We elaborate on the preliminary results presented in Beyrich et al. (in Boundary-Layer Meteorol 144:83–112, 2012), who compared the structure parameter of temperature (C2T) obtained with the unmanned meteorological mini aerial vehicle ( M 2 AV ) versus C2T obtained with two large-aperture scintillometers (LASs) for a limited dataset from one single experiment (LITFASS-2009). They found that C2T obtained from the M 2 AV data is significantly larger than that obtained from the LAS data. We investigate if similar differences can be found for the flights on the other six days during LITFASS-2009 and LITFASS-2010, and whether these differences can be reduced or explained through a more elaborate processing of both the LAS data and the M 2 AV data. This processing includes different corrections and measures to reduce the differences between the spatial and temporal averaging of the datasets. We conclude that the differences reported in Beyrich et al. can be found for other days as well. For the LAS-derived values the additional processing steps that have the largest effect are the saturation correction and the humidity correction. For the M 2 AV -derived values the most important step is the application of the scintillometer path-weighting function. Using the true air speed of the M 2 AV to convert from a temporal to a spatial structure function rather than the ground speed (as in Beyrich et al.) does not change the mean discrepancy, but it does affect C2T values for individual flights. To investigate whether C2T derived from the M 2 AV data depends on the fact that the underlying temperature dataset combines spatial and temporal sampling, we used large-eddy simulation data to analyze C2T from virtual flights with different mean ground speeds. This analysis shows that C2T does only slightly depends on the true air speed when averaged over many flights.