Well no, there is no 'easy' solution to this. It's a common plague of multi-environment systems.
There are partial solutions, but I am still to see a satisfactory one.
For example:
1) Keep different code branches for different database versions.
2) Have an 'adapter' layer on top of your data access layer taking care of (mocking) the differences between your updated data model and the old database implementation.