Processing of GNSS data requires proper handling of measurement biases originating from delays within the satellites and receivers. This is especially true for PPP with ambiguity resolution (PPP-AR) which relies heavily on code (pseudorange) observations. The objectives of the IGS Workshop on GNSS Biases 2015, held on November 5-6 in Bern, Switzerland, were therefore to better characterize GNSS biases, describe ways of handling such biases and define standard formats to exchange bias information.
While it is known that code observations are affected by larger noise and multipath that carrier-phase observations, several presentations at the workshop identified additional causes of biases affecting this type of observable. It was shown that changing the correlator spacing or enabling/disabling multipath mitigation techniques within a receiver can cause cm- to dm-level biases (André Hauschild, DLR). Receiver (and satellite) antennas also show elevation (nadir) dependent biases on the same order of magnitude (Lambert Wanninger, TU Dresden). Speaking with Martin Schmitz (Geo++), he confirmed that Geo++ have been calibrating both elevation- and azimuth-dependent code biases for ground antennas since 2008. This information is yet to be included in the ANTEX file since those biases are not entirely receiver-independent for P2 because L1 is often used to help tracking on L2.
Jean-Marie Sleewaegen (Septentrio) explained how frequencies at the edge of antenna filters (such as GLONASS frequencies) are more likely to be affected by biases. He also illustrated how changing the antenna cable length by only 5 cm can result in additional dm-level code biases. Other causes of biases include cable cross-talk and splitters. In a paper published in InsideGNSS in 2012, people from Septentrio had emphasized that a misalignment between code and carrier measurements within a receiver was responsible for GLONASS inter-frequency phase biases. It turns out that this misalignment is also present for GPS in some receivers, resulting in a frequency offset in PPP clock estimates (Pascal Defraigne, ROB).
A few years ago, the RTCM-SC104 proposed to include code and phase biases in the state-space representation (SSR) messages. This information would be transmitted as “undifferenced” biases as opposed to differential code biases (DCBs). People from the Center for Orbit Determination (CODE) also seem favorable to such a representation in a new Bias-SINEX format since the next version of their Bernese software (version 5.3) should replace DCB estimates with signal-specific estimates of biases. This representation has not yet been approved by members of the IGS Bias working group and will require further discussions. This signal-dependent formulation would also allow including code and phase biases for PPP-AR. Finally, it was decided that IONEX version 1.1 will not include DCB values anymore since the Bias-SINEX format will be used to carry this information instead.
All presentations should be made available at: www.biasws2015.unibe.ch