To investigate whether the results were sensitive to the
levels of control for seasonal patterns, we changed the df (3e5per year) for time to control for season. We also varied the df
(3e6) for mean temperature and relative humidity. We found
the results changed little.