CODE precise clock products have been noted to include the standard deviation at a lower rate than the clock bias. Did not spot Ginan interpolating these standard deviations, not in readRnxClk() or pclkMapClk() which appears to interpolate the variance between the epochs associated with the clock biases and thus have a zero standard deviation for most clock biases. Doubt these standard deviations make much difference to the results, and many products do not even include the standard deviation.
E.g. Using COD0MGXFIN_20260760000_01D_30S_CLK.CLK as an example, which has the software 'CCRNXC V5.5' in the header, and looking at G10, note that the final column of values is the standard deviation and is included only on the 5 minute epochs.
AS G10 2026 03 17 00 00 00.000000 2 -0.590324104522E-03 0.120688754584E-10
AS G10 2026 03 17 00 00 30.000000 1 -0.590324120886E-03
AS G10 2026 03 17 00 01 00.000000 1 -0.590324130143E-03
AS G10 2026 03 17 00 01 30.000000 1 -0.590324122385E-03
AS G10 2026 03 17 00 02 00.000000 1 -0.590324140349E-03
AS G10 2026 03 17 00 02 30.000000 1 -0.590324126967E-03
AS G10 2026 03 17 00 03 00.000000 1 -0.590324130847E-03
AS G10 2026 03 17 00 03 30.000000 1 -0.590324130616E-03
AS G10 2026 03 17 00 04 00.000000 1 -0.590324132658E-03
AS G10 2026 03 17 00 04 30.000000 1 -0.590324135630E-03
AS G10 2026 03 17 00 05 00.000000 2 -0.590324145071E-03 0.127851609557E-10
AS G10 2026 03 17 00 05 30.000000 1 -0.590324155434E-03
AS G10 2026 03 17 00 06 00.000000 1 -0.590324151671E-03
Fwiw, for RTKLIB it is proposed to interpolate the standard deviations to the epochs of the clock biases when reading the rinex clock files, see rtklibexplorer/RTKLIB#833
CODE precise clock products have been noted to include the standard deviation at a lower rate than the clock bias. Did not spot Ginan interpolating these standard deviations, not in readRnxClk() or pclkMapClk() which appears to interpolate the variance between the epochs associated with the clock biases and thus have a zero standard deviation for most clock biases. Doubt these standard deviations make much difference to the results, and many products do not even include the standard deviation.
E.g. Using COD0MGXFIN_20260760000_01D_30S_CLK.CLK as an example, which has the software 'CCRNXC V5.5' in the header, and looking at G10, note that the final column of values is the standard deviation and is included only on the 5 minute epochs.
AS G10 2026 03 17 00 00 00.000000 2 -0.590324104522E-03 0.120688754584E-10
AS G10 2026 03 17 00 00 30.000000 1 -0.590324120886E-03
AS G10 2026 03 17 00 01 00.000000 1 -0.590324130143E-03
AS G10 2026 03 17 00 01 30.000000 1 -0.590324122385E-03
AS G10 2026 03 17 00 02 00.000000 1 -0.590324140349E-03
AS G10 2026 03 17 00 02 30.000000 1 -0.590324126967E-03
AS G10 2026 03 17 00 03 00.000000 1 -0.590324130847E-03
AS G10 2026 03 17 00 03 30.000000 1 -0.590324130616E-03
AS G10 2026 03 17 00 04 00.000000 1 -0.590324132658E-03
AS G10 2026 03 17 00 04 30.000000 1 -0.590324135630E-03
AS G10 2026 03 17 00 05 00.000000 2 -0.590324145071E-03 0.127851609557E-10
AS G10 2026 03 17 00 05 30.000000 1 -0.590324155434E-03
AS G10 2026 03 17 00 06 00.000000 1 -0.590324151671E-03
Fwiw, for RTKLIB it is proposed to interpolate the standard deviations to the epochs of the clock biases when reading the rinex clock files, see rtklibexplorer/RTKLIB#833