This thesis describes the combination of geodetic observation techniques on-board satellites. This socalled co-location in space provides a considerable potential regarding the improvements needed to realize a long-term accurate and stable terrestrial reference frame. The space ties (i.e., the o set vectors between the on-board sensors) introduces new geometrical connections between sensors of di erent space geodetic techniques. This space ties can be provided easily to each fundamental site via space geodetic observations. Consequently, co-location in space allows to assess technique-speci c error sources as systematic e ects can be addressed either to a certain station or to a certain technique. Moreover, the additional introduced orbit dynamics improve the estimation of several geodetic parameters. Within this thesis the following core topics concerning co-location in space are discussed: orbit determination, the combination of ground and space GNSS observations, and VLBI Earth-orbiting satellite tracking. Highly accurate orbit determination is the prerequisite for a suitable co-location in space. Based on the Earth observation satellite missions GRACE, GOCE, and OSTM/Jason-2 orbit determination and the impact of modeling non-gravitational perturbations is studied. The overall reached orbit accuracies are at the level of a few centimeters. The combination of ground and space-geodetic GNSS observations is studied based on the GPS observations derived by 53 ground stations and the four LEOs (low Earth orbiter). Adding one LEO to the ground-only processing decreases the formal errors of weekly geocenter estimates by around 20% which is eight times more than expected due to the increased number of observations. This shows the considerable potential of the combination of ground and LEO data. Comparing the derived geocenter time series against results from satellite laser ranging (SLR) shows a good agreement for annual amplitudes, whereas the annual phases shows considerable discrepancies in the xand the z-component. Geocenter coordinates derived from surface load density coe cients estimated in a long-term solution show a better agreement to SLR solutions but without a signi cant impact of additional LEOs. Using the gravitational constraint GPS satellite antenna phase center o sets were estimated based on ground and LEO observations. The results show a signi cant bene t for the horizontal o sets as the introduced LEOs help to dissolve limiting correlations. Concerning single-frequency VLBI satellite tracking the L4R method is introduced to derive ionosphere delay corrections based on co-located GNSS observations. A 1 cm daily station coordinate repeatability is achieved in a single-frequency GNSS processing while introducing the L4R corrections. Di erences to ionospheric delays derived from VLBI observations show also a good agreement. As VLBI satellite tracking is currently in an experimental stage Monte-Carlo simulations were performed for eight di erent satellite orbit types. For a GNSS constellation tracking, station coordinate repeatabilities are at the level of 0.7 and 1.2 cm for a regional and a global network, respectively. Station coordinate repeatabilities of around 1 cm were derived for simulated VLBI observation to a ctitious LEO with an altitude of 2000 km. The station coordinates estimated from simulated observations to E-GRIP and E-GRASP/Eratosthenes show larger uncertainties. Based on the results suggestions for future action items regarding co-location in space were formulated. The most important recommendations are, that the combination of groundand space GNSS observations provides a considerable bene t for the determination of several parameters and that ionosphere delay corrections should be derived from co-located GNSS observations.

Co-location of Geodetic Observation Techniques in Space

Tracking L-band signals of GNSS satellites by radio telescopes became a new observation type in recent years and will be used to improve reference system realizations and links between Earth- and space-fixed frames. First successful test observations were done, with the drawback of being single-frequency only. In order to correct the ionospheric delay by using GNSS phase observations from co-located receivers, the L4R approach was developed. Based on residuals derived by a least-squares processing of the GNSS geometry-free linear combination corresponding corrections could be derived. As a first validation step L4R corrections were applied to GNSS $$L_1$$
 data analysis. Station coordinate repeatibilities at the 1-cm level were obtained for baselines of a few thousand kilometers. Comparing the derived delay corrections to VLBI ionospheric delays for quasars located in same directions, differences with a standard deviation of 2.2 TECU could be achieved.

Ionospheric corrections for single-frequency tracking of GNSS satellites by VLBI based on co-located GNSS

GBAS includes reference receivers, processing module, and communication device. Processing module checks GNSS satellite measurements to determine proximity of GNSS satellite measurement's IPP to IGPs derived from SBAS geostationary satellites. Processing module determines that GNSS satellite measurement is safe for mitigation using overbounded Vertical Ionosphere Gradient standard deviation sigma-vig (σ vig ) when IGPs possess acceptable GIVE values. Processing module determines whether number of GNSS satellite measurements determined safe for mitigation using σ vig are able to produce VPL that meets VAL required for precision approach. Communication device communicates overbounded σ vig along with differential corrections and indication of which GNSS satellite measurements that are safe for mitigation using at least one overbounded σ vig are able to produce VPL that meets VAL required for precision approach to GNSS receiver when number of GNSS satellite measurements determined safe for mitigation using overbounded σ vig are able to produce VPL that meets VAL required for precision approach.

Using space-based augmentation system (sbas) grid ionosphere vertical error (give) information to mitigate ionosphere errors for ground based augmentation systems (gbas)

Effect of biases in integrity monitoring for RTK positioning

The implementation of performance-based navigation allows advanced flight operations in areas without conventional navigation aids. Especially helicopter operations, which usually have been conducted under Visual Meteorological Conditions (VMC), may now operate more independently of the weather situation under Instrument Flight Rules (IFR). Nevertheless, at common helicopter IFR flight altitudes, icing conditions are a major threat. Therefore, helicopter operators have a signi¬ficant interest to fly at the lowest possible altitude. High navigation accuracy and reliability is imperative for such flight operations, this holds true especially in mountai¬nous areas. 

In the frame of the Swiss-wide implementation program to promote GNSS procedures and performance based navigation applications (CHIPS), a low-level flight net-work for helicopter operations is being implemented in Switzerland. The newly developed routes are still designed at relatively high altitudes and lowering them closer to the ground is definitively envisaged. For such operations, the RNP 0.3 navigation specification would however not suffice. A more ambitious performance would be in demand. 

The highest risk in low-level flight operations is the proximity to the terrain. Loss of navigation for instance due to an RF-interference event or even because of an insufficient satellite coverage may immediately lead to a hazardous situation. In order to analyze the performance of helicopter operations, 11 Agusta DaVinci helicopters have been equipped with a recording unit to obtain on-board GPS, EGNOS, FMS and flight plan information. As additional data recording measure, a geodetic receiver may be installed on-board the helicopter to record GPS and GLONASS dual-frequency data. These recordings allow the determination of the true flight path. 

Data is recorded in daily operations whenever possible. More than two dozens of flights have been analyzed so far. In addition, a dedicated high dynamic test routing has been developed using straight segments and RF legs featuring extremely narrow radii which are beyond the limitations of the current ICAO criteria. Furthermore, an approach procedure with RNP 0.1 performance including RF legs has been used. The resulting total system error usually remained below 0.02 NM. With few exceptions, all dedicated operations have been flown without issues. 

In summary, the investigation shows that a more ambi-tious RNP navigation accuracy requirement than 0.3 NM is feasible for advanced rotorcraft operations. Such performance-based navigation applications would allow safe flight operations at a lower flight altitude. For HEMS organizations, this would allow a significant increase of the efficiency and a more robust rescue service resulting in human life savings.

Flight Performance Investigations of Enhanced Rotorcraft Operations in Mountainous Areas – Towards a More Ambitious RNP Performance

In current Global Navigation Satellite System (GNSS) differential single-frequency data processing, ionospheric delays are considered to be considerably reduced for short baselines in regional network. However, under strong ionosphere activities the remaining ionospheric effects on the single-difference level and their short-term variations may not be neglected and may become critical for, e.g., the approach and landing of airplanes using a Ground Based Augmentation System (GBAS).

In this paper, the single-difference ionospheric residuals and spatial gradients are studied as a function of the Ionosphere Pierce-Point (IPP) distance under diverse ionosphere activities using data of the Automated GNSS Network for Switzerland (AGNES) over 15 years. A similar study has been performed by Lee et al. (2006) using the Wide Area Augmentation System (WAAS) “supertruth” data and the Jet Propulsion Laboratory (JPL) post-processed Continuously Operating Reference Stations (CORS) data by leveling the carrier-phase-based ionospheric measurements with the help of the code observables. Here, we provide an alternative method to estimate the single-difference ionospheric residuals and spatial gradients based on double-difference phase observations with resolved ambiguities.

After pre-processing of the phase observations, the phase ambiguities can be resolved with the help of the Bernese GPS Software. The geometry-free (GF) linear combination of the phase observations is then generated on the double-difference level without the code-related error sources like differential code biases (DCBs) and arc-wise biases resulting from code smoothing. Only the ionospheric delay, the phase noise and biases like multipath remain in the residuals. However, singularities exist when deriving single-difference ionospheric delays from double-difference residuals. This is corrected for by imposing, e.g., a zero-mean condition on the single-difference residuals per baseline and per observation epoch. In order to correct for the lack of absolute information in the double-difference residuals mentioned above, model values of the ionospheric delays are generated for each observation using the CODE’s global ionosphere maps. A mean model ionospheric offset is then added epoch-wise to the estimated single-difference ionospheric delays per baseline to supply absolute information and the zero-mean condition is then applied to the deviations from the mean ionospheric offset. Local ionosphere maps may be considered in the future to produce more appropriate mean model ionospheric offset. It should be mentioned that small epoch-wise biases may still exist in each of the ground baselines because of the spatial and temporal smoothing effects of the ionosphere model.

The single-difference spatial ionospheric gradients are estimated with the single-difference ionospheric delays and the pierce-point baseline length, namely the distance between the two pierce points of the “thin-shell” ionosphere model. We categorize the ionosphere activity into different levels based on geomagnetic indices, e.g., the Kp and Dst Index (see Lee et al. 2006), and analyze the correlation between the slant differential ionospheric delays and the geomagnetic indices as well as the daily Total Electron Content (TEC) provided by the Center for Orbit Determination in Europe (CODE). Using an elevation-dependent mapping function, the vertical single-difference ionospheric delays and gradients are also calculated and studied for all available days except the very stormy days. Very short ground baselines for the purposes of investigation of the non-ionospheric biases and ground baselines longer than 300 km are included in the study.

Studies with long-term time series and the entire network of the AGNES stations have delivered a general relationship between the ionospheric delays, gradients and the ionosphere activity. The differential slant and vertical ionospheric delays and gradients were calculated and analyzed from 1999 to 2013. We see that the change of the differential ionospheric delays and gradients corresponds to the ionosphere activities. For very short baselines, the mean single-difference ionospheric residuals are at mm-level with a standard deviation at cm-level.

In the end, the vertical ionospheric gradients are normalized in specific distance bins and an inflated sigma which bounds all the outliers and non-Gaussian tails of the normalized vertical ionospheric gradients has been determined. This part of the analysis is based on the work of Lee et al. (2006) for all the ionosphere nominal and active days. The daily peak slant ionospheric gradients are further analyzed for all available days under different ionosphere conditions with the help of the Conterminous United States (CONUS) ionosphere anomaly threat model (see Pullen et al. 2009).

This work has been financed by Flughafen Zurich AG as part of the Skyguide (Swiss Air Navigation Services Ltd.) project "Impact of Ionospheric Activities onto GNSS Signal during Approach and Landing" within the frame of the Swiss-wide program to implement new technologies (CHIPS). We would also like to thank swisstopo for providing the AGNES data for the processing.

References:
Lee J., Pullen S., Datta-Barua S. and Enge P.: Assessment of Nominal Ionosphere Spatial Decorrelation for LAAS; Position, Location, and Navigation Symposium, 2006 IEEE/ION, Coronado, CA, USA, pp. 506-514, April 2006.
Pullen S., Park Y.S. and Enge P.: Impact and mitigation of ionospheric anomalies on ground-based augmentation of GNSS, Radio Science 44, 2009, doi:10.1029/2008RS004084.

Assessment of Single-difference Ionospheric Residuals in a Regional Network for GBAS

Noise protection has an increased importance for approach und departure procedures in vicinity of airports. Flight tracks should be focused over sparsely populated areas to the greatest extent possible. ICAO's new amendment of the Performance-Based Navigation manual will allow Radius-to-Fix (RF) legs during initial and intermediate segments of an approach. Standards for procedure design are currently in development. However, before starting the final approach, a straight segment for aircraft stabilization is required in case of xLS approaches. 

Switzerland has many congested and extensively fragmented airspace due to various user requirements. Airspace changes are extremely complex and time-consuming. Consequently, the use of airspace should be optimized as far as possible, also in respect of noise-sensitive areas. Applications with RF legs are a major enabler for such advanced procedures. To minimize airspace use and to optimize fuel consumption, it is even desired to have various RF legs in one approach procedure with the last one ending directly at the beginning of the final approach segment. 

In the frame of the Swiss-wide implementation program to promote GNSS procedures and applications, test flights were carried out to analyze the use of various RF leg applications in an approach procedure. Two different approach scenarios were developed. The first scenario was designed according current ICAO standards with one RF leg and a straight stabilization segment of 2 NM before the final approach segment. The second scenario includes three RF legs, the first one to the right, followed immediately by a second and third leg, both, to the left. Latter ends directly at the start of the final approach segment. This scenario is not yet covered by current ICAO criteria. A maximum bank angle and wind conditions without safety margins dedicated to the deployed aircraft were applied to go to the limits of what is technically possible. Both scenarios were designed with an ILS and RNAV final approach segment. 

Test flights have been carried out with a Super King Air of the Swiss Air Force. This aircraft is equipped with the Pro Line 21 system having RF leg capability, a Rockwell Collins GPS4000S GPS/SBAS receiver and a RTCA DO260A-compliant ADS-B out transponder. 

A total of 13 approaches have been performed. The aircraft has been equipped with a geodetic GNSS receiver in order to record raw GPS measurements on both, the L1 and L2 band. These measurements have been post-processed using data of the permanent automated GNSS network for Switzerland to determine the true flight path with an accuracy of better than 10 centimeters. The navigation system flight path has been determined by recording ADS-B messages. Finally, the navigation system error and the flight technical error have been calculated and analyzed. 

The test flight results have been analyzed concerning compliance to the ICAO RNP1 specification. The flights have been performed with different settings of the autopilot and the flight director. Furthermore, configurations with deselected or switched off GPS and deselected DME have been tested. 

The performance boundaries of RNP 1 could be maintained for all flights. Even abnormal conditions with AHRS (Attitude Heading Reference System) only resulted in an acceptable performance. The scenario with an ICAO-compliant RF leg followed by a stabilization segment before the final approach segment did never show any significant deviation from the designed flight track. 

The scenario with three RF legs currently not covered by the ICAO design criteria did show surprising results. The transition from the last RF leg to the final approach segment without a straight intermediate segment did not show any significant deviation from the designed flight path. This holds true for the interception to an RNAV as well as to an ILS final approach segment. The so-called "S-design" of RF legs, i.e. an RF leg immediately followed by an opposite RF leg is in line with the current ICAO design criteria. However, the test flights showed significant lateral deviations of up to 0.3 NM for this scenario if the approach is performed with the autopilot. 

This paper describes the test flights carried out and the analyses performed to obtain important findings for the enhancement of flight design criteria in respect of flexibility of approach procedures and avoidance of flights over noise-sensitive areas.

M. Troller

Papers

Assessment of Single-difference Ionospheric Residuals in a Regional Network for GBAS

Curved Approach Segments for Noise Abatement