If you actually read the short paper you'll find they actually used reanalysis data sampled at a rate of every three hours across 42 years to compute their statistics:
> Global ERA5 reanalysis data (Hersbach et al., 2020) [...] were extracted on the 197 hPa pressure level with 0.25° horizontal resolution at three hourly intervals from 1 January 1979 to 31 December 2020. To allow the computation of CAT diagnostics that require vertical derivatives, fields on the 188 and 206 hPa levels were also extracted. The 21 turbulence diagnostics were then calculated from the extracted reanalysis fields every three hours.
How you go from that to "they just looked at 2 years" is beyond me.
Newer planes have sensors to measure eddy dissipation rates which are an objective measure of turbulence but I don't know how widespread those systems are and whether they get reported anywhere. They're mostly used for long distance transoceanic flights.
I’m a pilot and it’s been a while since I went over PIREPS but generally severe is rarely used, severe means the turbulence is so bad you can no longer control the aircraft. What most passengers imagine as severe is probably light turbulence. Most of the time it’s not even reported. As a side note if you’re ever on an aircraft and not secured at all times, you’re making a huge mistake.
Another question I have is what do you do in that scenario if you can't control it? Just ride it out and hope for the best?