Integral scale is an important variable in the theory of turbulence and in numerical weather prediction models. It is a good indicator of the size of the eddies that dominate the turbulence spectrum. Integral scales are estimated from the atmospheric wind speed measurements using autocorrelation functions (ACF) and Fourier spectrum. In theory, their ratio should be a constant. However, the values of integral scales obtained from ACF and Fourier spectrum are very sensitive to data filtering, which is in practice a first step in the analysis of the atmospheric turbulence in general. The measurements that were used were performed on a micrometeorological tower installed near the Maslenica Bridge north of the city of Zadar with a sampling frequency of 20 Hz. For the period from October 9, 2015 to October 9, 2016, 48 events of bora flow were abstracted. Bora is a strong, gusty wind, whose microscale characteristics are not fully explored yet. The main objectives of this work are to examine the effect of data filtering on integral scale values for bora flows, obtained from ACF and from Fourier spectrum, the effect on their ratio and to try to find out a suitable high-pass filter period. The results show that integral scales obtained by the filtered data using filter periods close to averaging time scale (30 min) are not significantly sensitive to filtering. When using shorter filter periods, it is better to estimate integral scales from Fourier spectrum. Given that the ratio is relatively conserved over the length of the filter, it is difficult to say whether the raw data should be filtered or not.