Euclid: Estimation of the impact of correlated readout noise for flux measurements with the Euclid NISP instrument

TL;DR

This paper models and characterizes correlated readout noise in Euclid's NISP detectors, deriving an analytical covariance expression that improves flux measurement accuracy and reduces bias in in-flight data analysis.

Contribution

It provides an analytical model for correlated readout noise in NISP detectors and demonstrates its effectiveness in reducing flux measurement bias.

Findings

Correlated noise significantly impacts flux measurements.

The analytical covariance model reduces bias by up to four times.

Bias remains negligible even with white noise approximation.

Abstract

The Euclid satellite, to be launched by ESA in 2022, will be a major instrument for cosmology for the next decades. \Euclid\ is composed of two instruments: the Visible (VIS) instrument and the Near Infrared Spectromete and Photometer (NISP). In this work we estimate the implications of correlated readout noise in the NISP detectors for the final in-flight flux measurements. Considering the multiple accumulated (MACC) readout mode, for which the UTR (Up The Ramp) exposure frames are averaged in groups, we derive an analytical expression for the noise covariance matrix between groups in the presence of correlated noise. We also characterize the correlated readout noise properties in the NISP engineering grade detectors using long dark integrations. For this purpose, we assume a $(1/f)^{\, \alpha}$-like noise model and fit the model parameters to the data, obtaining typical values of $\sigma = 19.7^{+1.1}_{-0.8}$ e$^{-} \rm{Hz}^{-0.5}$, $f_{\rm{knee}} = (5.2^{+1.8}_{-1.3}) \times 10^{-3} \, \rm{Hz}$ and $\alpha = 1.24 ^{+0.26}_{-0.21}$. Furthermore, via realistic simulations and using a maximum likelihood flux estimator we derive the bias between the input flux and the recovered one. We find that using our analytical expression for the covariance matrix of the correlated readout noise we diminish this bias by up to a factor of four with respect to the white noise approximation for the covariance matrix. Finally, we conclude that the final bias on the in-flight NISP flux measurements should still be negligible even in the white noise approximation, which is taken as a baseline for the Euclid\on-board processing