CANARY is the Multi-Object Adaptive Optics (MOAO) pathfinder for the future MOAO-assisted Integral-Field Units (IFU) proposed for Extremely Large Telescopes (ELT). The MOAO concept relies on tomographically reconstructing the turbulence using multiple measurements along different lines of sight.
Tomography requires the knowledge of the statistical turbulence parameters, commonly recovered from the system telemetry using a dedicated profiling technique. For demonstration purposes with the MOAO pathfinder CANARY, this identification is performed thanks to the Learn & Apply (L&A) algorithm, that consists in model-fitting the covariance matrix of WFS measurements dependant on relevant parameters:
Cn2(h) profile, outer scale profile and system mis-registration.
We explore an upgrade of this algorithm, the Learn 3 Steps (L3S) approach, that allows one to dissociate the identification of the altitude layers from the ground in order to mitigate the lack of convergence of the required empirical covariance matrices therefore reducing the required length of data time-series for reaching a given accuracy. For nominal observation conditions, the L3S can reach the same level of tomographic error in using five times less data frames than the L&A approach.
The L3S technique has been applied over a large amount of CANARY data to characterize the turbulence above the William Herschel Telescope (WHT). These data have been acquired the 13th, 15th, 16th, 17th and 18th September 2013 and we find 0.67"/8.9m/3.07m.s
−1 of total seeing/outer scale/wind-speed, with 0.552"/9.2m/2.89m.s
−1 below 1.5 km and 0.263"/10.3m/5.22m.s
−1 between 1.5 and 20 km. We have also determined the high altitude layers above 20 km, missed by the tomographic reconstruction on CANARY , have a median seeing of 0.187" and have occurred 16% of observation time.