Fig. 3 with Fig. 4), especially for unstable stratification. The differencesfor neutral and stable conditions are small. The dependenceof the peak-to-mean ratios from su/u, sv/u and sw/u is quitecomplex (Piringer et al., 2014). Higher su/u and sv/u compared toRobins (1979) increase, a higher sw/u compared to Robins (1979)decreases peak-to-mean ratios determined from ultrasonicanemometer data. As both is the case for the Kittsee data, largerpeak-to-mean ratios result for unstable conditions when using ultrasonicanemometer data compared to the Robins (1979)approach.For the determination of separation distances which comprise aconsiderable range (Figs. 5e10), the largest peak-to-mean ratios inthis range of distances are most important. From a visual inspectionof Fig. 3, neutral conditions are most relevant beyond 100 m. FromFig. 4a, the largest peak-to-mean ratios between 100 and 300 moccur for classes 3 to 5. The class 6 curve is most relevant beyond500m. Fig. 4b considers neutral and stable classes most relevant forhigh peak-to-mean ratios (apart from the factor 4 which is discussedlater). Maximum separation distances depend on the frequencyof the occurrence of these stability classes for each 10 winddirection sector and on the average wind speed. All peak-to-meanratios of the Robins (1979) approach assume the value of 1 before500 m (Fig. 3); this is the reason why this approach, for Kittsee,delivers the smallest separation distances, both with AODM and