ausblenden:
Schlagwörter:
Climate change; Atmospheric general circulation; Climate; Hadley circulation; Jet stream; Large single-model ensemble; Phase-space consideration of climate-pdf arising solely from internal variability
Zusammenfassung:
Based on a single-model 100-member ensemble of climate change simulations performed with the MPI-ESM, this paper investigates how the probability density function (PDF) of atmospheric general circulation (AGC)-a marginal distribution of the full climate PDF-responds to increasing CO2 concentration. The investigation provides a first assessment of CO2-induced changes in AGC-PDF derived from ensemble statistics (as oppose to time statistics in usual practice) that arise solely from internal variability consistent with the imposed CO2-forcing. By focusing on the first two moments of the AGC-PDF related to the Hadley circulation and the subtropical jets, we find that the major CO2-induced changes in the PDF are related to structure changes in the mean circulation and in the circulation variability. As CO2 concentration increases, different components of the AGC can co-evolve with changing forcing. In particular, the mean Hadley cell co-evolves with the mean subtropical jets. The co-evolution is characterized by the concurrent Hadley cell widening and poleward shifting of the subtropical jets (particularly strong in January), by a strengthening of both the Hadley cell and the subtropical jets in October, and by a weakening of the Hadley cell and a strengthening of the northern and southern jets in January. The changes in circulation variability are more subtle, and reveal only large and spatially coherent changes in places, where the variability is already strong without increasing CO2 concentration, such as at the northern and southern flanks of the summer southern subtropical jet, and at the northern flank of the southern polar night jet. There are some significant cross-variabilities between different components of the AGC. Overall, the CO2-induced AGC-changes can be described by linear trends in the first and second moments, with the trends in the means being much stronger than those in the variances and co-variances
