Spikes and Steps

However, there are still significant outliers: the minimum and maximum pixel values are $\sim$8.1 and 3.4, respectively. Using GAIA to examine the time series in the DC-subtracted cube reveals remaining data issues as shown in the Fig. .

Figure: Example time series in the DC-corrected data set showing spikes, steps, and noisy bolometers. The dominant common-mode signal is a variation due to a 30 s temperature cycle in the dilution fridge. Top-left: a bolometer with a typical time series. Variations due to the 30 s fridge oscillation are obvious. Since all bolometers share this oscillation it is a `common-mode' signal, but the amplitude may vary for different bolometers. Top-right: a bolometer with spikes in the time series. Middle-left: a bolometer with a moderate `step'. Steps can be introduced by the bias for that bolometer `rolling-over' as it reaches its limit. During the S2SRO the instrument software that flags these events was not active. Middle-right: a bolometer with a large step. Bottom-left: a bolometer with multiple steps and spikes. Bottom-right: a `noisy' bolometer.

sc2clean is quite efficient in finding steps, but its spike removal is of limited effectiveness in the presence of a strong common-mode signal. As an example sc2clean was re-run with the following configuration file:⁶

% cat my_sc2clean.def
order = 1
dcfitbox = 30
dcthresh = 25.0
fillgaps = 1
dcsmooth = 50
dclimcorr = 0
flagstat = 0
spikethresh = 5
spikebox = 50
noiseclip = 4.0

% sc2clean sc17_con sc17_concln config=^my_sc2clean.def

The results were that sc2clean left the time series of the top two plots in Fig.  unchanged, i.e. the common-mode variation was too large to cause the spikes in the right time series to be flagged. sc2clean completely flagged the two time series on the bottom row as bad. It did an excellent job correcting for the steps in the middle row time series, as shown in Fig. .

Figure: Original (white) and sc2clean step-corrected (red) time series.