The importance of using a comprehensive suite of tools for the validation of a gyrokinetic code is described. This is detailed by presenting experimental wavenumber spectra which show pronounced differences, although they are measured at the same toroidal, poloidal and radial location. They are obtained via Doppler reflectometry and the differences are due to the probing beam polarization. These differences are reproduced convincingly using turbulence from a gyrokinetic simulation as input for two-dimensional full-wave simulation. It is demonstrated that the application of synthetic diagnostics is indispensable if non-trivial diagnostics are used in the experiment. Furthermore, the measurement of wavenumber spectra via Doppler reflectometry with X-mode probing beam polarization might be problematic due to nonlinear wave-plasma interactions and should be regarded with care when used for quantitative statements or the validation of gyrokinetic codes.

Measurements of turbulent electron temperature fluctuation amplitudes, δT e⊥ /Te, frequency spectra and radial correlation lengths, Lr(T e⊥ ), have been performed at ASDEX Upgrade using a newly upgraded Correlation ECE diagnostic in the range of scales k ⊥ < 1.4cm −1 , kr < 3.5cm(k ⊥ ρs < 0.28 and krρs < 0.7). The phase angle between turbulent temperature and density fluctuations, αnT , has also been measured by using an ECE radiometer coupled to a reflectometer along the same line of sight. These quantities are used simultaneously to constrain a set of ion-scale nonlinear gyrokinetic turbulence simulations of the outer core (ρtor = 0.75) of a low density, electron heated L-mode plasma, performed using the gyrokinetic simulation code, GENE. The ion and electron temperature gradients were scanned within uncertainties. It is found that GK simulations are able to match simultaneously the electron and ion heat flux at this radius within the experimental uncertainties. The simulations were performed based on a reference discharge for which δT e⊥ /Te measurements were available, and Lr(T e⊥ ) and αnT were then predicted using synthetic diagnostics prior to measurements in a repeat discharge. While temperature fluctuation amplitudes are overestimated by > 50% for all simulations within the sensitivity scans performed, good quantitative agreement is found for Lr(T e⊥ ) and αnT . A validation metric is used to quantify the level of agreement of individual simulations with experimental measurements, and the best agreement is found close to the experimental gradient values.