Nothing you do is going to be very efficient, because the best you can hope for is an index scan on jDate, if there is an index on that column.
This should be faster than the LIKE, because the datediff and dateadd functions are much faster than conversions to strings. It works by finding the hour of the month for jDate and comparing it to the hour of the month for the month/day combination you are looking for:
You could also compute these numbers this way:
As with anyting, you should test to see what actually produces the best results.
CODO ERGO SUM