Why Delay Can Improve Precision
Waiting to compute a position is not merely a consolation prize for missing connectivity. In a very concrete sense, waiting provides access to genuinely better information than any live correction, RTK included, ever has available in the moment.
Satellite orbit and clock descriptions improve substantially in the hours and days after a signal is transmitted, as tracking networks refine their estimates. Waiting to process a position allows that improved information to be used, something no real-time method can access at the moment it is needed.
Broadcast Information Is a First Draft
The orbit and clock information a satellite broadcasts in real time is necessarily a prediction, computed and uploaded to the satellite in advance, describing where the satellite is expected to be and how its clock is expected to behave. It is good, but it is a forecast, not a final measurement. International tracking networks continuously observe the actual GNSS constellation from ground stations worldwide and, after the fact, compute a far more precise, retrospective description of exactly where each satellite really was and how its clock really behaved.
A satellite's broadcast position is the best available prediction at transmission time. A precise, after-the-fact orbit product is closer to a measurement, informed by tracking data that simply does not exist yet at the moment a live correction has to be generated.
A Range of Products, Not a Single Upgrade
These retrospective products are not released all at once or with a single fixed delay. An early, already-improved version, commonly called Ultra-Rapid, becomes available within hours, useful for near-term post-processing where some delay is acceptable but a full analysis cycle is not. A more refined version, Rapid, follows within a day or two. The most precise version of all, Final, incorporating the most complete tracking analysis, is not finalized for roughly two to three weeks. Post-processing software can select whichever of these products fits the project's own timeline and precision needs.
Waiting to process a position is not simply waiting for the same broadcast information to become available. It is waiting for a genuinely more accurate, independently verified description of the satellites' actual behavior, one that improves the longer the wait, up to a point.
Solving the Problem in Both Directions at Once
A live method is bound by something no amount of better hardware can change: at the moment a position is needed, only data from before that moment exists. A logged dataset has no such restriction. Software like RTKLIB can process the same recorded session forward, backward, and then combine both passes into a single solution, using data from both before and after any given instant to resolve ambiguities the forward-only or backward-only pass might have missed. In practice this often means a stretch of FLOAT that a forward-only pass never resolves gets FIXED once the same data is also processed in reverse. This advantage belongs to any method working from a complete, logged dataset. It's not unique to PPK; the same technique applies to post-processed PPP and any other delayed method built on recorded rather than streamed data.
What This Means in Practice
A project willing to wait even a day or two after collection often gains meaningfully improved satellite orbit and clock information over what any real-time method, RTK included, could have used live. A project willing to wait several weeks gains access to the most refined version available at all. This is the concrete trade PPK is built around: giving up the moment of collection as the moment of certainty, in exchange for a better final answer.
Two positions are computed from the exact same raw satellite observations, one processed immediately using broadcast information, one processed two weeks later using the most refined available orbit and clock products. Should these be expected to agree exactly? Not necessarily, and the delayed version is generally the more trustworthy of the two, precisely because of the better information it had access to.
Satellite orbit and clock information genuinely improves in the hours, days, and weeks after transmission, as tracking networks refine their estimates. Delaying a computation is not simply a workaround for missing connectivity. It provides access to better information than any real-time method has available at the moment it is needed, and it allows the solution itself to be computed using data from both directions in time, an option no live method can ever have.
Better information and no live connectivity requirement both sound like clear wins. The next chapter looks at exactly which situations make that trade worth making, and which do not.