The Penn State NCAR mesoscale model MM5, forced by the NCEP Eta model, was run at 6-kilometer resolution for nine years in the Southern California region. The simulation exhibits significant spatial structure in both forced and unforced surface air temperature (SAT) variability, and the goal of this study is to understand the reasons for this.

The SAT seasonal cycle amplitude is primarily determined by distance from the coast, and exhibits no elevation dependence. In contrast, the diurnal cycle is strongly affected by elevation, with the highest elevations having a diurnal amplitude up to three times smaller than low-lying areas. This effect occurs year round, but the magnitude is largest during summer months, when the stability in this region is at its peak. The magnitude of unforced day-to-day variability is similarly structured, with reduced variability at high elevations in the summer months. The winter day-to-day variability exhibits no such dependence. We also discuss the physical mechanisms behind these spatial structures.