You can check the alarm definition on the vCenter level... if the trigger is really being activated (i.e, you have it configured to trigger an alert with more than 75% of usage) and the space is still being used, you will need to deactivate the alarm or change the trigger to a higher % (if you do this on vCenter level, it will propagate to all datastores).
If you are not using anymore the space and the alarm *should not* be there, you can just edit the alarm and press OK. Usually, this works for me.