The ROWDROOP module is designed to remove an apparent excess of photocharge generated only along the fast clock direction (detector rows). The effect manifests itself as an additional output signal in a given pixel, which increases in proportion to the total signal from all pixels in the same row. As opposed to the MUX-bleed effect seen in the IRAC detectors, there is no evidence that the signal produced by IRS row droop depends on distance from the source of the signal. In other words, the row droop is constant in the fast-clock direction. The strongest source of row droop in astronomical data occurs in the SL module when a bright source is situated on the peak-up array. This situation can induce measurable row droop across the SL spectral orders. In-orbit data have shown that row droop is negligible in the other instrument channels (SH, LL, and LH), so the pipelines do not correct for it in these channels.
The row droop correction consists of subtracting from each sample the product of the row droop coupling constant and the total row photo-charge:
where Dmi is the observed value for the ith pixel of a row at sample k, and c is the row droop coupling constant.
The total row photo-charge must be measured on data that are not yet corrected for droop, or else we empirically find the row droop correction to be underestimated. This estimate should include any A/D saturation-corrected samples in the row, in analogy to DROOPOP for the entire array. The inputs for ROWDROOP are satcor.fits and droopop.fits. The total row photo-charge is estimated from satcor.fits, and the resulting row droop correction is subtracted from droopop.fits.
The coupling constant for row droop has been determined using in-orbit data to be c =(4.5±0.68)×10-5 for the SL array. It is independent of the coupling constant for array droop. There is no evidence for a dependency on read-out channel number.
The outputs of ROWDROOP are rowdroop.fits and rowdroop_unc.fits. In rowdroop.fits, the saturation-corrected samples are replaced with the arbitrary value 300,000 electrons, for the same reason as in DROOPOP. The output files are not available in the Spitzer Heritage Archive, but can be generated by running CUPID.