Data processing techniques for density profile evaluation from FM endash CW reflectometry

Microwave reflectometry is an important diagnostic to measure the plasma density profile on present fusion experiments and on next devices like ITER. Several data processing techniques, currently employed to evaluate the density profile, use filtering algorithms to overcome the effect of plasma fluctuations. However, in these processes the detailed profile structure may be altered or even lost. In order to study this problem we applied two different data analysis methods to experimental results obtained in ASDEX, when a locked magnetohydrodynamic mode is present. The first method uses a software implementation of a frequency discriminator to estimate continuously the evolution of the group delay versus frequency and the second applies a sliding fast Fourier transform method to determine an averaged group delay curve. It is shown that the profile exhibits a density plateau due to the magnetic island, which can be lost due to filtering or averaging procedures. Next we perform a numerical study for a profile with a small rotating density plateau, disturbed by random noise to simulate the effect of plasma fluctuations. The frequency discriminator is used to obtain group delay sweeps from consecutive sweeps and a two-dimensional (2-D) regularization technique is applied to the multiple samples. It is shown that the 2-D method has advantages in the routine evaluation of the detailed profile structure masked under the effect of plasma turbulence, due to the combined averaging (over space and time), based on the correlation between neighboring data points. copyright 1997 American Institute of Physics