alternatives. The reasons have to do with the strong diurnal cycle in
these data sets.
While meteorological models have often been preferred for
atmospheric forecasting, prior studies have found that a combination
of physics-based and statistical models is preferable to either
individually (Glahn and Lowry, 1972; Bacher et al., 2009; Reikard
et al., 2011). The next stage in this research is therefore a comparison
of time series and mesoscale models, and combinations of the
two approaches.