synopsis: (1) residual systematic uncertainty, in case the stability criterion is satisfied, the residual uncertainty is less than 5%, (2) residual systematic uncertainty, in case the stability criterion is not satisfied the systematic uncertainty is higher than 5%, (3) statistical uncertainty increase due to decrease in effective integration time for partial stable measurements, (4) residual systematic uncertainty: the independent FCS measurement can be stable, partly stable or unstable, causing a additional systematic uncertainty to the sky measurements.
limitations and applicability:
In case the stability criterion is not met, no signals are discarded.
A minimum of 7 signals per measurement or raster point is required to
perform the analysis.
description:
Transient recognition determines whether there is a significant systematic
signal variation in a measurement by applying the Mann algorithm. In case
such variation is present the first half of the signals is discarded,
this is repeated until a stability criterion is achieved or 7 signals are
remaining. The criterion for rejection is a variation in excess of 5%.
purpose correction:
Transient responsivity variation is an important cause of systematic
error which is minimized by this correction.
uncertainty/noise introduced:
The magnitude of the transient variation depends on the type of
detector, flux step (positive or negative), and exposure time.
A large flux step to a higher flux under-estimates and a flux step to
lower fluxes over-estimates the ``true'' (i.e. stabilized at t=
)
measurement signal.
In multi-filter and multi-aperture AOTs the flux step is tuned to become
positive. However, the flux of the FCS is often lower than the last sky
measurement (i.e. negative step) which yields a negative systematic
uncertainty because the detector responsivity is overestimated.
The null hypothesis for drift recognition is 5%. Therefore, if the
integration time is sufficiently long (>32 sec) and the transient
recognition algorithm yields a stable signal after the iteration, then
it is most likely that the derived signal is within 5% of the ``true''
signal level. The same holds for the FCS measurement signal, which adds
also a systematic uncertainty of at most 5% to the photometry via the
systematic uncertainty in the responsivity.
auxillary data:
Parameters for the transient recognition algorithm:
Nmin = 7, minimum number of signals for correction;
= 0.05, probability that the null-hypothesis is rejected.
keyword(s):
| DRALPHA: | significance level | SPD header |
| DRMINPTS: | min. no. of points for Mann-test | SPD header |
| FITDRn: | Drift fit flag (T or F, n=measurement number) | SPD header |