The same questions as in Issue 1 arise, but in this case both the mean differences and the spread seem larger. In theory different GPS-RO satellite systems should give similar results. A comparison of co-located soundings (again by latitude bands as well as land and ocean) from the different satellite systems could prove its reliability.
Response from Kevin Trenberth:
Even for the same satellite and sensor, different groups produce slightly different results and retrieval procedures are still advancing. It is for this reason that GPSRO is not yet close to being a benchmark.