Dear Olivia Shuk-ming Lee,
there is indeed an issue with the way the number of expected observations are calculated for Hong Kong International Airport. It is a combination of the particular way you introduced the schedules plus our algorithm doing the calculations.
As you do 7/24 3h observations, the "normal" way to introduce them would be 00:00->23:59 with a 3h interval (possibly 20h diurnal basetime). However, 20:00->19:59 with 3h can also be used, but presents the algorithm with some challenges in some cases, as can be seen.
We will work on a fix. In the meantime you could update (correct) in OSCAR/Surface the schedule definitions to the recommended format, unless there is a reason to use the 20:00->19:59 "style".
I am actually curious why you choose to represent the schedules in this way. We are aware of the lack of guidance on the schedules and will elaborate guidelines and for this reason it is important to understand how different users interpret the current schedules.
with very best regards and apologies for the complications,