AO4ELT5 Proceedings
AO error breakdown: anisoplanatism and bandwidth error correlation with ROKET
Ferreira, Florian (LESIA, Observatoire de Paris, PSL Research University, CNRS, Sorbonne Universités, UPMC Univ. Paris 06, Univ. Paris Diderot, Sorbonne Paris Cité), Gendron, Eric (LESIA, Observatoire de Paris, PSL Research University, CNRS, Sorbonne Universités, UPMC Univ. Paris 06, Univ. Paris Diderot, Sorbonne Paris Cité), Rousset, Gérard (LESIA, Observatoire de Paris, PSL Research University, CNRS, Sorbonne Universités, UPMC Univ. Paris 06, Univ. Paris Diderot, Sorbonne Paris Cité), Gratadour, Damien (LESIA, Observatoire de Paris, PSL Research University, CNRS, Sorbonne Universités, UPMC Univ. Paris 06, Univ. Paris Diderot, Sorbonne Paris Cité)
Future Extremely Large Telescope (ELT) adaptive optics (AO) systems will aim at wide field correction and high sky coverage. However, AO correction is only partial and images obtained with such systems can be improved by using post processing techniques for image inversion relying on a field-dependent point spread function (PSF). This requires an accurate knowledge of the PSF and its variations over the science field. The PSF estimation involves identifying the various sources of wave-front error from the AO system telemetry. Estimating and disentangling these error contributors is an issue due to the propagation and filtering process in the AO loop, and numerical simulations are a good way to address it. However, each simulation step at the ELT scale is very demanding in terms of computing power and data flow and the simulation of a long exposure PSF requires several thousands of instantaneous images. Moreover, the error breakdown estimation usually requires to perform several simulations to identify the different contributors. To lead this study, we used COMPASS (COMputing Platform for Adaptive opticS Systems), an end-to-end simulation tool especially developed to reach acceptable simulation time for ELT scales using GPU accelerators. We have developed in COMPASS an estimation tool called ROKET (erROr breaKdown Estimation Tool) that provides a comprehensive error breakdown as the output of a single simulation run. As the outputs of ROKET are the temporal buffers of each error contributor, we are able to study their behaviours and to take into account the possible correlation between them. We have, as well, implemented a GPU accelerated algorithm to reconstruct ELT scale PSF efficiently, so that the impact of the error contributors and their correlation could be seen on the PSF.
DOI: 10.26698/AO4ELT5.0055- Proceeding PDF
