Calibration of Wide Field Imagers - The SkyDICE Project

Between 1999 and 2001, three measurements changed Cosmology forever: the discovery of Cosmic Acceleration (Riess et al. 1998, Perlmutter et al. 1999) indicated that the density of the Universe is dominated by some kind of repulsive energy of unknown nature (namely Dark Energy). The measurement of the first acoustic peak in the CMB temperature anisotropy spectrum (de Bernardis et al. 2000) combined with the precise determination of H0 (Freedman et al. 2001) gave strong constraints on the flatness of space-time. These measurements contributed to solve the persisting disagreements between the observations that were favouring a low density of matter, and theoretical motivations for a higher-density (critical) Universe. It favoured the emergence of the Standard Model of Cosmology ( CDM) that describes nearly all of today's observations with only a handful of free parameters (Planck Collaboration et al. 2013b). Cosmology has now entered an era of precision measurements, and the goal of observations is now to hunt for "tensions" within the cosmological model. The case of Supernova cosmology is very characteristic of this situation. The measurement of luminosity distances to SNe-Ia as a function of their redshift allowed one to discover (with less than 100 supernovae) the acceleration of cosmic expansion. Today, SNe-Ia are still the most sensitive probe to w, the Dark Energy equation of state parameter, and growing number of SNe-Ia are being detected and studied by several Collaborations all over the world, in order to pin down the value of w, and to start ruling out Dark Energy models. The precision on w is now as low as 7% (Conley et al. 2011, Sullivan et al. 2011) with nearly 1000 SNe-Ia in the Hubble diagram. Unfortunately, the measurement is now dominated by systematic uncertainties, the dominant source of systematics being the photometric calibration of the imagers used to measure the SNe-Ia fluxes. This work is about photometric calibration. This is a rather esoteric subject, which is seldom chosen by PhD students. But the thing is that, to improve on the current results, astronomers have no choice but to revisit the ancient calibration schemes. Since 2005, most Dark Energy Collaborations (with the invaluable help of the HST calibration program) have launched ambitious calibration efforts, redefined primary standards and metrology between those standards and their science images and push down their error budget well below 1% (e.g. Betoule et al. 2012). One suspect however, that these techniques, which rely on observations of stellar calibrators, will not allow one to reach the calibration requirements of future surveys. For this reason, several groups in the world are working on experimental laboratory sources, that would allow one to inject very well characterised light into the telescope optics and derive, from these measurements, the telescope throughput as a function of wavelength. Since 2007, LPNHE cosmology group has been involved in the construction of a spectro-photometric calibration system for the last generation of wide field imagers (Barrelet and Juramy 2008). In particular, the team has designed and built two devices: SnDICE (Supernovae Direct Illumination Calibration Experiment) and SkyDICE (SkyMapper Direct Illumination Calibration Experiment), the first installed in the enclosure of the Canada France Hawaii Telescope (CFHT) on top of Mauna Kea, and the other in the dome of SkyMapper (Siding Springs Observatory, NSW, Australia). I started my PhD a few months after the project was funded. I was involved in nearly all stages of the project, in particular the integration and the calibration of the device on our test bench, as well as the installation and commissioning at Siding Springs. I then spent my third year analysing the commissioning data. We have shown that it is possible to build a LED based light source that samples evenly the full visible wavelength range. The stability of the source is remarkable, ranging from a few 10−4 for a few of the LEDs, to 10−3 for the less stable channels. I have detailed the spectrophotometric characterisation of the device on our test bench at LPNHE. More importantly, I have shown that it is possible to build a smooth spectrophotometric model of each LED, that can predict the LED spectrum at any temperature (in a temperature range representative of what is measured in the telescope enclosure). Each of these models comes with an uncertainty budget that accounts for (1)-the finite number of spectroscopic and photometric measurements and (2)-the test bench uncertainties. Finally, I have described a method to calibrate the effective passbands of the imager, and monitor their fronts from series of calibration frames taken with SkyDICE. This method takes into account all the test bench uncertainties are propagate them as exactly as possible to the final result. It is currently being applied to the real SkyDICE dataset, and what has been presented here is a set of tests performed on (realistic) simulated datasets. A important result of this work is that, despite the fact that the LEDs are not monochromatic sources, we are able to control the position of the filter fronts with an accuracy well below 1-nm. Regarding the passband inter-calibration, we have computed the expected uncertainties affecting our estimates of the passband normalisation, relative to the r-band. These uncertainties actually depend on how we interpret the uncertainties that affect the calibration of the NIST photodiode. In the best-case scenario, where the NIST uncertainties are all positively correlated, we have shown that after a few calibration runs, we get down to a precision of 0.4% in the u and v-bands (near-UV) and of 0.3% in the other bands. Depending on how we estimate the CALSPEC uncertainties (which are themselves uncertain), this result is either a major improvement on CALSPEC, or on par with what can be obtained with CALSPEC. In any case, this means that by using routinely a DICE source to calibrate a survey telescope, we should be able to test the CALSPEC flux scale. The analysis of the SkyDICE commissioning dataset is still ongoing. The main missing ingredient is the control of the relative positions of the telescope and the sources, as well as an estimate of the pollution of the calibration frames. These two aspects of the analysis are actively worked on, and the first constrains should be published soon.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.