Ins rumental Biases in Ionospheric Measurement s

I.inc-of-sight ionosphere measurements derived from differcncing dual-frequency Global Positioning System (GPS) range data are corrupted by instrumental biases in both the receiver and GPS satellite transmitters due to hardware delays in the L1 and L2 signal paths, The lineof-sight differential delay can be modeled as the sum of a rccciver bias, a satellite transmitter bias, and the line-ofsight ionospheric delay or TEC (total electron content). While the receiver bias can be calibrated directly for some types of receivers, the satellite biases must be estimated from the GPS data itself by using a model of the ionosphere. Ignoring the satellite (receiver) biases when computing TEC measurements from GPS will result in an error of i-9 (*30) TECU (1 TEC unit = 1016 electrons/mctcr2). Using a global ionospheric shell model to fit GPSbased ionospheric delay data from a world-wide network of 3040 receivers, we can estimate, with a single fit, satellite biases for the entire GPS constellation and receiver biases for all the uncalibrated receivers. Current studies indicate that the estimated receiver biases agree with the hardware calibrations at the level of 1 nanosecond (ns) and the dayto-day scatter of the estimated satellite biases ranges from 0.3 to 0.5 ns. Preliminary results show our estimated satellite biases agree with other reported values only at the level of 0.7 ns (RMS difference over all satellites). Further investigation will be required to reconcile these differences. If the true accuracy is 0.5 ns, as derived from day-to-day scatter, then the total uncertainty in line-ofsight TEC measurements derived from GPS is 0.6 ns or 1.8 TECU (1 ns corresponds to 2.85 TECU).