Speaker
Description
The Euclid mission is the European Space Agency's (ESA) cosmology mission designed to map approximately 1.5 billion galaxies over 14,000 deg² to investigate the nature of dark energy and dark matter through weak gravitational lensing and galaxy clustering, combining high‑resolution optical imaging (VIS) with near‑infrared imaging and spectroscopy (NISP).Image persistence in large-format HgCdTe H2RG detectors is a major instrumental systematic for the NISP, affecting redshift measurements, low-surface-brightness observations, and ultimately the accuracy of cosmological parameter estimation. The current Euclid persistence correction employs an empirical, pixel-dependent power-law model calibrated on ground data. While this model reproduces the overall nonlinear dependence of persistence on stimulus fluence and decay time, it still leaves residual biases of several electrons and realization-dependent spatial fluctuations, limiting its performance for short-term persistence correction. We present a new pixel-level persistence model based on Euclid's in-flight monthly calibration data. The model is constructed from sequences of 30 kADU LED-E flat-field exposures followed by dark observations acquired in both photometric and spectroscopic modes. For each pixel, the persistence signal is modeled as a time-integrated cubic B-spline representation of the decay rate. Integrating the spline basis over the actual exposure intervals naturally accounts for Euclid's mixed observing cadence while producing monthly maps of spline coefficients. To improve robustness, coefficient maps from eight independent calibration realizations per detector are combined using Principal Component Analysis (PCA), which isolates the dominant spatial component and suppresses realization-dependent noise. This model is applied to Wide Survey slew dark observations by integrating persistence contributions from the complete exposure history. Preliminary results demonstrate reduced systematic bias and small-scale spatial noise compared with the current model while preserving the physical behavior of the long-term persistence tail, providing a more accurate and physically motivated framework for persistence correction in Euclid NISP data.