This correction is currently only implemented for MIRI data and is only for integrations after the first integration (i.e. this step does not correct the first integration). It is assumed this step occurs before the dark subtraction, but after linearity.
The MIRI Focal Plane System (FPS) consists of the detectors and the electronics to control them. There are a number of non-ideal detector and readout effects which produce reset offsets, nonlinearities at the start of an integration, non-linear ramps with increasing signal, latent images and drifts in the slopes. The manner in which the MIRI readout electronics operate have been shown to be the source of the reset offsets and nonlinearities at the start of the integration. The reset offset, also described as the zero-point offset, are easily seen in multiple integration data, where the first integration starts at a lower DN level than subsequent integrations. The ampitude of this offset is proporational to the singnal level in the previous integration. Fortunately this offset is constant for all the groups in the integration, thus has no impact on the slopes determined for each integration.
The readout electronics have also been shown to be the source of the nonlinearities at the start of the integration. Basically the MIRI reset electronis use field effect transisitors (FETs) in their operation. The FET acts as switch to allow charge to build up and to also initialize (clear) the charge. However, the reset FETS do not instanteously reset the level, instead the expontenial adjustment of the FET after a reset causes the initial frames in an integration to be offset from their expected values. The Reset Switch Charge Decay (RSCD) step corrects for the slow adjustment of the FET output to its asymptotic level after a reset. This correction is made for integrations > 1 and is based on the signal level in the last frame of the previous integration in the exposure. Between exposures the MIRI detectors are conintually reset; however for a multiple integration exposure there is a single reset between integrations. The reset switch charge decay has an e-folding time scale ~ 1.7 * frame time so the affects of this decay are not measureable in the first integration because a number of resets have occurred from the last exposure and the effect has decayed away by the time it takes to readout out the last exposure, set up the next exposure and begin exposuring. There are low level reset effects in the first integration that are related to the strength of the dark current and can be removed with an integration dependent dark.
For MIRI multiple integration data, the reset switch decay causes the the initial groups in integrations after the first one to be offset from their expected linear accumulation of signal. The most significant deviations ocurr in groups 1 and 2. The amplitude of the different between the expected value and the measured value varies for even and odd rows and is related to the signal in the last frame of the last integration.
The rscd correction step applies the reset switch charge decay reference file. Based on READOUT pattern (FAST or SLOW) and Subarray type (FULL or one of MIRI defined subarray types) the reference file contains the scale factor and decay time (tau) for even and odd rows to corrected for the reset effects. The accumulated DN of the pixel from the previous integration is estimated from the linear fit of the ramp. For each pixel the group values are corrected according the formula:
corrected value = input vaule + dn_accumulated * scale * exp(-T / tau),
where T is the time since the last frame in the last integration.
Currently the rscd correction for subarray data is the same as it is for full array data. However, we anticipate a seperate set of correction coefficients in the future.