Euclid Quick Data Release (Q1). Characteristics and limitations of the spectroscopic measurements

The spectroscopy processing function (SPE PF) of the Euclid pipeline is dedicated to the automatic analysis of 1D spectra to determine redshifts, line fluxes, and spectral classifications. The first Euclid Quick Data Release (Q1) delivers these measurements for all ̋E<22.5 objects identified in the photometric survey. In this paper, we present an overview of the SPE PF algorithm and assess its performance by comparing its results with high-quality spectroscopic redshifts from the Dark Energy Spectroscopic Instrument (DESI) survey in the Euclid Deep Field North. Our findings highlight remarkable accuracy in successful redshift measurements, with a bias of less than $3 $ in $(z_ ̊m SPE -z_ ̊m DESI )/(1+z_ ̊m DESI )$ and a high precision of approximately 10^-3. The majority of spectra have only a single spectral feature or none at all. To avoid spurious detections, whereby noise features are misinterpreted as lines or lines are misidentified, it is therefore essential to apply well-defined criteria on quantities such as the redshift probability, or the flux, and signal-to-noise ratio. Using a well-tuned quality selection, we achieve an 89% redshift success rate in the target redshift range for cosmology ($0.9<z<1.8$), which is well covered by DESI for z<1.6. Outside this range in which the line is observable, redshift measurements are less reliable, except for sources showing specific spectral features (e.g. two bright lines or strong continuum). The classification based on spectroscopy alone is effective for galaxies (about 80% success rate), while it is less efficient for stars and quasars ($<60%$). Ongoing refinements along the entire chain of PFs are expected to enhance both redshift measurements and spectral classification, allowing us to define the large and reliable sample required for cosmological analyses. Taking into account the planned evolution of the spectroscopic pipeline, partially based on the limitations identified in this paper, these results are encouraging for Euclid's future galaxy clustering measurements, even though the requirements are not yet fulfilled.