This is how much unit strength is needed in a province to be at 0% chance of uprising as a function of morale.

The approximate relation is:
garrison_needed = 39.7576 - 1.3697 * Morale

Rounding down as in the following fits all the data points exactly:
garrison_needed = floor(39.75 - 1.35 * Morale)

The rounding function including jubs datapoint. There is no guarantee this will work for all other data points, but it works for all these points. I'm not sure if the garrison needed value is actually rounded when determining uprise chance. For example, if it says you need 6 and put a unit with exactly 6, but the internal value is really 6.4, it is conceivable that it could still have a chance to uprise. To be super safe you could shoot for displayed value + 1.

So I'm playing a game of Iron Order 1919 (another Bytro game) and came across a situation that implies that the garrison strength values are only rounded in the display, but not internally. I had a garrison troop strength of 2.2 and the displayed garrison needed was 2, but the uprise chance was still 1%. Thus, internally the garrison needed was probably like 2.3, e.g. The first equation posted by the OP (without rounding) is probably the closest to the internal value.