Low Order Adaptive Optics with Very Faint Reference Stars

TL;DR

This paper proposes a new adaptive optics method using faint reference stars and Lucky Imaging with photon-counting detectors to achieve near-diffraction-limited ground-based astronomical imaging, surpassing Hubble's resolution.

Contribution

It introduces a novel curvature wavefront sensor and software techniques adapted from medical imaging to enable high-resolution imaging with very faint reference objects on large telescopes.

Findings

Achieves near-diffraction-limited performance on large telescopes.

Uses reference stars fainter than I~17.5 mag routinely.

Provides a cost-effective alternative to space telescopes.

Abstract

It is widely believed that adaptive optics only has a role in correcting turbulent wavefronts on large telescopes using very bright reference stars. Unfortunately these are very scarce and many astronomical targets require wavefront correction to work over much of the sky. We therefore need to be able to use very much fainter reference objects. Laser guide stars in principle can allow 0.1 arcsecond resolution but have a number of severe technical problems that limit their application. Our aims are to provide imaging at even higher resolution than Hubble. Lucky Imaging completely eliminates the tip-tilt errors in astronomical wavefront detection. Most of the power that remains is in low order, large scale structures. These may be detected with high sensitivity using photon-counting EMCCD detectors working at high frame rate, up to ~100Hz. With a new design of curvature wavefront sensor, wavefront errors may be measured and corrected to give near diffraction-limited performance on large groundbased telescopes in the visible. Reference stars (and reference compact galaxies) fainter than I~17.5 mag may be used routinely. This paper will describe how these work, what detector and other hardware is needed and what software should be used to measure the wavefront errors and drive deformable mirror hardware. The software techniques that are used are those routinely applied for MRI and CT imaging. They are fast and relatively easy to implement. The net effect is that imaging systems can be constructed that improve substantially over Hubble resolution from the ground for a relatively modest sum of money.