| The research on the atmosphere of Earth, which provides the shield that human beings depend on for survival, has been always one of the hot topics. Comparing to the traditional atmospheric sounding methods, the radio occultation (RO) technique, developed in the1990s, has the advantages of high precision, high vertical resolution, global coverage, all-weather, self-calibration, and long-term stability, and is very suitable for exploiting Earth's atmosphere. The RO technique not only complements the deficiency of the traditional sounding methods, but also contributes significantly to improve numerical weather prediction (NWP) and the better ability to monitor global climate change. Hence, the RO technique plays an important role on the research in Meteorology and Climatology.In RO, the ionospheric error is one of the main error sources. However, in RO data processing, the error is only eliminated partly by the ionosphere-free dual-frequency linear combination, and the higher-order errors still remain and have an impact on the retrievals of neutral atmospheric parameters. During solar maximum condition, the higher-order errors may even affect the retrievals of the atmospheric parameters at25km, which limits its application in Atmospheric Sciences. In order to advance the RO product retrieval accuracy and improve its significance in Atmospheric Sciences, the research on the ionospheric error in RO is very important. Therefore, the thesis mainly focuses on the research of the ionospheric error, and its main contents are summarized as follows:1. Starting with the historical development of the RO plans, the research progresses of the RO technique and its significance in Atmosphere Sciences are reviewed. Then the progresses of the research on the ionospheric error in GNSS data processing, especially the research progresses in RO and their shortcomings, are also reviewed, and the research significance is pointed out as well.2. The theoretical foundation of the RO technique, Helmholtz equation, is introduced, as well as the eikonal equation and Bouguer law in geometrical optics. The basic processing procedures of the RO retrieval method based on geometrical optics approximation are described in detail, and the key steps are also explained. As a commonly-used simulation method in RO, the three-dimension (3D) ray tracing is also introduced in detail, which is the basis for the subsequent research.3. The error formulas of the atmospheric parameters including bending angles, refractivity, temperature and pressure, resulted by the errors in excess phase, are deduced elaborately. Based on the real RO orbits of GNSS and LEO with uniformly temperal and spatial distribution, as well as the neutral atmospheric model, the retrieval errors induced by the Earth's oblateness, the ephemeris error, the satellite clock oscillator stability, the local multipath and the thermal noise are statistically analyzed.4. According to the characteristics of the ionospheric error, it's divided into four parts:the neutral delay error and excess path length induced by the ray bending in the ionosphere, and the ionospheric first and second order term. Based on the classification above, their relationships with the perigee height, the solar activity level, the geographic location and local time are studied overall. And it's found that in RO the TEC difference induced error Asi and excess path length?slen dominate over the other two residual errors, however the residual neutral delay error?snen induced by the ray dispersity/bending in the ionosphere is almost negligible after the ionosphere-free dual-frequency linear combination. Moreover, a clear bulge due to the E layer is found in the two residual errors?s1and?slen, which are positively proportional to TEC rate.5. For the first time, the relationship between the second order residual error?s2and RO azimuth is discussed, and the sinusoidal variation of?s2with RO azimuth is found, whose maximum positive and negative values occur at RO azimuths0°(southward) and180°(northward) respectively. Also the retrieval biases of the neutral atmospheric parameters induced by?s2have the sinusoidal features with RO azimuth, but have opposite variation trends to that of As2, which means that the RO azimuths of the maximum positive and negative retrieval biases approximately correspond to the azimuths of maximum negative and positive second order errors respectively, and the maximum and negative retrieval biases are located at RO azimuths180°(northward) and0°(southward) separately. At RO azimuth0°, the second errors?s2are almost positive all over the world, and are characterized by three peaks due to the ionospheric equatorial anomaly, the inclination of the geomagnetic pole, and the feature of RO ray path propagating with low elevation, which are about1.2-1.6cm during moderate solar activity level and located at about30°N?45°N?0°N?15°N and15°S?30°S respectively.6. According to the classifications of the ionospheric error above, the differences between the excess phase combination at epoch (PCE) and the bending angle combination at impact parameter (BCI) are compared for the first time in RO. Based on the comparison between PCE and BCI, the excess phase combination at impact parameter (PCI) in RO is proposed, and the differences between PCE and PCI are also analyzed. Based on the comparison between PCE and PCI, a new ionosphere-free combination PCEI is proposed in RO, which mixes PCE and PCI together and has better accuracy than either of them. The PCEI method is validated with the simulated and real RO data from COSMIC. The results show that the PCEI method is comparable to the BCI method, but better than the PCE method, and the temperature profiles inverted by the PCEI method also agree well with the profiles at CDAAC, with differences less than1.0K at15-35km.7. As the systematic residual ionospheric errors (RIEs) still exist in the PCEI method and multi-frequency signals can be received in the future GNSS, such as GPS, Galileo and Beidou systems, the triple-frequency linear combination in RO is analyzed for the first time. And the results show that the systematic RIEs after the traditional dual-frequency linear combination on L-band can be eliminated at the millimeter level and the random error amplification is also very small, when including a C-band frequency. The almost unbiased temperature profile at0-40km can be inverted by the dual-or triple-frequency combinations on C-and L-bands, which is very suitable for monitoring decadal-scale climate change. However, comparing to the dual-frequency L-band-only combination, the retrieval accuracy of the temperature profile, inverted by the high-noise triple-frequency combination on L1, L2and L5, isn't improved well, which needs further development of a valid filtering algorithm.8. Based on the PCEI method, the RIEs and their impacts on temperatures in RO are both quantified for the first time. The results show that the RIEs and the temperature errors for day events are larger than those for night events, however, they have littile differences with each other for the day and night events at high latitude. The RIEs and the temperature errors at low and middle latitudes are largher than those at high latitude. |
PDF Full Text Request