (0.61) -0.37 (1.08)
#8 -1.39 (3.18) -0.002 (1.06) 0.39 (0.51) -0.31 (0.72)
#9 -2.43 (5.57) -1.225 (1.83) -2.17 (1.11) -0.42 (1.17)
#10 -1.70 (4.68) 1.220 (1.49) 2.12 (1.33) -0.43 (1.04)
Mean -1.62 (4.06) 0.344 (1.11) 0.39 (0.62) -0.35 (0.88)
65 There were four RO events (i.e. #2, #5, #6 and #10) that were located offshore (see Figure 5-3). There are no apparent differences found in the evaluation results between these four offshore events and the remaining land events. This suggests that the CHAMP RO meteorological observation technology can overcome the geographical limitation of conventional ground-based methods such as radiosonde observation stations. It includes cost savings for monitoring difficult-to-access areas such as the oceans.
Figure 5-8 shows the differences of mean differences and the maximum and minimum values of the four atmospheric parameters respectively between the two datasets. These differences are quite close to zero. For example, the temperatures in eight profiles have mean differences below 1 ºC and the differences of water vapour pressure are all very close to zero. The X-axis in Figure 5-8 is the geographical latitudes of the RO events.
Mean differences in refractivity Mean differences in temperature (Index) (with maximum and minimum bars) (C) (with maximum and minimum bars) 10 6 5 4 0 -5 2 -10 0 -15 -2 -20 -25 -4 -30 -6 -15.9 -18.0 -21.1 -25.2 -27.6 -28.4 -29.9 -35.1 -35.9 -45.5 -15.9 -18.0 -21.1 -25.2 -27.6 -28.4 -29.9 -35.1 -35.9 -45.5 Latitude Latitude
Mean differences in pressure Mean differences in water vapour pressure (hPa) (with maximum and minimum bars) (hPa) (with maximum and minimum bars) 8 2 6 0 4 2 -2
0 -4 -2 -6 -4 -6 -8 -15.9 -18.0 -21.1 -25.2 -27.6 -28.4 -29.9 -35.1 -35.9 -45.5 -15.9 -18.0 -21.1 -25.2 -27.6 -28.4 -29.9 -35.1 -35.9 -45.5 Latitude ( Latitude Figure 5-8 The mean differences of the four atmospheric parameters (i.e. refractivity, temperature, pressure and water vapour pressure) between CHAMP and NCEP. The error bars present the maximum and minimum differences between the two datasets.
66 5.3.2 Evaluation of FORMOSAT-3/COSMIC retrievals by using radiosonde
This study was designed to evaluate FORMOSAT-3/COSMIC GPS RO derived atmospheric profiles by using the Australian radiosonde data. The study area is bounded by latitudes [10˚S, 70˚S] and longitudes [75˚E, 170˚E]. This area covers all of the 38 Australian radiosonde stations geographically. There were 14,638 FORMOSAT-3/COSMIC RO events (Figure 5-9) recorded during a 3-month period from January 1 to March 31 2007. In this research, 42 collocated pairs were found by employing a 100 km radial distance buffer and a two hours temporal buffer based on the radiosonde records.
Figure 5-9 Distributions of the radiosonde stations (red triangles) and FORMOSAT-3/COSMIC RO events (green dots) for a period of three months from January 1 to March 31 2007
67 Mean differences in temperature (C) between COSMIC and radiosonde 3
2
1
0
-1
-2 -70 -60 -50 -40 -30 -20 -10 Latitude
Mean differences in pressure (hPa) between COSMIC and radiosonde 1
0
-1
-2
-3
-4 -70 -60 -50 -40 -30 -20 -10 Latitude Figure 5-10 Mean differences between FORMOSAT-3/COSMIC and radiosonde in temperature (upper plot) and pressure (lower plot) of the 42 study pairs against their latitudes
Temperature and pressure data derived from both radiosonde and FORMOSAT-3/COSMIC were compared for the 42 collocated samples. A good agreement between the two data sources has been found. 88% of the samples have mean temperature differences less than 1 ºC. The average of the mean temperature differences is about 0.1 ºC with a standard deviation of 1.5 ˚C. For pressure, 90% of the samples have mean differences less than 2 hPa and the average of mean differences is -1.1 hPa with a standard deviation of 0.9 hPa. Figure 5-10 shows the temperature (the upper plot) and pressure (the lower plot) mean differences for the 42 pairs against latitudes. The result suggests that FORMOSAT-3/COSMIC RO data generally have higher pressure than the radiosonde.
Figure 5-11 shows the differences of both temperature (the upper plot) and pressure (the lower plot) against altitudes with the 95% statistical confidence intervals. These estimates were obtained by transforming the variable in each case so that the assumptions required by the ordinary least squares estimation procedure for regression models (in particular the requirement of homoscedasticity for the residuals of the model)
68 were satisfied. Polynomials of sufficiently high degree were fitted, the prediction intervals calculated (that is, the confidence intervals for the individual response) and the inverse transformation applied to the fitted curve and the associated prediction intervals. The graphs shown in Figure 5-11 present the differences between FORMOSAT-3/COSMIC and radiosonde and patterns along their heights. The random errors of the FORMOSAT-3/COSMIC RO temperature retrievals along altitude are apparent since the mean difference line (green dotted line in the upper plot of Figure 5-11) is close to zero and nearly parallel to the 95% confidence interval lines. The pressure difference between FORMOSAT-3/COSMIC RO and radiosonde in the troposphere is larger than those in the stratosphere (shown in the lower plot in Figure 5-11). The result also indicates that FORMOSAT-3/COSMIC RO pressure is generally higher than the radiosonde (see the green mean difference line in the lower plot of Figure 5-11).
Figure 5-11 Differences in temperature (upper plot) and pressure (lower plot) between FORMOSAT-3/COSMIC and radiosonde against their altitudes and in the two plots, the green lines in the middle of each plot represent the mean differences and the blue and red lines are the 95% confidence intervals.
Spatial and temporal characteristics of the new data sources are vital for meteorological research and practical applications. Spatial representation maps illustrate the spatial
69 patterns for better interpretation of the geographical aspects of the FORMOSAT-3/COSMIC RO technique. In Figure 5-12, the temperature differences between FORMOSAT-3/COSMIC and radiosonde range between -0.9 ºC and 2.0 ºC. Most of the RO events studied (grey dots in Figure 5-12) gave a good agreement in temperature between FORMOSAT-3/COSMIC RO and radiosonde (i.e. differences < ±0.5 ºC), and only a couple of RO events (orange and red dots) have temperature differences larger than 1 ºC. It also appears that those FORMOSAT-3/COSMIC RO events with temperature differences larger than ±0.5 ºC were located at middle and high latitude areas (see Figure 5-12).
Figure 5-12 Temperature differences between radiosonde measurements and FORMOSAT-3/COSMIC derived values
Similarly, Figure 5-13 presents the pressure differences between FORMOSAT-3/COSMIC and radiosonde on map. The pressure differences between the two datasets range from -1.9 to 0.5 hPa. The results show that pressures derived from FORMOSAT-3/COSMIC are higher than radiosonde. Those events presented as grey dots in the map have differences smaller than 0.5 hPa between FORMOSAT-3/COSMIC RO and radiosonde.
70
Figure 5-13 Pressure differences between radiosonde measurements and FORMOSAT-3/COSMIC derived values
Continuous, accurate and high resolution measurements of the atmospheric profiles are important for numerical weather prediction analysis and climate related studies. GNSS RO derived atmospheric profiles have been considered as good data sources for atmospheric related research. In this study, the quality of the FORMOSAT-3/COSMIC data is assessed with statistical methods and the outcome of this study shows a very good agreement with the Australian regional radiosonde data. Such a large volume of atmospheric profiles obtained by GNSS RO technique will have a tremendous impact on the meteorological studies and applications. Most importantly, the GNSS RO derived atmospheric profiles are not restricted by geographic locations unlike the radiosonde technique (only 38 stations in Australia and most of them at coastlines). Therefore, the new data sources derived from the GNSS RO technique has a great potential to fill the gaps in current ground-based weather observational station networks.
5.3.3 Spatial analysis of FORMOSAT-3/COSMIC evaluation results
Another comprehensive evaluation study was undertaken to assess the FORMOSAT-3/COSMIC RO retrieved atmospheric profiles (temperature and pressure) in the whole Australian area for a thirteen-month period (from 1st July 2006 to 31st July 2007). Radiosonde records from 35 Australian radiosonde stations were used for this evaluation. A radial buffer of 100 km and a temporal buffer of 2 hours were chosen
71 to match the FORMOSAT-3/COSMIC profiles with the radiosonde records. A total number of 204 collocated pairs of FORMOSAT-3/COSMIC derived atmospheric profiles and radiosonde observations were identified (Figure 5-14).
Figure 5-14 Distributions of the 35 radiosonde stations (in red dots), 64,599 FORMOSAT-3/COSMIC derived atmospheric profiles (in dark dots) and 204 collocated pairs (in brown stars) for the period of 1st July 2006 to 31st July 2007.
Atmospheric temperature and pressure were analysed and evaluated. The overall results have demonstrated good agreements between the two datasets. Table 5-3 summaries the mean values and standard deviations of the differences between FORMOSAT-3/COSMIC retrievals and radiosonde observations. The mean difference in temperature is -0.43 ºC with a standard deviation of 1.53. The result shows that the FORMOSAT-3/COSMIC RO temperatures are generally warmer than radiosonde data. Pressure profiles also show a good agreement (mean at 1.39 hPa with standard deviation 1.71). It appears that FORMOSAT-3/COSMIC RO pressure values are generally lower than radiosonde measurements.
Spatial characteristics of the new data sources are vital for meteorological research and practical applications. In this study, the 204 pairs are classified into three geographical zones (i.e. inland, coastal and ocean) and three latitudinal zones (i.e. low altitude zone
72 [0°-23°S], middle latitude zone [23°-67°S] and high latitude zone [67°-90°S] respectively). It can be seen from Table 5-3 that the inland group has the smallest differences in both temperature and pressure when compared with the coastal and ocean groups. The high latitude zone (the Antarctic region) has the smallest differences in both temperature and pressure when compared with the low and middle latitude zones.
Table 5-3 The means and standard deviations of the differences (in temperature and pressure) between FORMOSAT-3/COSMIC and radiosonde.
Mean (Standard Deviation) Study Areas Temperature (ºC) Pressure (hPa)
Overall 0.43 (1.53) -1.39 (1.71)
Coastal 0.43 (1.48) -1.42 (1.80)
Inland 0.20 (1.77) -1.26 (1.44)
Ocean 0.48 (1.35) -1.94 (2.11)
Low Latitude Zone (0S to 23S) 0.30 (1.57) -1.78 (1.57)
Mid Latitude Zone (24S to 67S) 0.43 (1.53) -1.37 (1.87)
High Latitude Zone (68S to 90S) 0.29 (1.36) -0.03 (0.45)
The mean value of the differences in temperature between FORMOSAT-3/COSMIC and radiosonde and the standard deviations are represented in Figure 5-15. This map depicts that events in the coastal areas (i.e. eastern and western Australia) generally have larger differences and standard deviations in temperature than other areas.
73
Figure 5-15 Mean differences in temperature (and standard deviations) between FORMOSAT-3/COSMIC and radiosonde and the standard deviations of the differences are represented as density maps
Figure 5-16 Means of the differences in pressure (and standard deviations) between FORMOSAT-3/COSMIC and radiosonde and the standard deviations of the differences are represented as density maps
74 Figure 5-16 depicts a similar spatial representation of the pressure comparison results between the FORMOSAT-3/COSMIC and radiosonde measurements. It can be seen that both the inland and high latitude areas have smaller differences in pressure since these areas are relatively dry. The result also shows that events with larger mean variations and standard deviations were in the south-west and eastern Australia. It is interesting to note that those events where FORMOSAT-3/COSMIC has higher pressure than radiosonde are clustered in the south east and south west of Australia (see Figure 5-16).
Figure 5-17 shows the temperature differences (and corresponding standard deviations) between FORMOSAT-3/COSMIC RO and radiosonde against the altitudes, while Figure 5-18 shows the pressure result. It can be seen in Figure 5-17 that FORMOSAT-3/COSMIC data generally displays cooler temperatures at altitudes between 10 km and 20 km compared with radiosonde data. Larger pressure differences between the two datasets in the troposphere are found when compared with those in the stratosphere. The results also show that the FORMOSAT-3/COSMIC has generally lower pressure data compared with radiosonde observations.
Mean differences in temperature between COSMIC and radiosonde (C) with standard deviations 2
1
0
-1
-2 3.5 6 8.5 11 13.5 16 18.5 21 23.5 26 28.5 Height (km) Figure 5-17 Mean differences in temperature (and standard deviations) against the altitude from surface to 30 km with their standard deviations
75 Mean differences in pressure between COSMIC and radiosonde (hPa) with standard deviations 1
0
-1 0.5 3 5.5 8 10.5 13 15.5 18 20.5 23 25.5 28 -2
-3
-4 Height (km) Figure 5-18 Mean differences in pressure (and standard deviations) against the altitude from surface to 30 km with their standard deviations
5.3.4 Study of the co-location criteria
Evaluation of the GPS RO retrievals provides helpful information for the assimilation of the new data sources into the current meteorological systems and other applications. However, different evaluation studies use different co-location criteria to match the RO retrievals with atmospheric observational records. This study investigated the impact of different co-location criteria in terms of spatial and temporal separations (specifically, 100, 200 and 300 km radial buffers with 1, 2 and 3 hour temporal buffers). Radiosonde data from the 38 Australian weather observational stations (including three in the Antarctic) were used to compare the RO retrieved temperature profiles from both CHAMP (between 2001 and 2008) and FORMOSAT-3/COSMIC (between 2006 and 2009). The study regions were the Australasian region [between latitude 0-60º South and longitude 30-180º East] and the Antarctic region [latitude 60-90º South]. Table 5-4 summarises the statistical means, standard deviations and sample sizes of the temperature differences between the radiosonde measurements and RO profiles using different co-location criteria. In general, larger buffers (either spatially or temporally) result in greater differences in both mean and standard deviations. In order to test the differences among the three different temporal (1, 2 and 3 hours) and three different spatial co-location windows (100, 200 and 300 km), a two-way analysis of variance (ANOVA) is carried out. In two-way ANOVA, the total sum of squared deviations
3 3 n ( SS ) from the mean is calculated as jk (x − x )2 , where x denotes total ∑ j=1∑ k =1∑i = 1 ijk .. ijk
76 th th th the i temperature discrepancy at the j temporal and k spatial co-location, and x.. is the grand mean temperature discrepancy over all observations. The sum of squared deviations from the means ( SS ) due to temporal, spatial, their interaction and error, were however computed based on the summary statistics given in Table 5-4. For example the mean, standard deviation and the sample size from each cell, rather than the raw observations. The details of computation can be found from Cohen (2010). Mean squares ( MS ) are defined as sum of squares ( SS ) divided by their degrees of freedom ( DF ), which equal c-1 for c levels of effect. The test statistic F is defined as the ratio between two mean squares and the denominator equivalent to the mean squares error. The final decision was made based on the resulting P-value, which gives the probability of observing a test statistic greater than the observed F value when assuming no temporal or spatial effect.
Table 5-4 Means, standard deviations (STD) and the number of pairs (# Sample) of the temperature (ºC) discrepancies between radiosonde and RO data using different co- location criteria
Collocation 1 Hour 2 Hour 3 Hour criteria CHAMP COSMIC CHAMP COSMIC CHAMP COSMIC
Mean 0.37 0.34 0.37 0.35 0.35 0.35
100 STD 1.04 1.09 1.06 1.10 1.09 1.12 km
# Sample 93 341 170 672 239 1002
Mean 0.38 0.35 0.40 0.35 0.39 0.35
200 STD 1.10 1.16 1.13 1.17 1.15 1.18 km
# Sample 281 1266 602 2583 884 3851
Mean 0.39 0.36 0.40 0.36 0.39 0.37
300 STD 1.16 1.22 1.18 1.23 1.20 1.24 km
# Sample 556 2504 1189 5118 1776 7733
Table 5-5 shows the ANOVA results (i.e. sum of squares (SS), degrees of freedom (DF), mean squares (MS), F and P-value) of the temperature (ºC) discrepancies
77 between radiosonde and CHAMP data, as well as between radiosonde and FORMOSAT-3/COSMIC respectively. The results from the ANOVA test suggests that, at the three selected levels of temporal criteria (1, 2 and 3 hours), there are no statistical differences detected for the temperature discrepancies between radiosonde and CHAMP (P-value=0.956). The differences at the three levels of spatial buffers (100, 200 and 300 km) are also not statistically significant (P-value=0.834). The result also suggests that there is no interaction between the spatial and temporal effects (P-value=1.000). For the temperature discrepancies between radiosonde and FORMOSAT-3/COSMIC, there is no statistical difference detected at the selected range of spatial and temporal co-location criteria within 3 hours and 300 km (see Table 5-5).
Table 5-5 Analysis of variance results (i.e. sum of squares (SS), degrees of freedom (DF), mean squares (MSS), F and P-value) for temperature (ºC) discrepancies between radiosonde and CHAMP, and between radiosonde and FORMOSAT-3/COSMIC data.
CHAMP COSMIC
Source P-val F P-valu SS DF MS F SS DF MS ue e
Temporal 0.123 2 0.062 0.046 0.956 0.181 2 0.090 0.075 0.928
Spatial 0.490 2 0.245 0.182 0.834 1.337 2 0.669 0.555 0.574
Temporal 0.048 4 0.012 0.009 1.000 0.003 4 0.000 0.000 1.000 Spatial
7790. 30173. Error 5781 1.348 25061 1.204 297 364
7790. 30174. Total 5789 25069 958 880
In this study, FORMOSAT-3/COSMIC shows systematically smaller mean differences and standard deviations compared than the CHAMP RO (see Table 5-4). It suggests that FORMOSAT-3/COSMIC RO mission has a better agreement with radiosonde data than the earlier CHAMP mission.
78 Collocation using the 300 km spatial and 3 hour temporal buffers are further investigated, as this provided more pairs for the comparison between GPS RO data and radiosonde data compared with other co-location criteria. The mean difference between radiosonde and CHAMP data is 0.39 ºC with a standard deviation of 1.20 ºC, while the mean difference between radiosonde and FORMOSAT-3/COSMIC data is 0.37 ºC with a standard deviation of 1.24 ºC. There is no statistical difference detected between these two temperature differences (P-value=0.461). Figure 5-19 and Figure 5-20 show the evaluation study results for CHAMP and FORMOSAT-3/COSMIC atmospheric retrievals at the 16 pressure levels respectively. In these two figures, the 95% confidence intervals for the population mean temperature differences were obtained, along with the sample sizes at each pressure level.
All of the 95% confidence intervals for the mean differences contain the temperature difference of about ± 0.5 ºC. Therefore, the overall results show a significant statistical agreement between the GPS RO profiles (from both CHAMP and FORMOSAT-3/COSMIC) and radiosonde measurements. It can also be seen from both figures that the temperature profiles using GPS RO technique have a positive systematic difference between the 100 and 500 hPa pressure levels but a negative difference in the lower stratosphere (below 100 hPa pressure level) and troposphere (above 500 hPa pressure level) compared to the radiosonde data. At around the 100 and 500 hPa pressure levels, both the CHAMP and FORMOSAT-3/COSMIC data agree well with radiosonde observations. The results also show that the confidence interval ranges are generally wider at both very low and very high pressure levels than the middle pressure levels. There are a large number of the samples for evaluation.
79 Pressure levels 0
100
200
300
400
500
600
700 Mean
800 CI_low (95%) CI_high (95%) 900 Count/1000 1000 -0.5 -0.25 0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 Temperature differences and number of pairs (/1000)
Figure 5-19 CHAMP temperature profile comparison results (i.e. means, 95% confident levels and counts of comparison pairs divided by 1000) at standard pressure levels
Pressure levels 0
100
200
300
400
500
600
700 Mean 800 CI_low (95%) CI_high (95%) 900 Count/10000 1000 -0.5 -0.25 0 0.25 0.5 0.75 1 1.25 1.5 1.75 2
Temperature differences and number of pairs (/10000)
Figure 5-20 FORMOSAT-3/COSMIC temperature profile comparison results (i.e. means, 95% confident levels and counts of comparison pairs divided by 10000) at standard pressure levels
80 When compared Figure 5-19 and Figure 5-20, it appears that the overall confidence intervals in the FORMOSAT-3/COSMIC are more accurate than those in the CHAMP comparison. Once again it suggests that the FORMOSAT-3/COSMIC RO data agree with radiosonde better than the CHAMP RO data.
5.4 Summary
This chapter presented four various evaluation studies to assess the GPS RO retrievals from both CHAMP and FORMOSAT-3/COSMIC satellite missions using both radiosonde measurements and NWP model. Studies were focused on Australasian and the Antarctic regions which are areas of primary interests for Australian meteorologists. Overall, the satellite retrieved atmospheric profiles agree well with the profiles from both the radiosonde measurements and the NCEP weather forecasting model.
These evaluations show that GPS RO has poorer performance in the lower atmosphere (from surface to the lower stratosphere) about 6-8 km altitudes depending on the locations and the time of observations. The main reasons for this were discussed in Chapter Four. The results also suggest that the latest mission FORMOSAT-3/COSMIC provides better data than the older CHAMP mission, in terms of quantity and quality of profiles. Moreover, profiles generated from FORMOSAT-3/COSMIC have more detailed information in the lower atmosphere than CHAMP. Spatial characteristics of the evaluation results have been examined. No clear spatial patterns have been identified due to the limited number of RO events collocated with radiosonde observations. This also suggests that GPS RO technology can provide more information globally thus overcoming the geographical limitations of current weather monitoring networks. Research about the co-location criteria supported the evaluation studies that use spatial and temporal buffers to match the RO events with other datasets.
81 (This page intentionally left blank)
82 Chapter 6 Antarctic Temperature Trend and Australasian Tropopause Studies
6.1 Introduction
As discussed in the earlier chapters, the GNSS RO technique has global coverage and can overcome the inhomogeneous distribution of the group-based weather observational station network. The quality and resolution of GNSS RO data are comparable to radiosonde observations (see Chapter 5). Therefore, the GNSS RO technique can be a very important alternative to the current atmospheric observations. This chapter presents the climatological studies using the GNSS RO derived atmospheric information. These studies concentrate on two regions of interest, i.e. the Australasian and the Antarctic regions. Both regions have a very low population density (3.6 people per km 2 for Australia and 0.00007 people per km 2 for Antarctica) compared with other areas of the world (e.g. 86.7 per km 2 for Asia) (King and Turner, 1997; Sturman and Tapper, 1996). Due to the very low population density, weather and climatological information obtained from these areas is limited. Antarctica is key interests of Australian meteorologist as it plays important role to the Australian weather and climate.
A study was undertaken to examine the temperature trends over the Antarctic using both the radiosonde and GPS CHAMP data. A homogeneity trend analysis method was used for the first time to determine temperature trends from nine radiosonde stations located in Antarctica. Significant lower-stratospheric cooling and tropospheric warming have been found over the Antarctic for the past five decades (1956-2010) using radiosonde data. Temperature trends were also derived using CHAMP RO data collocated with these radiosonde stations and the results were compared with the radiosonde data. A study period of seven complete years, from 1st September 2001 (beginning of the austral spring) to 31st August 2008 (end of the austral winter) was selected for the analysis to ensure that equal number of seasons was presented in the time series. The CHAMP RO derived temperature trends were compared with radiosonde derived trends for the same period. The study demonstrated that the
83 CHAMP RO data provided comparable atmospheric information for climatological studies with the radiosonde measurements but better space coverage over the Antarctic. The CHAMP RO data were grouped into forty-five equal areas of 800km square over the whole Antarctic region and the temperature trends of these areas at different standard pressure levels were examined.
The tropopause is another important factor to weather and atmospheric phenomena (Seidel et al., 2001). Examining the tropopause structure and dynamics requires detailed vertical atmospheric profile information and is a challenging task (Fueglistaler et al., 2009; Randel et al., 2007). The GPS RO technique can retrieve high vertical resolution atmospheric profiles with a uniform global coverage. Therefore, enhanced tropopause information (i.e. better resolutions and distribution) can be expected from the new GPS-based technique (Lewis, 2009; Schmidt et al., 2005b; Steiner et al., 2009b). Australian meteorology and climate studies can take advantage of this technique as it has large unpopulated areas and also tropical areas which have a complex tropopause. A study was designed to retrieve tropopause heights over the Australasian region using GPS FORMOSAT-3/COSMIC RO temperature profiles. The FORMOSAT-3/COSMIC RO mission provides a high vertical resolution of upper air temperature profiles. The study results confirm that the FORMOSAT-3/COSMIC RO temperature profiles are able to show more details of the atmosphere and characterise the tropopause better than other observational methods.
6.2 Antarctic temperature change analysis
6.2.1 The Antarctic
The Antarctic continent has unique geographical and meteorological features. It is the only continent located in the Earth’s Polar area and it is totally within the Antarctic Circle (66°33 ′43 ″ South) (Johanson and Fu, 2007). Antarctica has an area of approximately 14 million square kilometres. It is the coldest, windiest and driest place on Earth. The thickness of average ice sheet is about 1.6 kilometres as a result of the high latitude and high average elevations (2,286 metres, higher than all other continents) (King and Turner, 1997). The ice sheet covers approximately 98% of the continent extending towards the Southern Oceans. The ice sheet itself also contributes to the
84 Antarctic weather as it reflects more solar radiation than other types of land coverage such as soil and vegetation.
The unique features of the Antarctica described above are important influence to the atmospheric/oceanic systems and circulations (Bargagli, 2007). Heat transportation between the tropics and the polar areas forms key energy circulations in the atmosphere. Antarctica contributes to the atmospheric circulations not only in the Southern Hemisphere but also on a global scale. Similarly, ocean water is cooling down rapidly in the Southern Ocean around Antarctica’s coast. This transfers warm water to the Antarctic area and then the cool water is cycled back to the tropics. Antarctica is the “engine” of both atmospheric and ocean energy circulations. Therefore, any changes of the Antarctic climate have direct impacts to all atmospheric and oceanic systems, and consequently influence the global (especially Southern Hemisphere) weather and climate (Roberts et al., 2010).
6.2.2 Temperature change study using radiosonde data
Changes in the temperatures over the Antarctic in the recent decades have been recorded in a number of studies. Variations (both warming and cooling) over different areas of the Polar Region south to 60°S have been reported. In the earlier studies, noticeable warming of the Antarctic Peninsula was observed (King and Turner, 1997; Turner et al., 2006). Recently, examining long-term trends in the monthly mean near-surface temperatures for 18 stations in Antarctica derived from the Reference Antarctic Data for Environmental Research (READER) project data set (Turner et al., 2006), it was found that 11 stations had warming trends and 7 had cooling trends. It was also noticed that only three stations show trends that are statistically significant: Faraday/Vernadsky on the peninsular, Novolazarevskaya in East Antarctic, which are both warming, and Amundsen-Scott at the South Pole, which has cooled. Also by examining changes in the Antarctic ice-sheet surface temperature since 1957, it was reported that significant warming extends well beyond the Antarctic Peninsula and covers most of West Antarctica. A warming rate exceeding 0.1°C per decade over the past 50 years was reported, which appears the strongest in the winter and spring (Steig et al., 2009).
85 Significant warming of the Antarctic winter troposphere has also been observed, based on radiosonde observations. Statistically significant increases in seasonal temperatures at many of the stations across the continent (including the coastal regions and at Amundsen-Scott) have been reported (Turner et al., 2006). Warming has occurred throughout the troposphere, with the maximum increase in temperature in the upper troposphere (UT), and maximum cooling in the lower stratosphere (LS).
Atmospheric temperature measurements in the Antarctic are performed at a small number of observational stations. For example, data from only nine stations, most of them in East Antarctica, were selected to be suitable for examining temperature trends in the UT and the LS (Turner et al., 2006). To obtain reliable information about temperature spatial distribution over the remote areas such as the Polar Regions, satellite remote sensing is essential. Using the Microwave Sounding Unit (MSU) observations from NOAA’s polar orbiting satellites, temperature variations over the Antarctic were analysed and it was reported that since 1979 the Antarctic troposphere has warmed in winter and spring and cooled in the summer and autumn seasons (Johanson and Fu, 2007). Note that the MSU-derived data cannot be used to analyse temperature changes over the near-polar area of the continent as there are no observations south of 82.5°S. In addition, the accuracy of the MSU temperature measurements over the polar regions were questionable due to the disagreement between the satellite observations and in-situ radiosonde data (Swanson, 2003).
Atmospheric temperature records can also be obtained using the emerging GNSS RO technique (Kursinski, 1997; Liou, 2010; Yuan et al., 1993). GNSS RO atmospheric observations provide global coverage, all-weather capability, long-term measurement stability and high-quality temperature measurements in the UT and the LS for a range of altitudes from about 5 to 25 km (Kuo et al., 2005; Rocken et al., 1997a). The high accuracy of the GNSS RO is of particular importance for reliable estimations of temperature variations over regions where radiosonde data are sparse or not available. It appears that the GNSS RO is currently the preferable satellite remote sensing methodology for the UT and the LS temperature monitoring. Detection of atmospheric temperature changes in the tropical UTLS using GNSS RO has been well documented (Steiner et al., 2009a). In this thesis, recent temperature trends in the Antarctic UT and the LS were estimated from GNSS RO observations.
86 6.2.3 Study datasets
Firstly, results derived from GNSS RO observations were compared with radiosonde temperature records. GNSS RO data were obtained from UCAR (http://www.cosmic.ucar.edu/). Raw radiosonde data for 38 Australian stations (including the three Antarctica stations, Casey, Davis and Mawson) were provided by the National Climate Centre, Australian BoM. In addition, radiosonde data from the other six Antarctic stations (Amundsen-Scott, Bellingshausen, Halley, McMurdo, Novolazarevskaya and Syowa), were obtained from the HadAT2 dataset from the Hadley Centre, UK Met Office (http://hadobs.metoffice.com/hadat/).
The changes in the UT and LS temperatures over the Antarctic region (south of latitude 60°S) were investigated. Temperature trends were examined using data from the available CHAMP mission from May 2001 to August 2008. The CHAMP temperature data is the level 2 data product from UCAR. It is worth to note that the data was generated by new software package which the maximum of solar activity from 2001 to 2003 has been taken into account. To investigate variations in the UT and LS temperatures, annual, seasonal and monthly mean temperatures at 600, 500, 400, 300, 200, 100 and 50 hPa pressure levels were derived from the CHAMP RO data. The four seasons defined are summer (December, January and February [DJF]), autumn (March, April and May [MAM]), winter (June, July and August [JJA]) and spring (September, October and November [SON]). Temporal and spatial UT and LS temperature variations were examined by calculating annual, seasonal and monthly temperature trends at each of the 5 degree latitude/longitude grids grids from latitudes 60°S to 90°S and for each of the studied pressure levels.
6.2.4 Temperature trends analysis
The long-term temperature trends over the Antarctic were examined from radiosonde data by using a multiple regression method. This method takes into account the variations and number of observations in each month. Raw radiosonde measurements from three Australian stations (i.e. Casey, Davis and Mawson) were obtained from the Australian BoM. Mawson station’s data are available from 1995 to 2010 while Casey
87 and Davis stations have data from 1958 to 2010. It can be seen that there are some gaps in the data set (Figure 6-1).
Temperature trends were calculated at the two key pressure levels (100 hPa and 500 hPa). The results show a general cooling trend at 100 hPa and a general warming trend at 500 hPa for all three stations from 1950s to 2010 ( Figure 6-1).
Figure 6-1 Radiosonde temperature trend analysis over three Australian stations in Antarctic
Radiosonde temperature data for eighteen weather monitoring stations in Antarctica were obtained from UK MetOffice. The data are monthly averages and the various record periods are between 1950 and 2010. Figure 6-2 shows the temperature trend analysis results at the 100 hPa pressure level. Nine stations with longer term records were selected for further temperature trend analysis presented in the next section. Downward (cooling) trends were found over all 18 stations (Figure 6-2). At 500 hPa pressure level, fourteen out of eighteen stations present upward (warming) trends (Figure 6-3). The stations with cooling trends are Mirnyj, Neumayer, Base Marambio and Domont Durville.
The temperature trends derived from both raw BoM data and monthly UK MetOffice data show a good agreement for the three Australian stations (i.e. Casey, Davis and
88 Mawson). The warming in the UT and cooling in the LS in the Antarctic have been confirmed in this study.
Figure 6-2 Radiosonde temperature trend analysis by using 18 Antarctic radiosonde observations at the 100 hPa pressure level
Figure 6-3 Radiosonde temperature trend analysis by using 18 Antarctic radiosonde observations at the 500 hPa pressure level
89 6.2.5 Homogeneity study of temperature time series
Homogeneity treatment of the climate time series
Estimation of atmospheric temperature trends over Antarctica is challenging. Long-term upper air data records are available from only a small number of observational stations. In addition, discontinuities are often unavoidable in long-term radiosonde data series due to changes in the types of instrumentation and data processing methods. The largest discontinuities appear to be related to solar heating of the temperature sensor and the changes in its design plus the data adjustments intended to deal with this problem (Karl et al., 2006).
The homogeneity of the climate data series is crucial for many research aspects, especially for a realistic estimate of historical climate trends. Statistical homogeneity tests (Wang et al., 2007; Wang, 2008a, b; Wang and Feng, 2010) are used to analyse long-term radiosonde temperature series. Some discontinuities are found in the data (Table 6-1). The hour column indicates the hour(s) of radiosonde observations analysed. Here, H indicates that the time series found is homogeneous at a 5% significance level. These often occurred at the times where data are missing (see both Figure 6-4 and Figure 6-5). In this study, discontinuities are considered when estimating temperature trends (Wang, 2008a). The trend estimates presented in this study are also robust against the effects of autocorrelation (Storch, 1999) and also the incomplete sampling of the annual cycle.
90 Table 6-1 Times of discontinuities (yyyymm) detected in the monthly mean radiosonde temperature series (H indicates that the time series was found to be homogeneous at 5% significance level)
Station Names Hour 30 hPa 50 hPa 100 hPa 150 hPa 200 hPa 300 hPa 500 hPa 700 hPa 850 hPa
197605 197812 Amundsen-Scott 0 H 198405 H H H No data No data 198302 198308 197409 195712 195803 198303 198107 198709 Halley 12 199307 196603 198407 H 197012 197507 199201 199201 199202 199904 198405 198012 197402 Syowa 0, 12 H 198407 H H H H H H 198110 198411 195801 Mawson 0, 12 H 200201 195804 H 195803 195806 195806 H 196407 200211 197605 198111 199206 197503 198209 Davis 0, 12 H H H H H 199301 H 198801 198906 199504 199112 199507 199610 Casey 0, 12 H H H H H 198804 H H H
197010 200003 199006 McMurdo 0, 12 H H H H H H 200003 200304 199303
197204 197605 198106 198107 199206 Bellingshausen 0 H H H H H H 199408 198812 199309 199504 199610 199202 199604 Novolazarevskaja 0 H 200204 198407 H H 19611100 H H 200508 200211
A homogeneity test and linear trend estimation was performed for each of the nine stations and at each of the nine pressure levels. For each station, the homogeneity test results were synthesised across the nine pressure levels. The data series were also visually inspected; dates of missing data in the time series were also used to verify the change points detected statistically and then to help determine the actual times of the shifts. Artificial shifts will have significant effects on temperatures at some pressure levels, but have little effect on temperatures at other pressure levels due to the extremely large temperature differences amongst the different pressure levels.
91
Figure 6-4 De-seasonalised series (i.e. anomalies from the annual cycle) of monthly mean atmospheric temperatures at the 500 hPa pressure level, recorded at the stations, and the corresponding multi-phase regression fits (red lines, which include the annual trends)
92
Figure 6-5 De-seasonalised series (i.e. anomalies from the annual cycle) of monthly mean atmospheric temperatures at the 100 hPa pressure level, recorded at the stations, and the corresponding multi-phase regression fits (red lines, which include the annual trends)
93 Temperature trends at different pressure levels
Based on the results of the trend analysis described above, there appears to have been significant tropospheric warming (at all pressure levels from 850 to 500 hPa) and lower stratospheric cooling (at all pressure levels from 200 to 30 hPa) at all nine stations analysed. The regional mean vertical profile of annual trends (estimated from monthly mean temperature series) is presented in Figure 6-6. It should be noted that unless specified otherwise, a regional mean refers to an average over 6 longer-term stations, excluding Bellingshausen, Davis and Syowa (see Table 6-1). The regional mean trends were found to be significantly different from zero at least at the 5% significance level at all except the 300 hPa level (which is approximately the warming-cooling transition layer) and the 30 hPa level (where trend estimates are more uncertain due to the quantity of missing values at this pressure level).
After further averaging over all levels from 850 to 300 hPa, it was found that the rate of warming in the UT is about 0.13°C per decade. This is in agreement with a global warming trend of about 0.14ºC per decade in the troposphere (from 850 hPa to 300 hPa) since 1958 (Karl et al., 2006). Excluding the warming-cooling transition layer (300 hPa), the mean warming rate (averaged over the levels from 850 to 500 hPa) is 0.17°C per decade. By averaging over all levels from 200 to 30 hPa, the rate of cooling in the Antarctic LS is about -0.47°C per decade. The Antarctic 100-50 hPa mean cooling rate is estimated to be -0.58°C per decade, while a global rate of lower stratospheric (from 100 hPa to 50 hPa) cooling since 1958 was estimated to be about -0.37ºC per decade (Karl et al., 2006). The strongest stratospheric cooling over the Antarctic was observed at the100 hPa pressure level, also different levels of the troposphere (from 850 hPa to 500 hPa) were warming at similar rates of about 0.14°C to 0.19° C per decade (Figure 6-6). In terms of annual trends, the stratospheric cooling occurred at faster rates than those for the tropospheric warming. If averaged over all the 9 stations analysed, the mean profile looks similar to that shown in Figure 6-6; the mean cooling rate in the LS (200 to 30 hPa) is about -0.60°C per decade, and the mean warming rate in the troposphere (from 850 to 500 hPa) is about 0.18°C per decade (both mean rates are slightly higher than the corresponding 6-station means).
94
Figure 6-6 The mean vertical profile of the annual temperature trends (°C per decade) and their standard deviations (STD) at nine atmospheric pressure levels. These are derived from the HadAT2 (homogenised) radiosonde data from the six longer-term Antarctic stations
Seasonal differences in the long-term trends
Seasonal differences in the long-term temperature trends were examined. The four seasons are defined as following: winter [JJA], spring [SON], summer [DJF] and autumn [MAM] (Ogawa et al., 2009). Trends from each seasonal mean temperature series were derived from the corresponding homogenized monthly mean series. This seasonal trend estimate was undertaken for each season, for each station and at each pressure level. The regional mean vertical profiles of seasonal trends are presented in Figure 6-7.
In the LS, statistically significant cooling trends were detected at all four pressure levels from 200 to 50 hPa in spring and summer (Figure 6-7). The fastest cooling rate (about -1.78°C per decade) was detected at 100 hPa for spring. The mean spring-summer cooling rate (averaged over the four pressure levels and over spring and summer) is -1.06˚C per decade. Seasonal cooling trends at 30 hPa are statistically significant only in summer. Note that the seasonal trend estimates at 30 and 50 hPa are
95 less reliable than those for the other pressure levels because the data time series have many missing values, especially in winter.
Figure 6-7 The mean vertical profile of seasonal temperature trends (°C per decade) and their standard deviations (STD) at nine atmospheric pressure levels derived from the HadAT2 (homogenised) radiosonde data from the six longer-term Antarctic stations
This is an apparent seasonality trend in the LS temperature. This and the observed maximum cooling at altitudes around 100 hPa, can be linked to the dramatic seasonal reduction in the stratospheric ozone concentration over the austral spring since the 1970s, known as “the Antarctic ozone hole” (Farman et al., 1985). As the stratospheric temperature depends on ozone absorption of ultraviolet radiation, rapid ozone destruction, which occurs in spring when sunlight returns to the polar region (Solomon et al., 1986) and lasts till early summer, results in this significant seasonal stratospheric cooling. The ozone depletion is largest in the LS around 100 hPa (about 95% of the ozone is destroyed between altitudes 15 and 20 km) but in the upper stratosphere (above 30 hPa) the decreases are small.
96 Tropospheric warming at the 850, 700 and 500 hPa pressure levels were detected in all four seasons but with slightly higher warming rates in winter (Figure 6-7). The regional mean tropospheric warming trends in Figure 6-7 are statistically significant (at least a 5% significance level) at all 3 pressure levels (i.e. 850, 700, and 500 hPa). The mean warming rate (averaged over the three pressure levels and over all four seasons) is 0.19°C per decade. Similar vertical profiles of seasonal trends are seen when the regional mean is obtained by averaging over all the 9 stations analysed, instead of just the 6 longer term stations. The fastest cooling rate (about -1.90°C per decade) was again detected at 100 hPa for spring. Note that the inclusion of the 3 shorter term stations increases the uncertainty in the regional mean trends, especially in the lower to mid-troposphere. The 9-station mean tropospheric warming trends are significant (at 5% level) only in winter. This uncertainty comes mainly from the inclusion of the shortest series (at station Bellingshausen), which has only 30 years of data which ended in January 1999.
Discussions
The results generally agree with earlier findings which identified warming in the UT and cooling in the LS over the Antarctic. In particular, the mean vertical profile of winter temperature trends over the past five decades (Figure 6-7) is similar to that estimated for the 1971-2003 period (Turner et al., 2006). However, our estimates of the winter tropospheric warming trends (0.20~0.31°C per decade for 9- or 6-station means) are smaller, and more uniform across the troposphere, than those reported in (Turner et al., 2006). This variation could arise because (i) data homogenisation was performed in this study but not in (Turner et al., 2006); (ii) in the period of analysis was different (e.g. shorter records are usually less representative of long-term trends), and (iii) the trend analysis methods were different particularly when the effects of autocorrelation and the incomplete sampling of the annual cycles were considered. The estimates of the LS cooling rates are faster for the 1971-2003 period (Turner et al., 2006), than for the whole period of the records analysed in this study. This is also apparent for the winter tropospheric warming rates at the 500 and 300 hPa levels. The latter indicates that the larger winter warming rates estimated in (Turner et al., 2006) than in this study are partly due to the difference in the period of the analysis. The raw data show lower cooling rates at the 30 hPa level than do the homogenized data. In addition, a few
97 erroneously low temperatures in the early period and an erroneously high temperature in the later period were recorded at the McMurdo station. This would result in a positive bias in a least squares estimate of the linear trend unless they were discarded. The positive bias in the trend estimate most likely exists in the annual and the spring estimates presented in (Turner et al., 2006), because these errors were not reported in that study. Autocorrelation in the data series analysed in this study is highly positive, and if ignored would result in a positive bias in the estimate with statistical significance of a warming trend (Storch, 1999).
The results shown in Figure 6-7 also demonstrated that, in each season, the tropospheric warming was somewhat uniform across the pressure levels from 850 to 500 hPa; the range of the warming rate (per decade) was about 0.20~0.31°C in winter, 0.16~0.22°C in spring, 0.13~0.16°C in summer, and 0.17~0.19°C in autumn respectively. Examining seasonal trends in the LS, it appears that the cooling occurred at much higher rates than it was previously reported: the largest decrease in temperature was observed in spring and summer at 100 hPa with cooling rates of about -1.8°C and -1.5°C per decade, respectively (Figure 6-7; the corresponding 9-station means are -1.9°C and -1.7°C per decade) respectively. No significant LS cooling trend was identified in winter and autumn, with the exception of winter at 30 hPa. However, this trend estimate was not reliable because of the quality of missing values in the data series.
In this study, long-term atmospheric temperature trends were derived from radiosonde data obtained over nine stations located in Antarctica. Over the past five decades significant lower-stratospheric cooling and tropospheric warming occurred over the Antarctic. While are the causes of the observed significant temperature trends beyond the scope of this study, it has been noted that the observed pattern of the tropospheric warming and stratospheric cooling is one of the expected consequences of the increase in concentration of greenhouse gases and stratospheric ozone depletion (Karl et al., 2006; Storch, 1999). Note that the decline of solar ultraviolet from 2001 to 2008 may have contribution to the climate change as well. Also, it has been demonstrated that human activities have caused significant warming in the Antarctic (Gillett et al., 2008).
The results also demonstrated clear seasonality of the observed stratospheric temperature trends. In spring and summer, cooling trends in the LS were highly significant while autumn and winter temperature trends are rather weak and statistically
98 insignificant. This seasonality of trends appears to be linked to the dramatic seasonal reduction in the stratospheric ozone concentration over the austral spring since the 1970s.
6.2.6 Temperature trends using CHAMP RO data
GNSS RO data from the CHAMP mission processed by UCAR (Level 2 wetPrf data available from May 2001 to August 2008; http://www.cosmic.ucar.edu/) were analysed over the Antarctic region (area south of the latitude 60°S). Satellite-derived temperature trends were estimated for ten pressure levels (900, 850, 700, 600, 500, 300, 200, 150, 100 and 50 hPa). Maps of the UT and the LS temperature trends were derived from the GNSS RO data by calculating the trends for each of the 45 boxes of the 800 km Equal Area Scalable Earth Grid (Brodzik and Knowles, 2002) (Figure 6-8). Radiosonde records from the HadAT2 dataset at the Hadley Centre, UK Met Office (http://hadobs.metoffice.com/hadat/) were also used to analyse the UT and the LS temperature trends. Six coastal Antarctic stations (Casey, Davis, Halley, Mawson, McMurdo and Novolazarevskaya) were selected for the analysis. They were chosen because of the completeness of their data records (on average, less than 2-3% of missing data at ten standard pressure levels from 900 to 50 hPa) for the same time period as the satellite data. Radiosonde data from the Amundsen-Scott station were also utilised as these data were representative of the interior of Antarctica. However, the percentages of missing data were relatively high for this station i.e. 21-24% for the six standard pressure levels from 500 to 50 hPa.
99
Figure 6-8 Location of seven Antarctic stations with the 800 km Equal Area Scalable Earth Grids superimposed over a map of the continent (45 grids in total) and 500 km radius circles centred on each station
Comparison of temperature trends derived from radiosonde and GPS RO observations
Atmospheric temperature trends derived from the radiosonde data at seven Antarctic stations and from the collocated CHAMP RO data were examined for the period 2001-2008 (Kuleshov et al., 2011b). For this comparison, CHAMP data were selected and co-location criterion of not more than 500 km distance between the radiosonde and GPS RO profiles was implemented (see Figure 6-8). Radiosonde and satellite data for seven complete years, from 1st September 2001 (the beginning of the austral spring) to 31st August 2008 (the end of the austral winter) were selected for the analysis to ensure that an equal number of seasons as presented in the time series. CHAMP RO and radiosonde data have demonstrated good agreements in the UT and the LS (for the pressure levels between 500 and 50 hPa), with respect to both the actual values and the seasonal cycle (see the comparison of in-situ radiosonde data and the corresponding CHAMP data for seven stations at 100 and 500 hPa in Figure 6-9 and Figure 6-10, respectively). Cooling trends through the UT and the LS were identified in both radiosonde and CHAMP data (Figure 6-11). Strong stratospheric cooling trends estimated as about -2° to -3°C per decade at the 200 and 100 hPa pressure levels (mean values for the seven localities) are statistically significant. Some differences in the trend
100 estimates can be explained because of the different methods of data sampling. Radiosonde measurements are “point” observations performed at the stations while GPS RO technique provides area-averaged data (in this comparison, over an area of about 785,000 km 2). Note that the high accuracy of the GPS RO data allows them to be used for the assessment of the quality of radiosonde temperature measurements in the UT and the LS (He et al., 2009).
Figure 6-9 Comparisons of i n-situ 100 hPa radiosonde data (black) and the corresponding CHAMP data (blue) at the stations indicated
101
Figure 6-10 Comparison of i n-situ 500 hPa radiosonde data (black) and the corresponding CHAMP data (blue) at the indicated stations
102
Figure 6-11 Comparison of linear trends estimated from in-situ radiosonde (RS) data and the corresponding CHAMP data for the period from September 2001 to August 2008
Temperature trend estimates: the upper troposphere
In order to quantitatively evaluate temperature trends derived from satellite remote sensing data, an area-averaged temperature trend over the Antarctic at a particular pressure level was computed as
∑(T iSi)/ ∑Si for i=1, N,
where T i is a temperature trend computed for each box i with corresponding area S i, and N is the total number of boxes (N=45; 800 km Equal Area Scalable Earth Grid was implemented – see Figure 6-8 for details). Maps created of the estimated atmospheric temperature trends from the CHAMP data for the period from September 2001 to August 2008 revealed significant variations in the regional cooling and warming at the examined pressure levels (Figure 6-11).
A map of the average annual 500-hPa temperature trend (Figure 6-12: 500 hPa) demonstrates that the upper tropospheric cooling was dominant over a large region of Antarctica between about 60°E and 135°W, in a sector where Amundsen-Scott, Casey,
103 Davis, Mawson and McMurdo stations are located. The average annual cooling rates over this region were generally below -1°C per decade; however, cooling trends in excess of -2°C per decade have been also identified over certain areas (e.g. coastal area and adjusted oceans around 180°, Figure 6-12: 500 hPa). Temperature trends over the other half of the Antarctic are mainly warming. There is a strong and statistically significant (at the 95% confidence level) warming trend near the Antarctic Peninsula (around 75°W) which exceeds 2°C per decade (Figure 6-12: 500 hPa). Overall, general tropospheric cooling over the entire Antarctic during 2001-2008 with an area-averaged temperature trend of about –0.14°C per decade (at 500 hPa) was estimated from the GPS RO data.
Temperature trend estimates: the lower stratosphere
The map of the average annual 100-hPa temperature trend over the Antarctic for 2001-2008 demonstrates dominant regional cooling (Figure 6-13: 100 hPa). Faster cooling rates which exceed -2°C per decade were observed over nearly two thirds of Eastern Antarctica and about one third of Western Antarctica (i.e. a sector between about 110°E and 60°W). Over the rest of the Antarctic, the LS was also mainly cooling over most of the polar region but at slower rates. From the satellite data, an area-averaged temperature trend over the Antarctic was about -1.7°C per decade at the 100 hPa pressure level. Note that all seven Antarctic stations are located in parts of the continent where strong stratospheric cooling is observed. Satellite data also demonstrated warming over two areas of Western Antarctica between 100°W and 170°W. However, there were no meteorological observational stations in that area for comparison.
The GPS RO-derived results confirmed that the cooling identified by Turner et al. 2005 in the LS over the Antarctic for earlier decades was also observed in recent years. Results show that it occurred at a higher rate than what was previously reported. The long-term radiosonde-derived trend estimates indicated cooling at a rate of -0.16°C per decade at the 100 hPa pressure level for 1971 – 2001 (Turner et al., 2005). GPS RO data for 2001-2008 demonstrated that (i) the mean temperature trend estimated for seven Antarctic stations was about -3.5°C per decade (in agreement with -3.51°C per decade average trend as estimated from the radiosonde data) and (ii) the area-average stratospheric temperature trend over the Antarctic was about -1.7°C per decade.
104
Figure 6-12 Temperature trends estimated from CHAMP RO data for the period from September 2001 to August 2008 at pressure levels 400, 500, 600 and 750 hPa
105
Figure 6-13 Temperature trends estimated from CHAMP RO data for the period from September 2001 to August 2008 at pressure levels 50, 100, 200 and 300 hPa
Examining the seasonal variations in temperature trends for 2001-2008, the strongest stratospheric cooling occurred in the austral spring. This corresponds with the recent findings about the statistically significant low stratospheric cooling over the Antarctic observed in the past five decades as derived from radiosonde data (Kuleshov et al.,
106 2011b). Using GPS RO data, warming in the UT and cooling in the LS reported in the past decade have been confirmed. The findings are in a good agreement with results obtained using long-term radiosonde data (see section 6.2.5). The Antarctic seasonal ozone depletion which has a maximum around November could be one of the important contributing factors in the strong stratospheric cooling reported (Thompson and Solomon, 2002).
Discussions
Using seven years CHAMP GPS RO data, the UT and the LS temperature trends and demonstrated significant regional variation in the spatial distribution of the temperature trends over the Antarctic for 2001-2008 were examined. The results demonstrated a general cooling of the atmosphere over the Antarctic in 2001-2008. This was at a rate of -0.14°C per decade in the UT and -1.7°C per decade in the LS. Temperature trends derived from CHAMP RO data were in agreement with the results of the earlier studies. The CHAMP RO data provided independent high-accuracy estimations of atmospheric temperature trends consistent with radiosonde observations. The study has demonstrated that the CHAMP RO observations can serve as a climate benchmark for examining changes in atmospheric temperature.
6.3 Tropopause study using FORMOSAT-3/COSMIC RO data
6.3.1 The tropopause
The tropopause is a layer of atmosphere which separates the troposphere and the stratosphere and is important for many weather and climatological studies. Examining the characteristics of the tropopause requires detailed vertical atmospheric profile information and this can be a challenging task in some situations (Randel et al., 2007). The Australian Bureau of Meteorology maintains a network of 38 radiosonde stations which provides the major upper-air information for the vast continent, surrounding oceans and also the Antarctic (Fu et al., 2009a). Most of the stations are located in the coastal regions and there are only a few inland. The vertical resolution of the radiosonde data is typically from a few hundreds meters to about a kilometre. This imposes certain limitations for tropopause studies. On the other hand, the GNSS RO
107 technique generates atmospheric profiles with a high vertical resolution (from tens of meters to a few hundred of meters depending on the layers of the atmosphere) and provides a more homogenous spatial distribution of the data. A study of the tropopause in the Australasian region using FORMOSAT-3/COSMIC data products (from 2007 to 2009) was carried out and the results showed the potential for detailed tropopause studies e.g. examining the multiple structure of the tropopause. The GNSS RO technique can retrieve high vertical resolution atmospheric profiles with a uniform global coverage. Therefore, enhanced tropopause information (i.e. better resolutions and distribution) can be expected from the GNSS RO technique (Lewis, 2009; Schmidt et al., 2005b).
6.3.2 Tropopause structure determination using RO data
A study was designed to retrieve tropopause heights over the Australasian region using GPS FORMOSAT-3/COSMIC RO temperature profiles (Fu et al., 2011). Currently, the FORMOSAT-3/COSMIC RO data provide high vertical resolution of upper air temperature profiles. The results confirmed that the FORMOSAT-3/COSMIC RO temperature profiles showed more details of the atmosphere and provided better characteristics of the tropopause than the conventional meteorological observations using radiosonde.
In Figure 6-14, two pairs of collocated temperature profiles obtained from radiosonde and FORMOSAT-3/COSMIC RO are presented. Radiosonde shows discontinuities in its temperature profiles. However, the FORMOSAT-3/COSMIC RO data show smooth profiles due to its high vertical resolution. The September (2009) FORMOSAT-3/COSMIC RO temperature profile (upper-right panel in Figure 6-14) indicates a double tropopause structures at 9 km and 17 km. The radiosonde profile does not clearly identify the double tropopause due to the lower spatial resolution of measurements. Average tropopause heights over the Australasian region have been calculated using FORMOSAT-3/COSMIC RO data from July 2006 to the end of 2010. The results (Figure 6-15) are grouped by four latitude zones and four time periods (day periods of years: 1-30; 90-120; 180-210 and 270-300). In general, the heights of tropopause are higher in the lower latitudes.
108 Radiosonde temperature profile COSMIC RO temperature profile (23:00, 08/09/2009, Lat/Lon: -28.7/114.7) (00:08, 09/09/2009, Lat/Lon: -27.8/114.4) 30 30
10 10 RS_temperature RO_temperature -10 -10
-30 -30
-50 -50 Temperature(C) Temperature(C) -70 -70 -90 -90 .7 .5 0.7 3.5 6.3 9.1 0.3 3.1 5.9 0.0 1.5 3.1 5.2 8.5 10.6 14.1 17.8 20.6 23.7 27.6 11.9 14 17 2 2 2 28.7 31.5 34.3 37.1 39.9 Altitude (km) Altitude (km)
Radiosonde temperature profile COSMIC RO temperature profile (23:00, 01/08/2009, Lat/Lon: -17.9/122.2) (01:03, 31/07/2009, Lat/Lon: -16.9/123.3) 30 30
10 10 RS_temperature RO_temperature -10 -10
-30 -30
-50 -50 Temperature(C) Temperature(C) -70 -70
-90 -90 0.0 1.4 3.6 6.3 9.1 12.8 17.0 19.4 21.4 24.0 27.9 0.6 3.4 6.2 9 11.8 14.6 17.4 20.2 23 25.8 28.6 31.4 34.2 37 39.8 Altitude (km) Altitude (km) Figure 6-14 Sample of temperature profiles obtained from radiosonde and FORMOSAT-3/COSMIC RO respectively
Average tropopause heights at different latitude zones and time period derived from COSMIC RO data (2006-2010) 18.0
16.0 Day period 14.0 of year 12.0 1-30 10.0 90-120 8.0 180-210 6.0 270-300 4.0 Tropopause height (km) height Tropopause
2.0
0.0 0 — -20 -20 — -40 -40 — -60 -60 — -90 Latitude Zone (degree)
Figure 6-15 Average tropopause heights at different latitude zones and time periods (day: 1-30; 90-120, 180-210 and 270-300) derived from FORMOSAT-3/COSMIC RO
109 Figure 6-16 , Figure 6-17 and Figure 6-18 show the map of annual average tropopause heights calculated from FORMOSAT-3/COSMIC RO data in 2009, 2008 and 2007, respectively.
Figure 6-16 Annual average tropopause heights derived from FORMOSAT-3/COSMIC RO data in the Australasian region - 2009
Figure 6-17 Annual average tropopause heights derived from FORMOSAT-3/COSMIC RO data in the Australasian region - 2008
110
Figure 6-18 Annual average tropopause height derived from FORMOSAT-3/COSMIC RO data in the Australasian region - 2007
6.4 Summary
This Chapter presented the results of the Antarctic temperature trend analysis studies and Australian tropopause structure studies using GNSS RO retrievals from two key recent missions CHAMP and FORMOSAT-3/COSMIC. The research demonstrated that the GNSS RO technique can provide stable and quality atmospheric information for climatological studies. The unique global coverage and high (vertical, horizontal and temporal) resolution are the advantages of the technique which overcomes the critical limitations of traditional station-based and satellite remote sensing atmospheric observations. The Antarctica temperature trend study using RO atmospheric profiles demonstrated that the technique has capability to monitor the Antarctica atmosphere with higher (temporal and spatial) resolutions and better sampling distribution. The space-based technique fills the gaps of the radiosonde network for important upper air observation. The tropopause study over the Australia region has proven that the high quality and vertical resolution atmospheric observation from RO technique benefit the in-depth tropopause structure studies. These studies confirm the value of the GNSS RO data and also demonstrate the potential for advancing the current studies.
111 (This page intentionally left blank)
112 Chapter 7 Conclusions and Recommendations
7.1 Summary
Monitoring the Earth’s atmosphere effectively and efficiently is one of the most important and most difficult tasks in meteorology. The spatial and temporal heterogeneity of the atmosphere contains valuable information. The rapid availability of atmospheric information in a high resolution and a global coverage is highly desirable. The GNSS RO technique has been proven to be an important alternative to the current atmospheric observation methods and can provide high quality and high resolution global atmospheric observations. It has the potential to revolutionise meteorological research and practice in the future.
This thesis presents the first comprehensive PhD study conducted in Australia on the GNSS RO atmospheric observation technique. The study is intended to address important questions and practical issues that are confronted by Australia to utilise the RO atmospheric products for better meteorological services. A series of pioneering studies were conducted and the outcomes provided valuable contributions to the BoM to assist its adoption of the state of the art technology in meteorological studies and applications.
The key contributions of this research are summarised as below:
• Established a solid database related to GNSS RO data products and publications;
• Conducted various evaluation studies of the GNSS RO retrievals focusing on the Australian and Antarctic regions;
• Investigated the Antarctic temperature trend using the CHAMP RO atmospheric data;
• Investigated a simulation study of the impact of future GNSS systems to the GNSS RO technique and Australian meteorology;
113 • Developed a framework of GNSS RO data processing system in a distributed computation environment;
• Conducted a comprehensive review of the GNSS RO technique including its fundamentals, history, missions, applications and the state of art technology;
• Investigated the structure of Australian tropopause using COSMIC RO atmospheric data.
7.2 Conclusions
The potential of the GNSS RO technique for Australian meteorology
This thesis has proven that the GNSS RO technique can provide benchmark atmospheric data for future atmospheric studies. The GNSS RO atmospheric data with a high resolution and uniform coverage are highly valuable for Australia which has large unpopulated areas and is located in the ocean-dominant Southern Hemisphere.
The review work, evaluation studies and investigations in climate have demonstrated the value of the GNSS RO retrieved atmospheric information to the Australian BoM and provided a solid basis for future studies. Australian meteorology can benefit significantly from the space-based technique.
An encouraging future with multiple GNSS
The simulation study shows that the current and next generation GNSS missions will provide thousands of daily RO events for the Australasian region. The quality of GNSS RO data will also be improved due to the increased number and strength of GNSS signals available for atmospheric retrieval processes.
The study of LEO satellite orbits and their impact on the number and distribution of RO events is important for establishing potential future Australian GNSS RO missions. There is a critical need of Australian involvement from the international GNSS RO community because of Australia’s unique geographical location for both new satellite missions and meteorological studies. The research results strongly support Australia’s
114 ongoing involvement, investigation and contribution to the GNSS RO atmospheric observations.
GNSS RO data processing infrastructure
It is important to deliver high quality GNSS RO retrievals with minimal delays. The retrieval process involves a large volume of data, complex algorithms and it relies on an advanced computational infrastructure. For Australia to benefit most from the technique and prepare for its own missions, a dedicated data processing centre needs to be established.
The developed prototype data processing system is designed to address its special computational requirements and then ensure quality and timely retrieval processes. The prototype system can be “up scaled” to an operational GNSS RO data processing system.
Evaluations of GNSS RO retrievals
It is critical to assess the GPS RO technique’s reliability and the quality of the atmospheric profiles derived. This then forms an important guideline for the assimilation of the new data sources into the current NWP systems and other meteorological applications. The four evaluation studies were designed with researchers from the Australian BoM and addressed their specific interests in accessing the GNSS RO retrievals in the Australian context. These studies present good results and agree well with other independent international evaluation studies.
Co-location criteria to match the RO profiles with other observations or modelled profiles are important for understanding the evaluation results and the data in the evaluation studies. The investigation into the impact of co-location criteria supports the evaluation studies and provides a guideline for future evaluation studies.
Applications of the GNSS RO retrievals in the Antarctic climate and the Australian tropopause studies
The Antarctic temperature trend studies, using both GNSS RO and their collocated Antarctic radiosonde data for 2001-2008, were found to be in good agreement in terms of monthly mean atmospheric temperature values and trends therein. This was
115 especially the case in the lower stratosphere to the upper troposphere. The study demonstrated overall cooling in the Antarctic lower stratosphere and warming in the troposphere, in agreement with the radiosonde measurements. This depicted the suitability of the GNSS RO data for climate change detection for the lower stratosphere and the upper troposphere.
The pilot study on the tropopause for the Australasian region using GNSS RO data confirmed the advantages of the GNSS RO methodology for detailed atmospheric structure studies.
Both studies demonstrated that the GNSS RO technique has the capability to provide high quality and continuous atmospheric observation in remote areas (e.g. Australian unpopulated area, the Antarctica and the oceans). The study suggested that the GNSS RO atmospheric retrievals can serve as a better benchmark for meteorological studies when compared with the traditional atmospheric observations.
7.3 Suggestions and recommendations
Further studies need to be carried out in this area to improve the technique and the usability of the retrievals. Some suggestions and recommendations are given below:
• Error quantifications and quality control . It is important to properly assess and quantify the errors at the various stages of the retrieval process. The assessment provides valuable information for assimilation of the GNSS RO data and also other meteorological applications.
• Satellite POD . The orbit information for both the LEO and the GNSS satellites is the one of the major inputs for retrieving accurate atmospheric profiles. High quality POD information is essential for real time applications of the GNSS RO retrievals. A dedicated study in POD specifically targeted at the GNSS RO retrievals should be carried out.
• Evaluation study . Continuous assessment of the RO retrievals is necessary using different sources of data with respect to different timeframes and
116 geographical locations. It is important to continue the evaluation studies over the data from new missions and from different data centres.
• GNSS RO data processing centre . Currently, many GNSS RO atmospheric products are freely available from several international data centres. However, it is clear that Australia needs its own data processing centre to be able process and utilise the data. This is critical for real-time applications and extensive assessment of data quality. The international community also needs Australia’s involvement in future RO satellite missions. It is important to continually study the data processing system to handle the ever increasing and complex satellite and meteorological data.
• LEO-LEO RO . The RO atmospheric observation technique will not be limited by just the GNSS signals, which are designed for positioning, navigation and timing services. A LEO-LEO technique has been proposed to capture more information from the atmosphere using selected microwave and infrared-laser frequencies (Yue et al., 2010). With the additional signals, the RO technique can not only retrieve thermo-dynamics (i.e. temperature, pressure and humidity)
but also greenhouse gases (e.g. CO 2, CH 4, N 2O and O 3) and wind fields. LEO-LEO RO can work well with the GNSS RO technique in the future RO missions and deliver a unique solution for multiple purpose atmospheric observations.
• Other atmospheric observations products . Studies need to be carried out to utilise the GNSS RO atmospheric retrievals with data from other observations such as satellite remote sensing observations. Different characteristics of these observation methods need to be taken into account and studied carefully for different purposes.
In summary, this thesis presented a series of studies in the GNSS RO atmospheric observation technique for Australian meteorology. The GNSS RO technique is an important revolution in atmospheric monitoring and adopting the technique is a major milestone. Since the first GNSS RO profile was retrieved on 16 April 1995, the development and acceptance of the technique has been encouraging and brought important benefits to the meteorological community. The technique will continue to be
117 developed to provide robust solutions for the Earth’s atmospheric monitoring. Australia should continue its investigations into GNSS RO for better meteorological services and also for better response to the climate change.
118
References
Alexander, P., de la Torre, A., and Llamedo, P.: Interpretation of gravity wave signatures in GPS radio occultations, Journal of Geophysical Research, 113, doi:10.1029/2007jd009390, 2008.
Alexander, S. P., Klekociuk, A. R., and Tsuda, T.: Gravity wave and orographic wave activity observed around the Antarctic and Arctic stratospheric vortices by the COSMIC GPS-RO satellite constellation, Journal of Geophysical Research, 114, doi:10.1029/2009jd011851, 2009.
Anthes, R. A., Rocken, C., and Kuo, Y. H.: Applications of COSMIC to Meteorology and Climate, Terrestrial, Atmospheric and Oceanic Sciences, Special issue on COSMIC, 39, 2000.
Anthes, R. A., Bernhardt, P. A., Chen, Y., Cucurull, L., Dymond, K. F., Ector, D., Healy, S. B., Ho, S. P., Hunt, D. C., Kuo, Y. H., Liu, H., Manning, K., McCleese, D. J., Meehan, T. K., Randel, W. J., Rocken, C., Schrener, W. S., Sokolovskiy, S. V., Thompson, D. C., Ternberth, K. E., Wee, T. K., Yen, N. L., and Zeng, Z.: The COSMIC/FORMOST-3 mission: Early results, Bulletin of the American Meteorological Society, 89, 313-333, doi:10.1175/BAMS-89-3-313, 2008.
Aragon-Angel, A., Hernandez-Pajares, M., Zornoza, J. M. J., and Subirana, J. S.: Improving the Abel transform inversion using bending angles from FORMOSAT-3/COSMIC, GPS Solutions, 14, 11, 2010.
Australian Bureau of Meteorology: Australia's global climate observing system, Australia Bureau of Meteorology, Melbourne, 47, 2002.
Australian Bureau of Meteorology: Annual Climate Summary, 2010a.
Australian Bureau of Meteorology: Climate snapshot: State of the Climate published, Australia Bureau of Meteorology and CSIRO, 6, 2010b.
Bargagli, R.: Antarctic ecosystems : environmental contamination, climate change, and human impact, Ecological Studies, Springer, Berlin ; London, 2007.
Beeton, R. J. S., Buckley, K. I., Jones, G. J., Morgan, D., Reichelt, R. E., and Trewin, D.: Australia State of the Environment 2006, Department of the Environment and Heritage, Canberra, 141, 2006.
Bevis, M., Businger, S., Herring, T. A., Rocken, C., Anthes, R. A., and Ware, R. H.: GPS Meteorology: Remote Sensing of Atmospheric Water Vapor Using the Global Positioning System, Journal of Geophysical Research, 97, 15787-15801, doi:10.1029/92jd01517, 1992.
119 Beyerle, G., Wickert, J., Galas, R., Hocke, K., Konig, R., Marquardt, C., Pavelyev, A. G., Reigber, C., and Schmidt, T.: GPS Occultation Measurements with GPS/MET and CHAMP, Taikiken Shinpojiumu, 15, 44-47, 2001.
Beyerle, G., Schmidt, T., Michalak, G., Heise, S., Wickert, J., and Reigber, C.: GPS radio occultation with GRACE: Atmospheric profiling utilizing the zero difference technique, Geophysical Research Letters, 32, 17, 2005.
Beyerle, G., Heise, S., Kaschenz, J., Konig-Langlo, G., Reigber, C., Schmidt, T., and Wickert, J.: Refractivity Biases in GNSS Occultation Data, in: Atmosphere and Climate: Studies by Occultation Methods, edited by: Foelsche, U., Kirchengast, G., and Steiner, A. K., Springer, 37-43, 2006.
Biadeglgne, B., Zhang, K., Hodet, G. d., Kuleshov, Y., and Wu, F.: Verification of Atmospheric Temperature and Moisture Profiles Derived by GPS Radio Occultation Technique, Proceedings of Spatial Science Coalition, Tasmania, Australia, may 14-18 2007.
Bizzarri, B., Bordi, I., Dell'Aquila, A., Petitta, M., Schmidt, T., Sutera, A., and Wickert, J.: Analysis of seasonal and daily mid-latitude tropopause pressure using GPS radio occultation data and NCEP-NCAR reanalyses, in: Atmosphere and Climate: Studies by Occultation Methods, edited by: Foelsche, U., Kirchengast, G., and Steiner, A. K., Springer, 253-263, 2006.
Boccara, G., Hertzog, A., Basdevant, C., and Vial, F.: Accuracy of NCEP/NCAR reanalyses and ECMWF analyses in the lower stratosphere over Antarctica in 2005, Journal of Geophysical Research, 113, doi:10.1029/2008jd010116, 2008.
Braun, J. J., and Hove, T. v.: Recent improvements in the retrieval of precipitable water vapor, ION GNSS, Long Beach, CA, Sept 2005.
Brodzik, M. J., and Knowles, K. W.: EASE-Grid: A versatile set of equal-area projections and grids, in: Discrete Global Grids, edited by: Goodchild, M., National Center for Geographic Information & Analysis, California USA, 2002.
Businger, S., Chiswell, S. R., Bevis, M. G., Duan, J., Anthes, R. A., Rocken, C., Ware, R. H., Exner, M. L., Vanhove, T., and Solheim, F. S.: The promise of GPS in atmopheric monitoring, American Meteorological Society, 77, 5-18, 1996.
Chao, B. F., Pavlis, E. C., Hwang, C., Liu, C. C., Shum, C. K., Tseng, C. L., and Yang, M.: COSMIC: Geodetic applications in improving Earth's gravity model, Terrestrial, Atmospheric and Oceanic Sciences, 11, 365-378, 2000.
Chevallier, F., and Engelen, R. J.: Assimilation of satellite data for the environment, ECMWF Seminar on recent development in the use of satellite observations in NWP, 3-7 Sept 2007.
Cho, S., Chung, J., Park, J., Yoon, J., Chun, Y., and Lee, S.: Radio occultation mission in Korea multi-purpose satellite KOMPSAT-5, New Horizons in Occultation Research Studies in Atmosphere and Climate, edited by: Steiner, A., Pirscher, B., Foelsche, U., and Kirchengast, G., Springer, 315 pp., 2009.
120 Chuvieco, E. S.: Earth observation of global change : the role of satellite remote sensing in monitoring the global enironment, Springer, [New York, NY], xii, 223 pp., 2008.
Cohen, S. J., and Waddell, M. W.: Climate change in the 21st century, McGill-Queen's University Press, Montréal ; Ithaca [N.Y.], xxvii, 379 pp., 2009.
Cornman, L. B., A.Weekley, Goodrich, R. K., and Frehlich, R.: Using airborne GNSS receivers to detect atmospheric turbulence, New Horizons in Occultation Research Studies in Atmosphere and Climate, edited by: Steiner, A., Pirscher, B., Foelsche, U., and Kirchengast, G., Springer, 315 pp., 2009.
COSMIC Data Analysis and Archive Center: Algorithms for processing GPS data into absolute total electron content (aTEC) Along LEO-GPS links, 2008.
Cucurull, L., Derber, J., Treadon, R., and Purser, R. J.: Preliminary impact studies using Global Positioning System radio occultation profiles at NCEP, Monthly Weather Review, 136, 2008.
Drusch, M., Scipal, K., Andersson, E., Balsamo, G., Holmes, T., and Rosnay, P. d.: Use of satellite data for land surface assimilation, ECMWF Seminar on recent development in the use of satellite observations in NWP, 3-7 Sept 2007.
Engeln, A. v., and Nedoluha, G.: Retrieval of temperature, water vapor, and a reference pressure from radio occultation refractivity and bending angle measurements using a 1D Var approach: A simulation study, Journal of Geophysical Research, 108, 26, 2003.
European Environment Agency: The European environment – state and outlook 2010: synthesis, 2010.
Eyre, J. R.: Progress achieved on assimilation of satellite data in numerical weather prediction over the last 30 years, ECMWF Seminar on recent development in the use of satellite observations in NWP, 3-7 Sept 2007.
Farman, J. C., Gardiner, B. G., and Shanklin, J. D.: Large losses of total ozone in Antarctica reveal seasonal ClOx/NOx interaction, Nature, 315, 207-210, 1985.
Feng, D. D., and Herman, B. M.: Remotely sensing the Earth's atmosphere using the Global Positioning System (GPS) - The GPS/MET data analysis, Journal of Atmospheric and Oceanic Technology, 16, 14, 1999.
Fjeldbo, G., Kliore, A. J., and Eshleman, V. r.: The neutral atmosphere of Venus as studied with the Mariner V radio occultation experiments, The Astronomical Journal, 76, 1971.
Foelsche, U., and Kirchengast, G.: Sensitivity of GNSS occultation profiles to horizontal variability in the tropsophere: A simulation study, Physics and Chemistry of the Earth: Part A, 29, 225-240, 2004.
Foelsche, U., Borsche, M., Steiner, A. K., Pirscher, B., Lackner, B. C., Gobiet, A., and Kirchengast, G.: CHAMP radio occultation based climatologies for global monitoring
121 of climate change, Wegener Center for Climate and Global Change, University of Graz, 58, 2006.
Fong, C., Yang, S., Chu, C., Huang, C., Yeh, J., Lin, C., Kuo, T., Liu, T., Yen, N. L., Chen, S., Kuo, Y., Liou, Y., and Chi, S.: FORMOSAT-3/COSMIC constellation spacecraft system performance: After one year in orbit, IEEE Trans. on Geoscience and Remote Sensing, 46, 2008.
Fong, C., Yen, N. L., Chu, C., Hsiao, C., Yang, S., Lin, Y., Chen, S., Liou, Y., and Chi, S.: In quest of global radio occultation mission for meteorology beyond 2011, Proceedings of Aerospace Conference, IEEE, 1-8 2009a.
Fong, C., Yen, N. L., Chu, C., Yang, S., Shiau, W., Huang, C., Chi, S., Chen, S., Liou, Y., and Kuo, Y.: FORMOSAT-3/COSMIC spacecraft constellation system, mission, results, and prospect for follow-On Mission, Terrestrial, Atmospheric and Oceanic Sciences, 20, 19, 2009b.
Fu, E., Zhang, K., Wu, F., Xu, X., Marion, K., Rea, A., Kuleshov, Y., and Weymouth, G.: An evaluation of GNSS radio occultation technology for Australian meteorology, Journal of Global Positioning Systems, 6, 74-79, 2007.
Fu, E.: Assessing space-based GNSS technology for meteorological studies in Australia, Proceedings of ION GNSS 2008, Savannah, Georgia, U.S. , September 16-19 2008.
Fu, E., Zhang, K., Marion, K., Xu, X., Marshall, J., Rea, A., Weymouth, G., and Kuleshov, Y.: Assessing COSMIC GPS radio occultation derived atmospheric parameters using Australian radiosonde network data, Procedia Earth and Planetary Science, 1, 1054-1059, doi:10.1016/j.proeps.2009.09.162, 2009a.
Fu, E., Zhang, K., Silcock, D., Zhou, X., and Kuleshov, Y.: Simulation of radio occultation atmospheric observation events with multiple GNSS for the Australasian region, Asia-Pacific Journal of Atmospheric Sciences, under review, 2011.
Fueglistaler, S., Dessler, A. E., Dunkerton, T. J., Folkins, I., Fu, Q., and Mote, P. W.: Tropical tropopause layer, Review of Geophysics, 47, doi:10.1029/2008rg000267, 2009.
Gao, X.: Towards navigation based on 120 satellites: Analyzing the new signals, PhD, Department of Electrical Engineering, Standford University, 161 pp., 2008.
Garnaut, R.: The Garnaut climate change review, Cambridge University Press, Melbourne, Australia, 2008.
Ge, S. J.: GPS radio occultation and the role of atmospheric pressure on spaceborne gravity estimation over Antarctica, PhD, Graduate School of The Ohio State University, The Ohio State University, 237 pp., 2006.
Gendt, G., Reigber, C., and Dick, G.: Near real-time water vapor estimation in a German GPS network-first results from the ground program of the HGF GASP project, Physics and Chemistry of the Earth: Part A, 26, 413-416, 2001.
122 Gillett, N. P., Stone, D. A., Stott, P. A., Nozawa, T., Karpechko, A. Y., Hegerl, G. C., Wehner, M. F., and Jones, P. D.: Attribution of polar warming to human influence, Nature Geoscience, 1, 750-754, doi:10.1038/ngeo338, 2008.
Gleason, S., and Gebre-Egziabher, D.: GNSS applications and methods, Artech House, Boston, Mass., xvii, 508 pp., 2009.
Gobiet, A.: Radio occultation data analysis for climate change monitoring and first climatologies from CHAMP, PhD, Wegener Center for Climate and Global Change, University of Graz, University of Graz, 149 pp., 2005.
Gorbunov, M., and Lauritsen, K. B.: Error Estimate of Bending Angles in the Presence of Strong Horizontal Gradients, Proceedings of OPAC-3, Graz, Austria, September 17-21, 11 2007.
Gorbunov, M. E., and Kirchengast, G.: Processing X/K band radio occultation data in presence of turbulence: An overview, in: Atmosphere and Climate: Studies by Occultation Methods, edited by: Foelsche, U., Kirchengast, G., and Steiner, A. K., Springer, 183-192, 2006.
Government Accountability Office.: Climate change : financial risks, Novinka Books, New York, viii, 74 pp., 2008.
GRAS Meteorology SAF: GRAS meteorology SAF user requirements, Danish Meteorological Institution; The Met Office of the United Kingdom; Institut d'Estudis Espacials de Catalunya, 36, 2001.
Gubenko, V. N., Pavelyev, A. G., and Andreev, V. E.: Determination of the intrinsic frequency and other wave parameters from a single vertical temperature or density profile measurement, Journal of Geophysical Research, 113, doi:10.1029/2007jd008920, 2008.
Guo, Y., Chang, E. K. M., and Leroy, S. S.: How strong are the Southern Hemisphere storm tracks?, Geophysical Research Letters, 36, doi:10.1029/2009gl040733, 2009.
Guzzi, R.: Exploring the atmosphere by remote sensing techniques, Lecture Notes in Physics, Springer, Berlin ; New York, xii, 262 pp., 2003.
Haan, d. S., Holleman, I., and Holtslag, A. M. A.: Real-Time water vapor maps from a GPS surface network: construction, validation, and applications, Journal of Applied Meteorology and Climatology, 48, 2009.
Hajj, G. A., Ibanez-Meier, R., Kursinski, E. R., and Romans, L. J.: Imaging the ionosphere with the Global Positioing System, Internatioanl Journal of Imaging Systems and Technology, 5, 174-184, 1994.
Hajj, G. A., Lee, L. C., Pi, X., Romans, L. J., Schreiner, W. S., Straus, P. R., and Wang, C.: COSMIC GPS ionospheric sensing and space weather, Terrestrial, Atmospheric and Oceanic Sciences, 11, 235-272, 2000.
123 Hajj, G. A., Ao, C. O., Iijima, B. A., Kuang, D., Kursinski, E. R., Mannucci, A. J., Meehan, T. K., Romans, L. J., Juarez, M. d. l. T., and Yunck, T. P.: CHAMP and SAC-C atmopheric occultation results and intercomparisons, Journal of Geophysical Research, 109, 45, 2002a.
Hajj, G. A., Kursinski, E. R., Romans, L. J., Bertiger, W. I., and Leroy, S. S.: A technical description of atmospheric sounding by GPS occultation, Journal of Atmospheric and Solar-Terrestrial Physics, 64, 451-469, 2002b.
Hayashi, H., Furumoto, J.-I., Lin, X., Tsuda, T., Shoji, Y., Aoyama, Y., and Murayama, Y.: Validation of refractivity profiles retrieved from FORMOSAT-3/COSMIC radio occultation soundings: Preliminary results of statistical comparisons utilizing balloon-borne observations, Terrestrial, Atmospheric and Oceanic Sciences, 20, 8, 2009.
He, W., Ho, S.-p., Chen, H., Zhou, X., Hunt, D., and Kuo, Y.-H.: Assessment of radiosonde temperature measurements in the upper troposphere and lower stratosphere using COSMIC radio occultation data, Geophysical Research Letters, 36, L17807, doi:10.1029/2009gl038712, 2009.
Healy, S. B.: A statistical comparison of GPS/MET radio occultation data with numerical weather prediction analysis, Meteorological Office, 44, 1998.
Healy, S. B., Haase, J., and Lesne, O.: Abel transform inversion of radio occultation measurements made with a receiver inside the Earth's atmosphere, Annales Geophysicae, 20, 1253-1256, 2002.
Ho, S.-p., Kirchengast, G., Leroy, S., Wickert, J., Mannucci, A. J., Steiner, A., Hunt, D., Schreiner, W., Sokolovskiy, S., Ao, C., Borsche, M., von Engeln, A., Foelsche, U., Heise, S., Iijima, B., Kuo, Y.-H., Kursinski, R., Pirscher, B., Ringer, M., Rocken, C., and Schmidt, T.: Estimating the uncertainty of using GPS radio occultation data for climate monitoring: Intercomparison of CHAMP refractivity climate records from 2002 to 2006 from different data centers, Journal of Geophysical Research, 114, doi:10.1029/2009jd011969, 2009.
Hocke, K.: Inversion of GPS meteorology data, Annales Geophysicae, 15, 443-450, 1997.
Hocke, K., and Pavelyev, A. G.: General aspect of GPS data use for atmospheric science, Advances in Space Research, 27, 1313-1320, 2001.
Hocke, K., and Tsuda, T.: Gravity Waves and Ionospheric Irregularities over Tropical Convection Zones Observed by GPS/MET Radio Occultation, Geophysical Research Letters, 28, 2815-2818, 2001.
Hoeg, P., Jensen, A. S., Nielsen, A. S., Lohmann, M. S., Schroder, T. M., Grove-Rasmussen, J., and Larsen, G. B.: Radio Occultation Signal Analysis, Danish Meteorological Institute, Copehagen, Denmark, 38, 2002.
Hoeg, P., and Kirchengast, G.: ACE+ atmosphere and climate explorer based on GPS, GALILEO, and LEO-LEO radio occultation, Institute for Geophysics, Astrophysics and Meteorology, University of Graz, 132, 2002.
124 Hofmann-Wellenhof, B., Lichtenegger, H., and Collins, J.: Global Positioning System : theory and practice, 5th, rev. ed. ed., Springer-Verlag, New York, xxii, 382 pp., 2001.
Hofmann-Wellenhof, B., Lichtenegger, H., and Wasle, E.: GNSS--global navigation satellite systems : GPS, GLONASS, Galileo, and more, Springer, Wien ; New York, xxix, 516 pp., 2008.
Huang, C.-Y., Cheng, C.-Z., Lin, P.-H., Fong, C.-J., Wickert, J., and Hsiao, C.-C.: Radio occultation retrieval of atmospheric profiles from the FORMOSAT-3/COSMIC mission: Early results, Terrestrial, Atmospheric and Oceanic Sciences, 20, 14, 2009.
Huang, C. Y., Kuo, Y. H., Chen, S. H., and Vandenberghe, F.: Improvements in typhoon forecasts with assimilated GPS occultation refractivity, American Meteorological Society, 20, 931-953, 2005.
Huang, C. Y., Kuo, Y. H., Chen, S. Y., Rao, A. S. K. A. V. P., and Wang, C. J.: The assimilation of GPS radio occultation data and its impact on rainfall predication along the West Coast of India during Monsoon 2002, Pure and Applied Geoghysics, 164, 1577-1591, 2007.
Hwang, C., Tseng, T.-P., Lin, T., Svehla, D., and Schreiner, b.: Precise Orbit Determination for the FORMOSAT-3/COSMIC Satellite Mission Using GPS, Journal of Geodesy, accepted: 18 July 2008, 7, 2008.
IPCC: InterGovernmental panel on climate change fourth assessment report: Climate change 2007, InterGovernmental Panel on Climate Change, 2007.
Jacobson, L., and Ebooks Corporation.: GNSS Markets and Applications, GNSS Technology and Applications, Artech House, Norwood, 231 pp., 2007.
Japan Aerospace Exploration Agency: Interface specification for QZSS, Japan Aerospace Exploration Agency, 216, 2009.
Johanson, C. M., and Fu, Q.: Antarctic atmospheric temperature trend patterns from satellite observations, Geophysical Research Letters, 34, L12703, doi:10.1029/2006gl029108, 2007.
Juarez, M. d. l. T., Marcus, S., Dornbrack, A., Schroder, T. M., Kivi, R., Iijima, B. A., Hajj, G. A., and Mannucci, A. J.: Detection of temperatures conducive to Arctic polar stratospheric clouds using CHAMP and SAC-C radio occultation data, Journal of Geophysical Research, 114, doi:10.1029/2008jd011261, 2009.
Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Gandin, L., Iredell, M., Saha, S., White, G., Woollen, J., Zhu, Y., Leetmaa, A., Reynolds, R., Chelliah, M., Ebisuzaki, W., Higgins, W., Janowiak, J., Mo, K. C., Ropelewski, C., Wang, J., Jenne, R., and Joseph, D.: The NCEP/NCAR 40-year reanalysis project, Bulletin of the American Meteorological Society, 77, 437-471, doi:10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2, 1996.
125 Kaplan, E. D., and Hegarty, C.: Understanding GPS : Principles and applications, 2nd ed. ed., Artech House Mobile Communications Series, Artech House, Boston, Mass. ; London, xvii, 703 pp., 2006.
Karl, T. R., Hassol, S. J., Veasey, S. W., and McKay, E. E.: Temperature trends in the lower atmosphere: Steps for understanding and reconciling differences, Washington DC, 180, 2006.
Karoly, D. J., and Braganza, K.: Attribution of recent temperature changes in the Australian region, Journal of Climate, 18, 457, 2005.
Kidd, C., Levizzani, V., and Bauer, P.: A review of satellite meteorology and climatology at the start of the twenty-first century, Progress in Physical Geography, 33, 2009.
King, J. C., and Turner, J.: Antarctic meteorology and climatology, Cambridge Atmospheric and Space Science Series, edited by: Dessler, A. J., Houghton, J. T., and Rycroft, M. J., Cambridge University Press, 1997.
King, M. D.: Our changing planet : the view from space, Cambridge University Press, Cambridge ; New York, xvii, 390 pp., 2007.
Kirchengast, G., and Hoeg, P.: The ACE+ mission: An atmosphere and climate explorer based on GPS, GALILEO, and LEO-LEO radio occultation, in: Occultations for Probing Atmosphere and Climate, edited by: Kirchengast, G., Foelsche, U., and Steiner, A. K., Springer, 201-220, 2004.
Kirchengast, G.: The ACE+ phase a scientific support study ACEPASS: Summary report, Institute for Geophysics, Astrophysics and Meteorology, University of Graz, 20, 2005.
Kirchengast, G., Schweitzer, S., Ramsauer, J., and Fritzer, J.: EGOPSv52 software user manual: Overview manual, Institute for Geophysics, Astrophysics and Meteorology, University of Graz, 70, 2007.
Knudsen, B. M., Christensen, T., Hertzog, A., Deme, A., Vial, F., and Pommereau, J. P.: Accuracy of analyzed temperatures, winds and trajectories in the Southern Hemisphere tropical and midlatitude stratosphere as compared to long-duration ballon flights, Atmospheric Chemistry and Physics Discussions, 6, 5391-5397, 2006.
Kolosov, M. A., Yakovlev, O. I., Efimov, A. I., Pavelyev, A. G., and Matyugov, S. S.: Radio occultation of the Venusian atmosphere and bistatic radiolocation of the surface of Venus using the Venera-9 and Venera-10 satellites, Radio Science, 14, 163-173, doi:10.1029/RS014i001p00163, 1979.
Kramer, H. J.: Observation of the earth and its environment: survey of missions and sensors, Springer, Verlag Berlin Heideberg New York, 2002.
Kueh, M., Huang, C., Chen, S., Chen, S., and Wang, C.: Impact of GPS radio occultation refractivity soundings on a simulation of typhoon Bills (2006) upon landfall, Terrestrial, Atmospheric and Oceanic Sciences, 20, 17, 2009.
126 Kuleshov, Y., Fu, E., Wang, X., Zhang, K., Marion, K., Zhou, X., Liou, Y.-A., and Pavelyev, A.: Spatial patterns of recent atmospheric temperature trends over Antarctica derived from GPS radio occultation observation validated with in-situ radiosonde data Geophysical Research Letters, submitted, 2011a.
Kuleshov, Y., Wang, X., Fu, E., Zhang, K., Feng, Y., and Zwiers, F. W.: Significant lower-stratospheric cooling and tropospheric warming observed over Antarctica since 1956, Natural Atmosphere, Submitted, 2011b.
Kuo, Y. H., Zou, X., and Huang, W.: The impact of Global Positioning System data on the prediction of an extratropical cyclone: an observing system simulation experiment, Dynamics of Atmospheres and Oceans, 27, 439-470, 1997.
Kuo, Y. H., Sokolovskiy, S. V., Anthes, R. A., and Vandenberghe, F.: Assimilation of GPS radio occultation data for numerical weather prediction, Terrestrial, Atmospheric and Oceanic Sciences, 11, 157-186, 2000.
Kuo, Y. H., Wee, T. K., and Bromwich, D. H.: Potential impact of COSMIC GPS radio occultation data on regional weather analysis and prediction over the Antarctic, Radio Occultation Science Workshop, Boulder, Colorado, USA 2002.
Kuo, Y. H., Schreiner, W. S., Wang, J., Rossiter, D. L., and Zhang, Y.: Comparison of GPS radio occultation soundings with radiosondes, Geophysical Research Letters, 32, 4, 2005.
Kursinski, E. R., Hajj, G. A., Hardy, K. R., Romans, L. J., and Schofield, J. T.: Observing tropospheric water vapor by radio occultation using the Global Positioning System, Geophysical Research Letters, 22, 2365-2368, 1995.
Kursinski, E. R.: The GPS radio occultation concept: Theoretical performance and initial results, PhD, California Institute of Technology, California Institute of Technology, Pasadena, California, 377 pp., 1997.
Kursinski, E. R., Hajj, G. A., Leroy, S. S., and Herman, B. M.: The GPS radio occultation technique, Terrestrial, Atmospheric and Oceanic Sciences, 11, 53-114, 2000.
Kursinski, E. R., Flittner, D., Herman, B. M., Feng, D. D., Syndergaard, S., and Ward, D.: An active microwave limb sounder for profiling water vapor, ozone, temperature, geopotential of pressure surfaces and clouds, Proceedings of NASA Earth Science Technology Conference, Pasadena, CA, June 11-13, 2002, 9 2002.
Kursinski, E. R., Ward, D., Otarola, A., Frehlich, R., Groppi, C., Albanna, S., Shein, M., Bertiger, W., Pickett, H., and Ross, M.: The active temperature, ozone and moisture microwave spectrometer (ATOMMS), New Horizons in Occultation Research Studies in Atmosphere and Climate, edited by: Steiner, A., Pirscher, B., Foelsche, U., and Kirchengast, G., Springer, 315 pp., 2009.
Larsen, G. B., Lauritsen, K. B., Rubek, F., and Sorensen, M. B.: Processing of CHAMP radio occultation data using GRAS SAF software, in: Earth Observation with CHAMP: Results from Three Years in Orbit, edited by: Reigber, C., Luhr, H., Schwintzer, P., and Wickert, J., Springer, 543-548, 2005a.
127 Larsen, G. B., Syndergaard, S., Hoeg, P., and Sorensen, M. B.: Single frequency processing of Orsted GPS radio occultation measurements, GPS Solutions, 9, 144-155, 2005b.
Lewis, H. W.: A robust method for tropopause altitude identification using GPS radio occultation data, Geophysical Research Letters, 36, doi:10.1029/2009gl039231, 2009.
Liou, Y.-A., Huang, C.-Y., and Teng, Y.-T.: Precipitable water observed by ground-based GPS receivers and microwave radiometry, Earth Planets Space, 52, 6, 2000.
Liou, Y.-A., Pavelyev, A. G., Liu, S.-F., Pavelyev, A. A., Yen, N., Huang, C. Y., and Fong, C.-J.: FORMOSAT-3/COSMIC GPS radio occultation mission: Preliminary resutls, IEEE Trans. on Geoscience and Remote Sensing, 45, 14, 2007.
Liou, Y. A., Pavelyev, A. G., Huang, C. Y., Igarashi, K., Hocke, K., and Yan, S. K.: Analytic method for observation of the gravity waves using radio occultation data, Geophysical Research Letters, 30, 5, 2003.
Liou, Y. A., Pavelyev, A., Huang, C. Y., and Pavelyev, A. A.: FORMOSAT-3 GPS radio occultation mission, coordinates, 2, 4, 2006.
Liou, Y. A., Pavelyev, A. G., Matyugov, S. S., and Yakovlev, O. I., and Wickert, J.: Radio occultation method for remote sensing of the atmosphere and ionosphere, Intech, Vukovar, Crotia, 2010.
Loiselet, M., Stricker, N., and Menard, Y.: GRAS - Metop's GPS-based atmospheric sounder, European Space Agency Bulletin, 102, 7, 2000.
Loscher, A.: Assimilation of GNSS radio occultation data into GCM fields for global climate analysis, PhD, The University of GRAZ, 217 pp., 2004.
Loscher, A., Foelsche, U., and Kirchengast, G.: CHAMP radio occultation data assimilation into ECMWF fields for global climate analyses, Wegener Center for Climate and Global Change, University of Graz, 61, 2006.
Ma, Z., Kuo, Y.-H., Ralph, F. M., Neiman, P. J., Wick, G. A., Sukovich, E., and Wang, B.: Assimilation of GPS Radio Occultation Data for an Intense Atmospheric River with the NCEP Regional GSI System, Monthly Weather Review, 53, 2010.
Mannucci, A. J., Ao, C. O., Meehan, T. K., Iijima, B. A., Komjathy, A., Yunck, T. P., Pestana, M. K., and Wilson, B. D.: GPS radio occultation as part of the global Earth observing system, Proceedings of Geoscience and Remote Sensing Symposium, 2008. IGARSS 2008. IEEE International, I-308-I-311 2008.
McDonald, A. J., and Hertzog, A.: Comparison of stratospheric measurements made by CHAMP radio occultation and Strateole/Vorcore in situ data, Geophysical Research Letters, 35, doi:10.1029/2008gl033338, 2008.
Meehan, T. K.: The TriG digital beam steered sounder, Proceedings of Aerospace Conference, IEEE, 1-5 2009.
128 Melbourne, W. G., Davis, E. S., Duncan, C. B., Hajj, G. A., Hardy, K. R., Kursinski, E. R., Meehan, T. K., Young, L. E., and Yunck, T. P.: The application of spaceborne GPS to atmospheric limb sounding and global change monitoring, JPL Publication, 1994.
Misra, P., and Enge, P.: Global positioning system : signals, measurements, and performance, 2nd ed. ed., Ganga-Jamuna Press, Lincoln, Mass., 569 pp., 2006.
Mohr, T., and Schmetz, J.: Weather observations from space, in: Utilization of Space: Today and Tomorrow, edited by: Feuerbacher, B., and Stoewer, H., Springer, 20, 2006.
Moran, J. M., and Morgan, M. D.: Meteorology, the atmosphere and the science of weather Burgess Publishing, MN. US., 1986.
Mostert, S., and Koekemoer, J. A.: The science and engineering payloads and experiments on SUNSAT, Acta Astronautica, 41, 401-411, 1997.
National Oceanic and Atmospheric Administration: Federal meteorological handbook: Rawinsonde and Pibal observations, Department of Commerce/National Oceanic and Atmospheric Administration, Washington, 191, 1997.
National Research Council: Earth science and applications from space: National imperatives for the next decade and beyond, The National Academies Press, Washington, 2007.
Nedoluha, G. E., Alfred, J., Benson, C. M., Hoppel, K. W., Wickert, J., and Koenig-Langlo, G.: A comparison of radiosonde and GPS radio occultation measurements with meteorological temperature analyses in the Antarctic vortex, Journal of Geosphysical Research, 112, 9, doi:10.1029/2007JD008928, 2007.
Nilsson, T., Gradinarsky, L., and Elgered, G.: Water vapour tomography using GPS phase observations: Results from the ESCOMPTE experiment, Tellus, 59A, 674-682, 2007.
Notarpietro, R., Zin, A., Perona, G., Corgnati, L., and Gabella, M.: ROSA: The Italian space agency GPS radio occultation reciver singal tracking characteristics and terrestrial measurement Campaign, in: Atmosphere and Climate: Studies by Occultation Methods, edited by: Foelsche, U., Kirchengast, G., and Steiner, A. K., Springer Berlin Heidelberg, 157-168, 2006.
Ogawa, T., Otsuka, Y., Shiokawa, K., Tsugawa, T., Saito, A., Hoshinoo, K., Matunaga, K., Kubota, M., and Ishii, M.: Medium-Scale traveling ionospheric disturbances and plasma bubbles observed by an all-sky airglow imager at Yonaguni, Japan, Terrestrial, Atmospheric and Oceanic Sciences, 20, 9, 2009.
Pavelyev, A., and Kucherjavenkov, I. A.: Refraction attenuation in the planetary atmospheres, Radio Engineering and Electronic Physics, 23, 13-19, 1978.
Pavelyev, A. G., Y.A. Liou, J. W., Gubenko, V. N., Pavelyev, A. A., and Matyugov, S. S.: New applications and advances of the GPS radio occultation technology as recovered by analysis of the FORMOSAT-3/COSMIC and CHAMP data-base, New
129 Horizons in Occultation Research Studies in Atmosphere and Climate, edited by: Steiner, A., Pirscher, B., Foelsche, U., and Kirchengast, G., Springer, 315 pp., 2009.
Piccinini, P., Engberg, J., Colless, E., and Yamagata, H.: Atmosphere, autosphere, biosphere, Drome Pty Ltd, [Fitzroy, Vic.], 47 pp., 2000.
Pingel, D., and Rhodin, A.: Assimilation of radio occultation data in the global meteorological model GME of the German weather service, New Horizons in Occultation Research Studies in Atmosphere and Climate, edited by: Steiner, A., Pirscher, B., Foelsche, U., and Kirchengast, G., Springer, 315 pp., 2009.
Prasad, R., and Ruggieri, M.: Applied satellite navigation using GPS, GALILEO, and augmentation systems, Artech House Mobile Communications Series, Artech House, Boston, cm. pp., 2005.
Randel, W. J., Seidel, D. J., and Pan, L. L.: Observational characteristics of double tropopauses, Journal of Geophysical Research, 112, D07309, doi:10.1029/2006jd007904, 2007.
Rennie, M.: Report on the assimilation of GPS radio occultation data into the Met Office global model, Met Office, 2008.
Rieder, M. J., and Kirchengast, G.: Error Analysis and Characterization of Atmospheric Profiles Retrieved from GNSS Occultation data, Journal of Geophysical Research, 106, 31755-31770, 2001.
Roberts, A., Cassano, J., Doscher, R., Hinzman, L., Holland, M., Mitsudera, H., Sumi, A., and Walsh, J. e.: A science plan for regional arctic system modeling, International Arctic Research Center, University of Alaska Fairbanks, 55, 2010.
Rocken, C., Ware, R. H., Hove, T. v., Solheim, F. S., Alber, C., and Johnson, J. M.: Sensing atmospheric water vapor with the Global Positioning System, Geophysical Research Letters, 20, 2631-2634, 1993.
Rocken, C., Anthes, R. A., Exner, M. L., Hunt, D. C., Sokolovskiy, S. V., Ware, R. H., Gorbunov, M. E., Schreiner, W. S., Feng, D. D., Herman, B. M., Kuo, Y. H., and Zou, X.: Analysis and validation of GPS/MET data in the neutral atmosphere, Journal of Geophysical Research, 102, 29849-29866, 1997a.
Rocken, C., Hove, T. v., and Ware, R. H.: Near real-time GPS sensing of atmospheric water vapor, Geophysical Research Letters, 24, 3221-3224, 1997b.
Rocken, C., Johnson, J. M., Hove, T. v., and Iwabuchi, T.: Atmospheric water vapor and geoid measurements in the open ocean with GPS, Geophysical Research Letters, 32, 13, 2005.
Schmid, A.: Advanced Galileo and GPS receiver techniques : enhanced sensitivity and improved accuracy, Nova Science, Hauppauge, NY, xi, 235 pp., 2009.
130 Schmidt, R., Heise, S., Wickert, J., Beyerle, G., and Reigber, C.: GPS radio occultation with CHAMP and SAC-C: Gloabl monitoring of thermal tropopause paramters, Atmospheric Chemistry and Physics Discussions, 5, 1473-1488, 2005a.
Schmidt, T., Wickert, J., Beyerle, G., and Reigber, C.: Tropical tropopause characteristics from CHAMP, in: Earth Observation with CHAMP: Results from Three Years in Orbit, edited by: Reigber, C., Luhr, H., Schwintzer, P., and Wickert, J., Springer, 561-566, 2005b.
Schwarz, M.: Joint temperature, humidity, ozone, and sea surface temperature retrieval from infrared atmospheric sounding interferometer data, Institute for Geophysics, Astrophysics, and Meteorology, University of Graz, 198, 2004.
Seidel, D. J., Ross, R. J., and Angell, J. K.: Climatological characteristics of the tropical tropopause as revealed by radiosondes, Journal of Geophysical Research, 106, 7857-7887, 2001.
Seko, H., Shoji, Y., Kunii, M., and Aoyama, Y.: Impact of the CHAMP occultation data on the rainfall forecast, in: Data Assimilation for Atmospheric, Oceanic and Hydrologic Applications, 197-217, 2009.
Sgrigna, V., Console, R., Conti, L., Galper, A. M., Malvezzi, V., Parrot, M., Picozza, P., Scrimaglio, R., Spillantini, P., and Zilpimiani, D.: The ESPERIA project: A mission to investigate the near-Earth space, in: Earth Observation with CHAMP: Results from Three Years in Orbit, edited by: Reigber, C., Luhr, H., Schwintzer, P., and Wickert, J., Springer, 407-412, 2005.
Sokolovskiy, S. V., Rocken, C., Lenschow, D. H., Kuo, Y. H., Anthes, R. A., Schreiner, W. S., and Hunt, D. C.: Observing the moist troposphere with radio occultation signals from COSMIC, Geophysical Research Letters, 34, 6, 2007.
Solomon, S., Garcia, R. R., Rowland, F. S., and Wuebbles, D. J.: On the depletion of Antarctic ozone, Nature, 321, 755-758, 1986.
Staten, P. W., and Reichler, T.: Use of radio occultation for long-term tropopause studies: Uncertainties, biases, and instabilities, Journal of Geophysical Research, 113, doi:10.1029/2008jd009886, 2008.
Steig, E. J., Schneider, D. P., Rutherford, S. D., Mann, M. E., Comiso, J. C., and Shindell, D. T.: Warming of the Antarctic ice-sheet surface since the 1957 International Geophysical Year, Nature, 457, 459-462, doi:10.1038/nature07669, 2009.
Steiner, A. K., and Kirchengast, G.: Error analysis for GNSS radio occultation data based on ensembles of profiles from end-to-end simulations, Journal of Geophysical Research, 110, 22, 2005.
Steiner, A. K., Kirchengast, G., Borsche, M., and Foelsche, U.: Lower stratospheric temperatures from CHAMP RO compared to MSU/AMSU records: An analysis of error sources, New Horizons in Occultation Research Studies in Atmosphere and Climate, edited by: Steiner, A., Pirscher, B., Foelsche, U., and Kirchengast, G., Springer, 315 pp., 2009a.
131 Steiner, A. K., Kirchengast, G., Lackner, B., Pirscher, B. M., and Foelsche, U.: Atmospheric temperature change detection with GPS radio occultation 1995 to 2008, Geophysical Research Letters, 36, L18702, doi:10.1029/2009GL039777, 2009b.
Stern, N.: The Economics of Climate Change: The Stern Review, Cambridge University Press, 2007.
Storch, H. v.: Misuses of statistical analysis in climate research, in: Analysis of Climate Variability: Applications of Statistical Techniques edited by: Storch, H. v., and Navarra, A., Springer, 1999.
Sturman, A., and Tapper, N.: The weather and climate of Australia and New Zealand, Oxford University Press, Melbourne, xi, 476 pp., 1996.
Swanson, R. E.: Evidence of possible sea-ice influcence on Microwave Sounding Unit tropospheric temperature trends in polar regions, Geophysical Research Letters, 30, 2040, doi:10.1029/2003GL017938, 2003.
Terzo, O., Mossucca, L., Ruiu, P., Molinaro, M., Notarpietro, R., Cucca, M., and Stigliano, S.: A distributed environment approach for the GPS radio occultation analysis, Proceedings of ICNS '09. Fifth International Conference on Networking and Services, 487-492 2009.
Thompson, D. W. J., and Solomon, S.: Interpretation of Recent Southern Hemisphere Climate Change, Science, 296, 895-899, doi:10.1126/science.1069270, 2002.
Torre, A. d. l., Alexander, P., and Menendez, C. G.: Are we observing mountain waves above the andes range from GPS occultation profiles?, in: Atmosphere and Climate: Studies by Occultation Methods, edited by: Foelsche, U., Kirchengast, G., and Steiner, A. K., Springer Berlin Heidelberg, 243-252, 2006.
Torre, A. d. l., Alexander, P., Llamedo, P., Schmidt, T., and Wickert, J.: Recent advances in gravity wave analysis from long term Global GPS radio occultation observations, New Horizons in Occultation Research Studies in Atmosphere and Climate, edited by: Steiner, A., Pirscher, B., Foelsche, U., and Kirchengast, G., Springer, 315 pp., 2009.
Troller, M., Geiger, A., Brockmann, E., Bettems, J. M., Burki, B., and Kahle, H. G.: Tomographic determination of the spatial distribution of water vapor using GPS observations, Advances in Space Research, 37, 2211-2217, 2006.
Turner, J., Colwell, S., Marshall, G. J., Lachlan-Cope, T. A., Carleton, A. M., Jones, V. L., Reid, P. A., and Iagovkina, S.: Antarctic climate change during the last 50 years, Internatioanl journal of Climatology, 25, 279-294, 2005.
Turner, J., Lachlan-Cope, T. A., Colwell, S., Marshall, G. J., and Connolley, W. M.: Significant Warming of the Antarctic Winter Troposphere, Science, 311, 1914-1917, doi:10.1126/science.1121652, 2006.
Vespe, F., Perona, G., Cosmo, V. D., Petitta, M., Materassi, M., Tartaglione, N., Zin, A., Notarpietro, R., Benedetto, C., Casotto, S., Speranza, A., and Sutera, A.: ROSA – The
132 Italian radio occultation mission onboard the Indian OCEANSAT-2 satellite, New Horizons in Occultation Research Studies in Atmosphere and Climate, edited by: Steiner, A., Pirscher, B., Foelsche, U., and Kirchengast, G., Springer, 315 pp., 2009.
Vey, S., and Dietrich, R.: Validation of the atmospheric water vapour content from NCEP using GPS observations over Antarctica, in: Geodetic and Geophysical Observations in Antarctica, 125-135, 2008.
Wang, K.-Y., Lin, S.-C., and Lee, L.-C.: Immediate impact of the Mt Chaiten eruption on atmosphere from FORMOSAT-3/COSMIC constellation, Geophysical Research Letters, 36, doi:10.1029/2008gl036802, 2009.
Wang, X. L., Wen, Q. H., and Wu, Y.: Penalized maximal t test for detecting undocumented mean change in climate data series, Journal of Applied Meteorology and Climatology, 46, 916-931, doi:10.1175/JAM2504.1, 2007.
Wang, X. L.: Accounting for autocorrelation in detecting mean shifts in climate data series using the penalized maximal t or F test, Journal of Applied Meteorology and Climatology, 47, 2423-2444, doi:10.1175/2008JAMC1741.1, 2008a.
Wang, X. L.: Penalized maximal F test for detecting undocumented mean shift without trend change, Journal of Atmospheric and Oceanic Technology, 25, 368-384, doi:10.1175/2007JTECHA982.1, 2008b.
Wang, X. L., and Feng, Y.: RHtestsV3 user manual, Environment Canada, 28, 2010.
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.: GPS sounding of the atmosphere from low earth orbit: Preliminary results, American Meteorological Society, 77, 22, 1996.
Ware, R. H., Exner, M. L., Herman, B. M., Kuo, Y. H., Meehan, T. K., and Rocken, C.: GPS/MET preliminary report, The University Corporation for Atmospheric Research, 20, 1995.
Ware, R. H., Alber, C., Rocken, C., and Solheim, F. S.: Sensing integrated water vapor along GPS ray paths, Geophysical Research Letters, 24, 417-420, 1997.
Ware, R. H., Solheim, F. S., Carpenter, R., Gueldner, J., Lijegren, J., Nehrkorn, T., and Francois, V.: Radiometric profiling of tropospheric temperature, humidity and cloud liquid, Radio Science, 38, 11, 2001.
Wee, T. K., Kuo, Y. H., Bromwich, D. H., and Monaghan, A.: Assimilation of GPS radio occultation refractivity data from CHAMP and SAC-C missions over high southern latitudes with MM5 4DVAR, Monthly Weather Review, 136, 2008.
Wickert, J., Reigber, C., Beyerle, G., Konig, R., Marquardt, C., Schmidt, T., Meehan, T. K., Grunwaldt, L., and Galas, R.: GPS radio occultation with CHAMP: Initial results, Beacon Satellite Symposium, Boston, U.S., June 4-6 2001a.
133 Wickert, J., Schmidt, T., Marquardt, C., Reigber, C., Neumayer, K. H., Beyerle, G., Galas, R., and Gnmwaldt, L.: GPS radio occultation with CHAMP: First results and status of the experiment, Scientific Assembly, Budopest, Hungary, September 2-8, 2001 2001b.
Wickert, J., Beyerle, G., Schmidt, T., Healy, S. B., Heise, S., Konig, R., Michalak, G., and Rothacher, M.: GPS based atmospheric sounding with CHAMP: Results achieved after four years, EUMETSAT Meteorological Satellite Conference, Dubrovnik, Croatia, September 19-23, 2005 2006.
Wickert, J., Michalak, G., Schmidt, T., Beyerle, G., Cheng, C.-Z., Healy, S. B., Heise, S., Huang, C.-Y., Jakoski, N., Kohler, W., Mayer, C., Offiler, D., Ozawa, E., Pavelyev, A. G., Rothacher, M., Tapley, B., and Arras, C.: GPS radio occultation: Results from CHAMP, GRACE and FORMOSAT-3/COSMIC, Terrestrial, Atmospheric and Oceanic Sciences, 20, 16, 2009a.
Wickert, J., Schmidt, T., Michalak, G., Heise, S., Arras, C., Beyerle, G., Falck, C., K¨onig, R., Pingel, D., and Rothacher, M.: GPS radio occultation with CHAMP, GRACE-A, SAC-C, TerraSAR-X, and FORMOSAT-3/COSMIC: Brief review of results from GFZ, New Horizons in Occultation Research Studies in Atmosphere and Climate, edited by: Steiner, A., Pirscher, B., Foelsche, U., and Kirchengast, G., Springer, 315 pp., 2009b.
World Meteorological Organization: Guide to meteorological instruments and methods of observation, Seventh ed., Publications Board, Geneva, Switzerland, 2008.
World Meteorological Organization: Progress report on the implementation of the global observing system for climate in support of the UNFCCC 2004-2008, World Meteorological Organization, Switzerland, 112, 2009.
Xie, F. Q.: Development of a GPS occultation retrieval method for characterizing the marine boundary layer in the presence of super-refraction, PhD, Department of Atmospheric Sciences, The University of Arizona, 134 pp., 2006.
Xu, G.: GPS : Theory, algorithms and applications, Springer, Dordrecht, 359 pp., 2007.
Xu, X., Zhang, K., Wu, F., and Fu, E.: The impact of LEO satellites' orbital parameters on the pattern of the GPS-LEO radio occultation occurrence, Proceedings of International Global Navigation Satellite Systems 2007, Sydney, Australia, December 4-6, 2007 2007.
Yuan, L. L., Anthes, R. A., Ware, R. H., Rocken, C., Bonner, W. D., Bevis, M. G., and Businger, S.: Sensing climate change using the Global Positioning System, Journal of Geophysical Research, 98, 14925-14937, 1993.
Yue, X., Schreiner, W. S., Lei, J., Sokolovskiy, S. V., Rocken, C., Hunt, D. C., and Kuo, Y. H.: Error analysis of Abel retrieved electron density profiles from radio occultation measurements, Annales Geophysicae, 28, 217-222, 2010.
Yunck, T. P., and Melbourne, W. G.: Spaceborne GPS for Earth science, IUGG Meeting, Boulder, Colorado, July 1995.
134 Yunck, T. P., Liu, C. H., and Ware, R. H.: A history of GPS sounding, Terrestrial, Atmospheric and Oceanic Sciences, 11, 20, 2000.
Yunck, T. P., and Hajj, G. A.: Atmospheric and ocean sensing with GNSS, in: Earth Observation with CHAMP: Results from Three Years in Orbit, edited by: Reigber, C., Luhr, H., Schwintzer, P., and Wickert, J., Springer, 421-430, 2005.
Zeng, Z., Randel, W. J., Sokolovskiy, S. V., Deser, C., Kuo, Y. H., Hagan, M., Du, J., and Ward, W.: Detection of migrating diurnal tide in the tropical upper troposphere and lower stratosphere using the Challenging Minisatellite Payload radio occultation data, Journal of Geophysical Research, 113, 12, 2008.
Zhang, K., Fu, E., Liou, Y.-A., Marshall, J. L., and Kuleshov, Y.: GNSS Radio Occultation Technique and Detailed Evaluation Study Using FORMOSAT-3 Data for Australia, Proceedings of 28th Surveying and Geoinformatics Symposium, Taibei, August 27 - 28 2009.
Zhang, K., Fu, E., Silcock, D., Pavelyev, A., Liou, Y.-A., and Kuleshov, Y.: The investigation of seasonal temperature trends in the Antarctic using CHAMP GPS radio occultation, Atmospheric Measurement Techniques, submitted, 2011.
Zou, X., Liu, H., Anthes, R. A., Shao, H., Chang, J. C., and Zhu, Y. J.: Impact of CHAMP radio occultation observations on global analysis and forecasts in the absence of AMSU radiance data, Journal of the Meteorological Society of Japan, 82, 533-549, 2004.
135