Attenuating the Absorption Contribution on \({C_{n^{2}}}\) Estimates with a Large-Aperture Scintillometer

Large-aperture scintillometers (LAS) are often used to characterize atmospheric turbulence by measuring the structure parameter of the refractive index \({C_{n^{2}}}\) . However, absorption phenomena can lead to an overestimation of \({C_{n^{2}}}\) . By applying an accurate numerical filtering technique called the Gabor transform to the signal output of a LAS, we improved our knowledge of the accuracy of the measured \({C_{n^{2}}}\) by determining and attenuating the contribution of absorption. Two studies are presented on a 12-day dataset using either fixed band pass or adaptive filtering. The first consists of evaluating the best-fit filter for which the resulting \({C_{n^{2}}}\) is independent of meteorological conditions, especially crosswind, and the second consists in accurately reconstructing the signal to remove absorption, without losing information on \({C_{n^{2}}}\) . A reference \({C_{n^{2}}}\) (hereafter ‘reconstructed \({C_{n^{2}}}\) ’) is created by accurately removing absorption from the scintillation spectrum, and is used to evaluate each filter. By comparing the ‘reconstructed \({C_{n^{2}}}\) ’ with a raw \({C_{n^{2}}}\) measured with a scintillometer, in the presence of absorption, we found that the average relative contribution of absorption to the measurement of \({C_{n^{2}}}\) is approximately 9%. However, the absorption phenomenon is highly variable; occasionally, in the worst cases, we estimate that the absorption phenomenon could represent 81% of the value of \({C_{n^{2}}}\) . Some explanations for this high variability are proposed with respect to theoretical considerations. Amongst the fixed band-pass filtering used, we conclude on the preferential use of a band-pass filter [0.2–400 Hz] for \({C_{n^{2}}}\) , as its performance is only slightly affected by the crosswind, and that the mean absorption contribution is reduced to 5.6%, with a maximum value of 60%. Using an adaptive filter on the 12-day dataset really improves the filtering accuracy for both points discussed—the independence of meteorological conditions and the quality of signal reconstruction.