The NRCan PPP Service: Towards Fast Convergence

Abstract

The Canadian Geodetic Survey (CGS) of Natural Resources Canada (NRCan) has been operating an online positioning service since 2003. This application processes user-submitted global navigation satellite system (GNSS) data using the precise point positioning (PPP) methodology. With precise satellite orbit and clock corrections, dual-frequency users can achieve cm-level positioning accuracies with respect to a global reference frame, regardless of the density of nearby reference stations. The online PPP service offered by CGS currently processes over a thousand submissions daily, a clear indication of the usefulness of this approach. Augmentation is an emerging trend in PPP data processing. It is characterized by additional state-space corrections such as carrier-phase and code biases for integer ambiguity resolution (AR), as well as atmospheric delay (tropospheric and ionospheric) corrections. An important benefit of these corrections is a reduction in the initial convergence time of the PPP filter, which otherwise limits the applicability of PPP for several applications with short on-site occupation times. Since the current implementation of the NRCan PPP service does not benefit from augmentation, modernization efforts are in development to: 1) enable ambiguity resolution based on the decoupled-clock model (DCM); 2) generate precise slant ionospheric delay corrections across Canada; 3) estimate additional equipment delays for single- and triple-frequency PPP-AR users. NRCan has been among the first organizations to generate satellite clock corrections preserving the integer nature of the carrier-phase ambiguities on the user end. This product is termed decoupled clocks since combining signal-dependent equipment delays (such as carrier-phase and code biases) with clock parameters effectively leads to a different clock correction for each signal. As a result, ambiguity resolution can be achieved, which is a key component in reducing the initial convergence period. GPS DCM clocks are routinely being estimated in both real time and post processing to eventually serve users submitting data with various latencies. The use of DCM clocks typically reduces PPP convergence times from approximately 60 to 30 minutes. To further reduce the convergence period of PPP solutions, slant ionospheric delays are generated daily from a cooperative network of approximately 200 permanent stations in Canada. The spatial distribution of these stations varies widely across the country, with inter-station distances of about 50 km in certain regions to more than a few hundred kilometers in northern territories. Ionospheric delays are computed at 30-second intervals, directly from PPP-AR solutions. These precise slant delays are interpolated at the user position on a satellite-by-satellite basis using the three nearest stations. This approach was preferred to standard ionospheric maps due to limitations in the spatial and temporal resolutions of these products, thereby leading to better ionospheric predictions. The gain in performance strongly depends on the proximity of a user to the network of reference stations, with quasi-instantaneous convergence in ideal scenarios. For users located outside of the cooperative network geographical limits, ionospheric constraints are applied using global ionospheric maps (GIMs). To ensure compatibility with the DCM clock products, CGS estimates daily differential phase biases in addition to differential code biases (known as DCBs). Theoretically, these biases also allow single-frequency users to perform ambiguity resolution. While this concept is demonstrated with geodetic-quality receivers, it has not yet been tested with low-cost receivers. Nevertheless, with precise slant ionospheric constraints, precise positioning can be achieved for single-frequency users located in areas populated by reference stations. RTK-like accuracies can even be obtained if a user is located within a few kilometers of a reference station. Additional satellite phase and code biases are also derived to serve triple-frequency users and enable fast (e.g., 10 minutes) convergence, regardless of the proximity of reference stations. The performance of these new implementations is assessed for single-, dual-, and triple-frequency users, and is compared against the actual service offered by NRCan. While an emphasis is made for Canadian users, data sets from globally distributed stations are also analyzed.