The easiest answer answer is that you can average the daytime and nighttime MODIS observations to get more complete coverage. Mask with at least a QL>=4 (QL 4 and 5 are the best data).
However you must balance that approach with the fact that you may be introducing daytime diurnal SST warming into average if the wind speed is so low that the upper water column is not well mixed but heavily stratified. So if that is a concern its better to just use nighttime skin-SST temperature.
Hope this answer helps.
PO.DAAC Data Engineer