Euclid photo-z data challenges
The data challenges are an on-going effort lead by the photo-z Organisation Unit (OU-PHZ) in the Euclid consortium. This page presents the photo-z data challenges, gather useful information and is the entry point for downloading the data.
The data challenges are the main activity of the work package "testing" (a team of ~30 people), whose overall goal of is to set up a framework to test photo-z algorithms using Euclid-like catalogues (simulated or based on real data, e.g. CANDELS data prepared by OU-MER), distributed to people to run their photo-z codes. We also develop quality control tools to compare the different algorithms and propose photo-z requirement updates for cosmic shear when necessary.
it is strongly encouraged to join the work package mailing list to get the most updated information about the challenges.
Your ESA "cosmos" login is required to access protected material (see detailed instructions here). The main repository is located in https://obswww.unige.ch/data/euclid-testing/testing/.
For any question on these pages, you may contact This email address is being protected from spambots. You need JavaScript enabled to view it..
Goal of the challenges
The photo-z data challenges aim at measuring the performance of photo-z codes to be used for the cosmic shear analysis of the Euclid mission. The data are prepared under the lead of the "testing" work package with people from OU-EXT, OU-MER and OU-PHZ teams.
Data challenges
-
DC1: The Data Challenge 1 is the first of a planned series of data analysis efforts which are coordinated among the various Euclid OU's.
-
DC2: The main goal of the Data Challenge 2 is to measure the performance of photo-z codes.
-
DC3: The goal of the Data Challenge 3 is to focus on the method and the calibration set, using simulations.
Data Challenge 1
The Data Challenge N.1 (DC1) is the first of a planned series of data analysis efforts which are coordinated among the various Euclid OU's, and have the goal of testing various steps of the Science Ground Segment data processing. In the first exercise, OU-MER will release a set of data (images and catalogs) that simulates data obtained by Euclid. The simulated data has been obtained by using CANDELS data from 8 different filters to mimic Euclid observations in VIS and NIR and four ground based filter bands. Objects from these images will be extracted with standard processing tools to obtain multi-wavelength catalogs that can be used to test photometric redshifts tools.
Documentation
The DC1 documentation is located on the Euclid "REDMINE" wiki page: http://euclid.roe.ac.uk/projects/sgw/wiki/Data_Challenge
Data
As for the documentation, the data (catalogues and images) are located on the "REDMINE" wiki page: http://euclid.roe.ac.uk/projects/sgw/wiki/Data_Challenge
Results
The final report can be found here.
Data Challenge 2
The main goal of this challenge is to measure the performance of photo-z codes using a realistic set of real images in the COSMOS field, including DECam images at the DES depths, a VIS-like image emulated from HST i-band data (the detection image), and deep near-IR images from the ultraVISTA survey at similar depths as those defined in the Euclid red book (Laureijs et al., 2011).
Documentation
The README file contains a detailed description of how the catalogues were made, how to run your photo-z code and return your results.
Data
All you need to train and run you photo-z code (reference/photometric catalogues, filter curves, mask files, readme file, etc.) is contained in a single archive (the link points towards the latest version, for older versions, see https://obswww.unige.ch/data/euclid-testing/testing/DC2 and for a version log, see https://obswww.unige.ch/data/euclid-testing/testing/DC2/version_log.ascii).
Images may be requested to OU-MER and OU-EXT groups.
Scripts used to analyse the data by WP leader are given "as is" for information.
Results
Individual results are stored in https://obswww.unige.ch/data/euclid-testing/testing/DC2/results:
- CATALOGUES: photo-z catalogues with all entries produced by each participant
- ALL: performance checks in Euclid metric (in PDF plots) for each participant, in the case where no filter is applied ("all" objects, see documentation for more details)
- USE: performance checks in Euclid metric (in PDF plots) for each participant, in the case where a custom filter is applied ("all" objects, see documentation for more details)
Global summaries and analysis can be found in form of presentations given at OU-PHZ meetings.
Data Challenge 3
The goal of this third challenge is to focus on the method and the calibration set, using simulations. Since the DC2 has the advantage of providing a catalogue made from real images and allows us to directly link the photo-z group efforts to the data reduction and photometry (OU-EXT and OU-MER) , we will leave it open for the time we use simulations.
The catalogues include all the information (photometry, position and true redshift) and were produced from best-fit SEDs in the COSMOS field, from which we emulated fluxes and flux errors at the required depths as reported from Jean-Charles Cuillandre last year. The training data set has 5 times higher SNR.
Catalogues, filter transmission and documentation are here: https://obswww.unige.ch/data/euclid-testing/testing/DC3/
The login and password to access the page can be found in https://wiki.cosmos.esa.int/