Ah I realize my first comment is maybe not written so well. I meant to convey, it looks to me that some irrationally high dummy testing values (put there because we're seeing historically high winds and they wish to test that behaviour) were accidentally put into the production data source.