Dark Energy Survey Deep Field photometric redshift performance and training incompleteness assessment

Toribio San Cipriano, L., De Vicente, J., Sevilla-Noarbe, I., Hartley, W. G., Myles, J., Amon, A., Bernstein, G. M., Choi, A., Eckert, K., Gruendl, R. A., Harrison, I. ORCID: https://orcid.org/0000-0002-4437-0770, Sheldon, E., Yanny, B., Aguena, M., Allam, S. S., Alves, O., Bacon, D., Brooks, D., Campos, A., Carnero Rosell, A., Carretero, J., Castander, F. J., Conselice, C., da Costa, L. N., Pereira, M. E. S., Davis, T. M., Desai, S., Diehl, H. T., Doel, P., Ferrero, I., Frieman, J., García-Bellido, J., Gaztañaga, E., Giannini, G., Hinton, S. R., Hollowood, D. L., Honscheid, K., James, D. J., Kuehn, K., Lee, S., Lidman, C., Marshall, J. L., Mena-Fernández, J., Menanteau, F., Miquel, R., Palmese, A., Pieres, A., Plazas Malagón, A. A., Roodman, A., Sanchez, E., Smith, M., Soares-Santos, M., Suchyta, E., Swanson, M. E. C., Tarle, G., Vincenzi, M., Weaverdyck, N. and Wiseman, P. 2024. Dark Energy Survey Deep Field photometric redshift performance and training incompleteness assessment. Astronomy & Astrophysics 686 , A38. 10.1051/0004-6361/202348956

Preview

PDF - Published Version Available under License Creative Commons Attribution. Download (11MB) | Preview

Abstract

Context. The determination of accurate photometric redshifts (photo-zs) in large imaging galaxy surveys is key for cosmological studies. One of the most common approaches is machine learning techniques. These methods require a spectroscopic or reference sample to train the algorithms. Attention has to be paid to the quality and properties of these samples since they are key factors in the estimation of reliable photo-zs. Aims. The goal of this work is to calculate the photo-zs for the Year 3 (Y3) Dark Energy Survey (DES) Deep Fields catalogue using the Directional Neighborhood Fitting (DNF) machine learning algorithm. Moreover, we want to develop techniques to assess the incompleteness of the training sample and metrics to study how incompleteness affects the quality of photometric redshifts. Finally, we are interested in comparing the performance obtained by DNF on the Y3 DES Deep Fields catalogue with that of the EAzY template fitting approach. Methods. We emulated – at a brighter magnitude – the training incompleteness with a spectroscopic sample whose redshifts are known to have a measurable view of the problem. We used a principal component analysis to graphically assess the incompleteness and relate it with the performance parameters provided by DNF. Finally, we applied the results on the incompleteness to the photo-z computation on the Y3 DES Deep Fields with DNF and estimated its performance. Results. The photo-zs of the galaxies in the DES deep fields were computed with the DNF algorithm and added to the Y3 DES Deep Fields catalogue. We have developed some techniques to evaluate the performance in the absence of “true” redshift and to assess the completeness. We have studied the tradeoff in the training sample between the highest spectroscopic redshift quality versus completeness. We found some advantages in relaxing the highest-quality spectroscopic redshift requirements at fainter magnitudes in favour of completeness. The results achieved by DNF on the Y3 Deep Fields are competitive with the ones provided by EAzY, showing notable stability at high redshifts. It should be noted that the good results obtained by DNF in the estimation of photo-zs in deep field catalogues make DNF suitable for the future Legacy Survey of Space and Time (LSST) and Euclid data, which will have similar depths to the Y3 DES Deep Fields.

Item Type:	Article
Date Type:	Publication
Status:	Published
Schools:	Physics and Astronomy
Publisher:	EDP Sciences
ISSN:	0004-6361
Date of First Compliant Deposit:	11 July 2024
Date of Acceptance:	24 February 2024
Last Modified:	19 Jul 2024 09:31
URI:	https://orca.cardiff.ac.uk/id/eprint/170515

Actions (repository staff only)

Edit Item