Torge Mar)n 1,*,+ and Thomas W. N. Haine 2,+
1
GEOMAR Helmholtz Centre for Ocean Research Kiel, Kiel, Germany
2Earth & Planetary Sciences, The Johns Hopkins University, Bal)more, MD USA
*corresponding author: torge.mar)n@gmail.com
+these authors contributed equally to this work
The heat is on
What is Arc/c Amplifica/on?
References Conclusions
SST anomaly on Oct 25, 2016, NASA OISST V2
source: NSIDC Arc/c Sea Ice News Nov. 2016 / climatereanalyzer.org
The CMIP5 model ensemble* projects
autumn mean Arc+c Amplifica+on to peak in the 2010s
Figure 5: Time series of the autumn mean surface Arc)c Amplifica)on factor defined as the ra)o of Arc)c to global SAT anomalies as shown in Figure 1.
*Models included in the ensemble are listed in the box on the bo_om le`.
The Arc)c region warms rapidly accompanied by shrinking sea ice coverage. Last autumn unusual high SST hindered sea ice regrowth which lead to record low sea ice area and volume this past winter.
In the Arc)c (here defined as area >70˚N) surface air temperature (SAT) rises faster than the global mean under increasing greenhouse gas concentra)ons (see Figure 1). While this Arc)c Amplifica)on is present throughout the troposphere we focus on the strong surface effect
1.
What causes it?
Several processes are involved:
• albedo contrast between ice and ocean
• seasonal heat storage in ocean and exchange with atmosphere (moderated by sea ice)
• year-to-year memory in sea ice thickness
• temperature and radia)on (lapse rate and Planck) feedbacks
• change in cloud type and coverage
• snow cover and vegeta)on change on land
• heat and water vapor advec)on from subpolar region
Note, as sea ice vanishes, its effec)veness in Arc)c amplifica)on decreases.
Models
This study is based on CMIP5 simula)ons including the following climate models:
ACCESS1-0, ACCESS1-3, BNU-ESM, CCSM4, CESM1-BGC, CESM1-CAM5, EC-EARTH, HadGEM2-AO, HadGEM2-CC, HadGEM2-ES, MIROC-ESM-CHEM, MIROC5, MIROC-ESM, MPI-ESM-LR, MPI-ESM-MR, which have been verified by different metrics to best simulate the present sea ice decline
2,3. We include runs of the historical and RCP8.5 scenarios.
Figure 1: Autumn (September to November) Arc)c mean (blue) and global mean SAT anomalies (red) referenced to 1960–1980 mean and smoothed by a 30-year boxcar filter.
Figure 3: Arc)c September mean sea ice area from CMIP5 models (gray lines) and the mul)-model mean (black line).
Circles mark the year in which each model run passes through Se = 1 (see Fig. 6 on right). The ver)cal purple dashed line marks the mul)-model mean peak year of sea ice related feedbacks and is explained under .
Figure 2: Rates of SAT and sea ice area (SIA) change based on the 30-year boxcar filter smoothed mul)-model mean )me series shown in Figures 1 and 3 above and below.
For the yellow star and purple dashed line markers see and respec)vely.
16
19000 1920 1940 1960 1980 2000 2020 2040 2060 2080 2100
2 4 6 8 10 12 14 16 18
Sea ice extent (x106 km2 )
NSIDC, HadISST, PIOMAS, and historical + RCP8.5 runs
(a)
NSIDC HadISST PIOMAS CMIP5
1900 1920 1940 1960 1980 2000 2020 2040 2060 2080 2100
Year
0.5 1 1.5 2 2.5 3
Seasonality number, Se
P e r e n n i a l Seasonal
Episodic(b)
Antarctic
extent area
NSIDC HadISST PIOMAS CMIP5
80
o N
NSIDC + Multi-model mean
150
o W 120 oW
90 W o
60 o
W 30oW 0
o
30
o E
60o E
90 E o
120 o
E 150oE
180
o W
Figure 1. Increasing seasonality of Arctic sea ice extent. (a) An- nual range from data and model simulations, (b) The correspond- ing seasonality numbers. NSIDC satellite sea ice index (HadISST) data are shown since (prior to) 1978. The PIOMAS estimates are from a data assimilation product. The multi-model mean from three CMIP5 coupled climate models are also shown (three realizations of CESM1-CAM5, and one realization each of MIROC-ESM-CHEM, MPI-ESM-MR from the historical and RCP8.5 runs). In (b) season- ality numbers for NSIDC and HadISST Antarctic sea ice extent are plotted. The thin lines show the seasonality numbers for NSIDC sea ice area. Inset: Map of sea ice extent when Se = 1 ± 0.05 (the brown line shows the NSIDC ice edge and the turquoise patch is from the CMIP5 models). The yellow patch shows the CMIP5 distribution of sea ice extent when the total extent decreases to (1 ± 0.6) ⇥ 10
6km
2. In each case, the composite multi-model mean distribution is shown.
F i g u r e 6 : I n c r e a s i n g seasonality of Arc)c sea ice extent and area: We define the seasonality number Se as
the ra)o of seasonal range to annual mean sea ice extent/
area4. Seasonality numbers presented are from NSIDC satellite data5 (Arc)c and Antarc)c) as well as PIOMAS6 Arc)c hindcast w/assimila)on for the observa)onal period, HadISST7 (Arc)c & Antarc)c) prior to 1978, and the CMIP5 mul)-model Arc)c mean for 1900-2100 (historical + RCP8.
Arc+c and global warming
Year-to-year change in temperature and sea ice
Sea ice area decline
Peak in coincident sea ice and SAT year-to-year change
Figure 4: Peak in coincident sea ice retreat and Arc)c SAT warming (indicated by purple ‘x’ in le` panel). This year is marked by a dashed purple line in Figures 2, 3 & 5. Color-filled circles depict year-to-year changes of Arc)c September mean sea ice area and autumn mean SAT. Right panel shows the same but for global autumn mean SAT. Here, the 30- year boxcar filter smoothed mul)-model means of Figure 2 are shown.
The northern hemisphere sea ice system has entered the seasonal regime
1
2
3
4 Peak in Arc+c Amplifica+on shiCs from
2010s in early autumn to 2080s in winter
CMIP5 models*
mul/-model mean
year of ice feedback peak year of Se = 1
1 Screen, J. A. & Simmonds, I. (2010) The central role of diminishing sea ice in recent Arc)c temperature amplifica)on. Nature 464, 1334–
1337. doi:10.1038/nature09051.
2 Wang, M. & Overland, J. E. (2015) Projected future dura)on of the sea-ice-free season in the Alaskan Arc)c. Prog. Oceanogr. 136, 50–59.
doi:10.1016/j.pocean.2015.01.001.
3 Liu, J., Song, M., Horton, R. & Hu, Y. (2013) Reducing spread in climate model projec)ons of a September ice-free Arc)c. Proc. Nat. Acad. Sci.
110 (31), 12571–12576 (2013). doi:10.1073/pnas.1219716110.
4 Haine, T. W. N. & T. Mar/n (2017) The Arc/c-Subarc/c sea ice system is entering a seasonal regime: Implica/ons for future Arc/c amplifica/on. Scien/fic Reports, submiRed, revised.
5 Fe_erer, F., Knowles, K., Meier, W. & Savoie, M. (2016) Sea ice index, version 2. Tech. Rep., Na)onal Snow and Ice Data Center, Boulder, Colorado USA (updated daily, accessed 6 April 2017).
6 Zhang, J. & Rothrock, D. A. (2003) Modeling global sea ice with a thickness and enthalpy distribu)on model in generalized curvilinear coordinates. Mon. Weather. Rev. 131, 845–861.
7 Rayner, N. A. et al. (2003) Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res. 108.
Figure 7: Time series of mul)-model mean Arc)c Amplifica)on factors by month from early autumn to mid-winter. Clearly, Ar)c Amplifica)on i s s t r o n g e s t i n N o v e m b e r . A progression of the peak in each month f r o m 2 0 0 8 f o r September to the late 2 1s t c e n t u r y f o r February can be seen.
For the yellow star and purple dashed line markers see and respec)vely.
Here, the peak in SAT and SIA changes is calculated for each month. From mid- winter to spring sea ice retreat becomes less dominant and Arc)c Amplifica)on is driven by other processes preven)ng a meaningful computa)on of this metric.
2 3
2 3
year
• Surface Arc)c Amplifica)on peaks under con)nuous global warming.
• For autumn months this peak occurs in the 2010s but later in 2060-2080 for the winter season according to model projec)ons under the CMIP5 RCP8.5 climate scenario.
• The autumn peak coincides with the Arc)c sea ice cover entering the seasonal regime …
• and a maximum in effec)veness of sea ice in associated feedback mechanisms, such as the ice-albedo feedback.
• The seasonality number
4is thus a more meaningful measure of the changing Ar)c than the 10
6km
2sea ice extent threshold o`en cited in the literature to iden)fy and ice-free Arc)c.
Se(T) = (t2 −t1) x(t2)− x(t1) x(t')dt'
t1 t2