One of the challenges for users working with MIRI data is the presence of coherent pattern noise resulting from electromagnetic interference (EMI). While a subtle 10 Hz heater noise signal impacts all MIRI data, a more prominent 390 Hz noise impacts short integrations for MIRI subarray data such as that obtained by the Low Resolution Spectrometer (LRS).

In Build 10.1 of the JWST Science Calibration Pipeline (released on February 28, 2024), a new step, "emicorr," was added to the pipeline to correct for the correlated noise by masking the source signal and removing periodic amplitude variations at the known EMI frequencies. This correction is illustrated in Figure 1 below, and significantly improves extracted spectra. This pipeline step is now executed by default for all MIRI FASTR1 and SLOWR1 science exposures, but not for target acquisition images.

While the emicorr step is currently in the pipeline, development work is still ongoing in order to improve its performance. For example, we are working to improve the runtime, which can currently take over a minute for some long exposures. Additionally, the current algorithm does not completely remove the EMI signal for observations with multiples of 8 groups because the 390 Hz noise is in phase with the rate images used for source masking. Therefore, users whose integrations consist of multiples of 8 groups (i.e. 7, 15, 23,… 8*n-1 science groups plus the single group read-reset used by the FASTR1 and SLOWR1 MIRI readout patterns) should subtract this signal as an additional background component in the meantime.

Users seeking more information about the emicorr step, or who have questions about the impact on their observations, are encouraged to reach out to the JWST Help Desk using the MIRI card.