Including volcanic forcing in the MiKlip prediction system has increased the prediction skill for global surface temperature (Timmreck et al., 2016). However, the regional impacts of volcanic forcing, in particular on the Northern Hemisphere (NH) winter climate, which are controlled largely by dynamical changes, are neither fully understood nor well represented in the current prediction system. ALARM-II will therefore explore and improve the representation of the climate response to aerosol perturbations caused by volcanic eruptions in the MiKlip prediction system, in order to prepare for future events and be able to forecast their effects immediately after a potential future eruption.
ALARM II consists of three workpackages and is coordinated by Dr. Claudia Timmreck (MPI-M). Dr. Hauke Schmidt (MPI-M) and Prof. Dr. Kirstin Krüger (UiO), as an external partner, contribute to the project.
A multilinear regression analysis of five historical runs with the MPI-ESM-HR and CMIP5 forcing shows that the MPI-ESM-HR model reacts qualitatively similar, in the middle atmosphere to natural and anthropogenic forcings, to the MPI-ESM-LR and the MPI-ESM-MR model (see Schmidt et al., 2013). The temperature response to volcanic eruptions shows the typical pattern of a tropospheric cooling and a warming in the low to mid-latitude lower stratosphere. The meridional temperature signal from volcanic forcing is reflected in the westerly wind anomalies in large parts of the stratospheric high-latitudes. It is, however, difficult to interpret the difference between the model versions, as the volcanic signal is masked by the high interannual variability in NH winter. The large ensemble (≥25) of the planned VolMIP (Model Intercomparison Project on the climatic response to Volcanic forcing) Pinatubo simulations (Zanchettin et al., 2016) with the MPI-ESM in LR (low resolution) and HR (high resolution) will offer a great possibility to study the effect of an increased model resolution with an ensemble sufficiently large to get a significant response.
The MPI Grand Ensemble, a 100-member ensemble of historical (1850–2005) simulations with the MPI-ESM-LR, has been analysed with focus on the stratospheric temperature and wind response in the 1st post volcanic NH winter (Bittner et al., 2016). Approximately 15 ensemble members are needed to get a significant (95%) response of the NH polar vortex in December to February (DJF) after the Pinatubo eruption. The amount of necessary ensemble members for a significant response depends on the magnitude of the anomaly and the interannual variability. Including smaller eruptions to increase the sample size does not necessarily improve the detectability of the volcanic signal. Analysing the dynamical response to volcanic eruptions in too small ensembles might therefore lead to false conclusions. Hence, the CMIP5 models do not generally fail to capture the dynamical response to tropical volcanic eruptions (Figure 1). Large uncertainties remain in the response of the real atmosphere to volcanic eruptions due to the small number of observed events.
To address the question how does the pre-eruption climate state influence the impact of the volcanic signal on the prediction, we have performed together with Module D FLEXFORDEC/INTEGRATION decadal forecasts with the MiKlip prediction system, for the initialisation years 2012 and 2014, which differ in the Pacific Decadal Oscillation (PDO) and North Atlantic Oscillation (NAO) phase llling et al (2018). Each forecast contains an artificial Pinatubo-like eruption starting in June of the first prediction year and consists of 10 ensemble members. Our results show that the average global cooling response over 4 years of about 0.2 K and the precipitation decrease of about 0.025 mm day−1 is relatively robust throughout the different experiments and seemingly independent of the initialisation state. However, on a regional scale, we find substantial differences between the initialisations. The cooling effect in the North Atlantic and Europe lasts longer and the Arctic sea ice increase is stronger in the simulations initialised in 2014. In contrast, the forecast initialised in 2012 with a negative PDO shows a prolonged cooling in the North Pacific basin.
Schurer, A.P. | Hegerl, G.C., Luterbacher, J., Brönnimann, S., Cowan, T., Tett, S.F.B., Zanchettin, D., Timmreck, C.
Niemeier, U. | Timmreck, C., Krüger, K.
Toohey, M. | Krüger, K., Schmidt, H., Timmreck, C., Sigl, M., Stoffel, M., Wilson, R.
Zanchettin, D. | Timmreck, C., Toohey, M., Jungclaus, J., Bittner, M., Lorenz, S., Rubino, A.
Claudia Timmreck | Graham W. Mann, Valentina Aquila, Rene Hommel, Lindsay A. Lee, Anja Schmidt, Christoph Brühl, Simon Carn, Mian Chin, Sandip S. Dhomse, Thomas Diehl, Jason M. English, Michael J. Mills, Ryan Neely, Jianxiong Sheng, Matthew Toohey, and Debra Weisenstein