Estimated ZTD shows negative values

[sandeepasl]

The output ZTD estimated by goGPS shows negative values when processed with an observation weighting scheme based on square of sine of satellite elevation angle. The following figure shows the ZTD plots obtained by processing a RINEX observation from a U-blox device with three different weighting schemes (left - uniform, middle - sin of satellite elevation, right - square of sine of satellite elevation). The observation file was of total duration 30 minutes, and it was generated one day before processing. Hence, goGPS was configured to use rapid/real-time products. Neill mapping function with an a-priori zenith delay based on Saastamoinen model was chosen for troposphere. The output rate for exporting tropospheric parameters was configured to be 30s (sampling time in the RINEX file). The observation file was processed with GPS constellation alone.

As can be seen, the right figure (corresponding to a weighting scheme using square of sine of the satellite elevation) shows negative values for ZTD. There is a ZTD difference of approximately 10 cm between the first two (left and middle) figures also. Is this expected when we change the weighting scheme from uniform to sine of the satellite elevation angle?

The following figure shows the output when the same data is processed with GPS+GLONASS constellations. The order of the plots is same as above.

Here, we can see that the ZTD is on the higher side even with uniform weighting. It shows very large and unexpected values when the weighting scheme is square of sine of the satellite elevation.