Characterization of seasons over the extratropics based on the annual daily mean temperature cycle

A proposal to characterize seasons based on the annual cycle of daily mean temperature over extratropical regions is presented. Four metrics are computed, based on the dates along the year when the maximum and minimum temperature values and the maximum warming and cooling (when the rate of temperatura change is maximum or minimum) are obtained. These four metrics are computed for 1980-2014 period for ERA-Interim, JRA-55 and NCEP2 reanalysis products. Results indicate that these four metrics are able to represent, with geographical coherence over any extratropical region, the start and end of subperiods along the year and that can be seen and used as a consistent procedure to define seasons. The two more extreme metrics (dates for the maximum and minimum temperature) are closely influenced by the variability of the net radiation and near-surface circulation. Regions with different climatic regimes, defined in terms of land-atmosphere coupling conditions, have different contributing factors influencing the interannual variability of the metrics. In humid regions the net surface radiation is critical for the timing of the maximum temperature date, while the variability of the atmospheric circulation is relevant for the minimum temperature metric. In transitional regions, land-atmosphere interaction plays a key role in the timing of the annual cycle, both in summer and winter. Variations in net radiation are relevant for dry regions. Therefore, a robust methodology to establish seasons based on the annual temperature cycle is presented. We also inspect how the metrics timing responds in terms of the interannual variations of different relevant atmospheric variables.