July 2017 SIO Data Cleaned
 Share
The version of the browser you are using is no longer supported. Please upgrade to a supported browser.Dismiss

 
View only
 
 
ABCDEFGHIJKLMNOPQRST
1
TimestampNameLeadBlankNameOrgCitizenReUseExecSumTypeSICDataSITDataEmail AddressComponentsBriefExplainArcticExtentAntarcticExtentBlankUncertaintyUncertaintyBasisPostProcessingAlaskaExtent
2
6/3/2017 9:13:12BosseFrank Bosse
Yes this contribution is from a "Citizen Scientist"
Yes automatically include my contributions in July and August 2017
Just as in the last years my guess is derived from the 0...700m OHC of the Atlantic part of the arctic. For the detailed mapping of the data from Argo and the physical explanation see https://www.arcus.org/files/sio/23220/bosse_july2015.pdf . The average of the monthly June...September OHC-data of the year n is the input for a linear regression to estimate the september mean of the sea ice extent for the year n+1. In this year I do not consider the PIOMAS- data due to some doubts of the validity of this model in the light of the Cryosat data.
Mixeddh7fb@gmx.de
The arctic melting is strong influenced by the waterinflow of the Atlantic and the forcing. Both parts are represented in the OHC of the 60°N...65°N part which is used ( the JJAS data) for a linear regression The data are available via the "KNMI Climate Explorer", thanks to Geert-Jan. The skill of the guess can be shown with this plot: https://picload.org/image/riliowwr/image002.gif
It shows running 15a-Trends for the guess ( in brown) and for the observed data(blue). The guess -trends reflect the observed trend-changes to a high degree. The "rest" ( 19%) are the weather patterns of the melting season which influence the melting less than thought, see: http://discovery.ucl.ac.uk/1519678/1/Serreze_et_al-2016-Journal_of_Geophysical_Research__Atmospheres.pdf .
4.620.45
The uncertainty is calculated with the single std.dev. of the residuals 1979...2016. The year 1996 is excluded as an outlier.
3
6/4/2017 22:40:00
Sanwa elementary school
Arata Iihoshi,Naoto Aokihara,Manato Ikegami,Hinata Umayahara,Kou Umeoka,Yusaku Kaihara,Yusei Kishimoto,Harutsugu Sadakiyo,Taisei Sugihara,Ryomei Tamura,Kou Teragauchi,Tetsuya Fujii,Taiyo Yamamoto,Kota Yokoyama,Kazuki Wakabayashi,Mizuki Kawakami,Mao Kojo,Momoka Saegusa,Sayo Shigeto,Yuna Hayashida,Miharu Fukushima,Sakuya Fukuman and Miki Hisanaga. Total number is 23.(The total number of students is 21.)
Yes this contribution is from a "Citizen Scientist"
Yes automatically include my contributions in July and August 2017
Monthly mean ice extent in September will be about 4.43 million square kilometers. We estimated the minimum ice area through discussion among 21 students based on the ice map from 2004 to 2016.
Heuristic
Sea ice Velocity is not used.
Sea ice Thickness is not used.
sanwa-sho@town.jinsekikogen.hiroshima.jp
A dynamic model is not used.
We first estimated the total ice area for September of 2004,2006,2008,2010,2012,2014 and 2016 from the ice concentration map,by approximating the ice cover with a triangle or trapezoid and so on.Based on this rough estimation, we discussed a yearly change of the ice area and calculated the ice area of this September.
4.43
4
6/12/2017 10:22:14MorisonJames Morison
Yes automatically include my contributions in July and August 2017
My June 2017 projection is for a new record low average September, 2017 Arctic sea ice extent of 3.4 million square kilometers. This heuristic estimate is based on what must be the worst pack ice conditions entering the summer season, namely:

A) Analysis from Ron Kwok had most of the multiyear ice off Ellesmere Island being swept out of Fram Strait by a persistent low over the central Arctic, and the January 1 multiyear fraction for 2017 was an all time low. The total ice volume must be at a record low for this time of year.

B) Temperatures over the Atlantic side of the Arctic Ocean up to the Pole were reportedly warm in late 2016 into early 2017.

C) High winter AO should negatively correlate with following September ice extent [Rigor et al., 2002]. Winter (NDJFMA) 2016-17 AO was 9th highest since 1950 and 1.1 above the 1950-88 average. This should influence to ice extent negatively.

As always, everything ultimately depends in the summer’s weather, but the ice initial conditions starting the summer melt must be the worst ever so I’m predicting a new record minimum September average of 3.4 million square kilometers.
Heuristic
analysis of multiyear ice over the winter 2016-17 by Ron Kwok, NSIDC ice extent record, and NOAA AO record
morison@apl.washington.edu
My method is heuristic based on experience, analysis of multiyear ice over the winter 2016-17 by Ron Kwok, NSIDC ice extent record, and NOAA AO record
3.41Experience
5
8/12/2017correctedaccordingtoemailfromJinlun
Jinlun Zhang and Axel Schweiger
University of Washington/APL
Yes this contribution is from a "Citizen Scientist"
Yes automatically include my contributions in July and August 2017
Driven by the NCEP CFS forecast atmospheric forcing, PIOMAS is used to predict the total September 2017 Arctic sea ice extent as well as ice thickness field and ice edge location, starting on June 1. The predicted September ice extent is 3.7± 0.5 million square kilometers. The predicted ice thickness fields and ice edge locations for September 2017 are also presented.
Dynamic Model
Satellite sea ice concentration data (NASA team)
PIOMAS hindcast of 12-category ice thickness distribution fields on 6/1/2017
zhang@apl.washinghton.edu
Pan-Arctic Ice-Ocean Modeling and Assimilation System (PIOMAS, Zhang and Rothrock, 2003), with coupled sea ice and ocean model components. The ocean model is the POP (Parallel Ocean Program) model and sea ice model is the TED (Thickness and Enthalpy Distribution) model. Atmospheric forcing is from the NCEP Climate Forecast System (CFS) version 2 (Saha et al., 2014) hindcast and forecast.
These results are obtained from a numerical seasonal forecasting system. The forecasting system is based on a synthesis of PIOMAS, the NCEP CFS hindcast and forecast atmospheric forcing, and satellite observations of ice concentration. The CFS forecast ranges from hours to months: there are a total of 16 CFS ensemble forecast runs every day, of which four ensemble runs go out to 9 months, three runs go out to 1 season, and nine runs go out to 45 days (Saha et al., 2014). These ensemble runs all create 6-hourly forecast atmospheric data that are widely accessible in real time, thus ideal for forcing PIOMAS forecasts on daily to seasonal time scales. Here we used four CFS forecast ensemble members to drive the PIOMAS ice–ocean ensemble forecasts. Ensemble mean values from these four members are considered to be the prediction. To obtain the “best possible” initial ice-ocean conditions for the forecasts, we conducted a retrospective simulation that assimilates satellite ice concentration and sea surface temperature data through the end of May 2017 using the CFS hindcast forcing data. After that, four ensemble PIOMAS forecast runs were conducted using atmospheric forecast forcing from four CFS ensemble runs. Additional information about PIOMAS prediction can be found in Zhang et al. (2008).
4.50.5
6
6/12/2017 18:58:57
McGill (Tremblay, Brunette, Williams)
Bruno Tremblay [1], Charles Brunette [1] (Primary contact), James Williams [1], [1]: McGill University, Department of Atmospheric and Oceanic Sciences, Montreal, Qc., Canada.
Yes automatically include my contributions in July and August 2017
We are studying predictability of sea ice in the Arctic Ocean, with a focus on improving our understanding of predictability on a seasonal timescale. To this end we take approaches entirely based on observations. We are interested in the preconditioning effect that winter sea ice dynamics has on the following summer melt. Williams et al (2016) showed that anomalous Fram Strait sea ice export during winter accounts for roughly 40% of the interannual variability of the September sea ice extent (SIE). We use the DJFMA Arctic Oscillation (AO) index as a predictor variable as a proxy for the Fram Export. The DJFMA AO index is significantly correlated with the winter Fram Strait export anomaly (r=0.55) as well as the monthly mean September SIE anomaly (r =-0.51). We have been following the reports of the Sea Ice Outlook as it is the most comprehensive evaluation of the skill of the sea ice prediction community in predicting the September SIE. We are submitting a forecast this year to add to the number of contributors and to see how we perform alongside other types of predictions.
Statisticaln/an/a
charles.brunette@mail.mcgill.ca
n/a
Our prediction for the monthly mean Arctic sea ice extent (SIE) of September 2017 is 3.95 million km^2. We produce the prediction as a sum of the linear trend (climatology) and departure from the trend (interannual variability). We take the long-term linear trend for the 1993-2016 period. We use the mean winter (DJFMA) Arctic Oscillation (AO) index in a linear least squares fit model as a predictor for the anomaly of monthly mean September SIE over the same period. This builds on the idea of winter dynamic preconditioning - see Williams et al. (2016). Since we use the mean DJFMA AO index, our prediction is made on May 1 and does not change during the summer. We use the Sea Ice Index V2 monthly SIE dataset (NSIDC) and the monthly AO index distributed by NOAA.
3.950.472
We produce and compare hindcasts to the observed monthly mean September SIE for the 1993-2016 period. The adjusted r^2 between the two time series is 0.766. The root mean square error is 0.472 million km^2.
n/a
7
6/13/2017 12:13:29
Kay, Bailey, Holland (NCAR/CU)
30 scientists affiliated with National Center for Atmospheric Research and/or University of Colorado at Boulder
Yes automatically include my contributions in July and August 2017
An informal pool of 30 climate scientists in early June 2017 estimates that the September 2017 ice extent will be 4.08 million sq. km. (std. dev. 0.48, min. 3.10, max. 5.09). Since its inception in 2008, the NCAR/CU sea ice pool has easily rivaled much more sophisticated efforts based on statistical methods and physical models to predict the September monthly mean Arctic sea ice extent (e.g. see appendix of Stroeve et al. 2014 in GRL doi:10.1002/2014GL059388 ; Witness the Arctic article by Hamilton et al. 2014 http://www.arcus.org/witness-the-arctic/2014/2/article/21066). We think our informal pool provides a useful benchmark and reality check for Sea Ice Prediction efforts based on more sophisticated physical models and statistical techniques.
Heuristicn/a
Jennifer.E.Kay@colorado.edu
An informal pool of 30 climate scientists in early June 2017 estimates that the September 2017 ice extent will be 4.08 million sq. km. (std. dev. 0.48, min. 3.10, max. 5.09). Guesses were collected by sending an e-mail out to the scientists.
4.08
std. dev. 0.48, min. 3.10, max. 5.09
The uncertainty estimate is based on the scatter in entries in our informal pool.
8
6/15/2017 8:13:07Cawley
Gavin Cawley, School of Computing Sciences, University of East Anglia
Yes automatically include my contributions in July and August 2017
This is a purely statistical method (Gaussian Process, related to Kriging) to estimate the long-term trend from previous observations of September Arctic sea ice extent. As this uses only September observations, the prediction is not altered by observations made during the Summer of 2016.
Statistical
ftp://sidads.colorado.edu/DATASETS/NOAA/G02135/north/monthly/data/N_09_extent_v2.1.csv
gcc@cmp.uea.ac.uk
Non-linear Gaussian process regression model with squared exponential covariance function, hyper-parameters optimised by marginal likelihood maximisation.
4.23
4.23 +/- 1.14 million square kilometers (95% Bayesian credible interval)
Gaussian process models provide credible intervals on model prediction.
9
6/19/2017 9:25:27Druckenmiller & Eicken
Matthew Druckenmiller, SEARCH Sea Ice Action Team, National Snow and Ice Data Center
Hajo Eicken, International Arctic Research Center, University of Alaska Fairbanks

With contributions from:
Stefan Hendricks, Alfred-Wegener Institute
Chris Polashenski, Cold Regions Research and Engineering Laboratory (CRREL)
Andy Mahoney, Geophysical Institute, University of Alaska Fairbanks
Josh Jones, Geophysical Institute, University of Alaska Fairbanks
Yes automatically include my contributions in July and August 2017
See provided regional outlook write-up
Observational
druckenmiller@nsidc.org
See provided regional outlook write-up
10
7/1/2017 17:09:49Yizhe ZhanUniversity of Auckland
Yes automatically include my contributions in July and August 2017
We estimated the September sea ice extent in 2017 to be much larger than last year. Compared to June 2016, MISR (firstlook product) indicates a significant increase in RSR over both the Canadian Arctic Archipelago and Svalbard, which consistent with the sea ice concentration anomaly. The area-weighted Pan-Arctic mean June RSR 2017 is much larger than that in 2016 and only slightly smaller than 2014.
Statisticaly.zhan@auckland.ac.nz
The statistical prediction is based on the significant 3-month lag correlation between June top-of-atmosphere reflected solar radiation (RSR) and September sea ice extent (SIE). See our JGR publication in http://onlinelibrary.wiley.com/doi/10.1002/2016JD025819/full

Here we used MISR firstlook product that have been released on 2nd June and its hindcast model, which is established from the detrended June RSR and September SIE data (2002-2016). The June RSR 2016 anomaly is calculated by subtracting it from the 2002-2016 RSR trend (+14.56 W/m2). This RSR anomaly is then applied to the model to estimate the September SIE anomaly of this year (14.56 * 0.077). Lastly, our final estimation is made by adding the anomaly to the September SIE from the 2002-2015 trend (4.34 + 1.12).
5.46±0.3
The resulting uncertainty is the prediction error of the MISR hindcast model for the years 2002 to 2016. It is the standard deviation of the differences between modeled and observed September SIE.
11
7/4/2017 5:15:53Christian Johnnone
Yes this contribution is from a "Citizen Scientist"
No do not use my prediction this month in later months
My contribution is based on the idea that Temperature of Sea-Surface and Air Temperature (70-90N) can be used to explain most of the Year to Year Variability of September Sea-Ice-Extent, it is able to show that both can work as a proxy for the strength of Extent Melt to September. While Longterm-Trend is mainly caused by human induced climate change, i therefore argue that the Year to Year variability can explained by Temperature-Variability
Statistical
ChristianJohn1@gmx.net
The used Method is based on linear Regression for all Variables and rewrite the Residuals back on the Longterm-Trend of September Sea Ice Exent. For the Outlook i use data from NSIDC(September-Extent), ERSSTv4 and NCEP-Reanalysis on the Domain 70-90N from 1979-2016

For scheme: http://www.directupload.net/file/d/4772/znvo8u8x_png.htm
For Performance: http://www.directupload.net/file/d/4772/gkq9usbq_png.htm
4.903
Uncertainty: (standard deviation)
June: 0.37 Mio km^2
June/July: 0.34 Mio km^2
The uncertainty is the standard deviation of the residual error between observation and model
12
7/6/2017 22:50:53Rob DekkerIndividual
Yes this contribution is from a "Citizen Scientist"
Yes automatically include my contributions in July and August 2017
My projection is based on an estimate of how much heat the Northern Hemisphere absorbs during spring and early summer. I use three variables (land snow cover, ice concentration, ice area) that are available in June, in a formula which shows particularly strong correlation with Sept sea ice extent. Regressed over the1992 - 2015 period, the formula projects 4.1 M km^2 for September 2016, with a standard deviation of only 340 k km^2.
Past performance of this June forecast method for September ice extent over the past 24 years shown in a graph here :
http://i1272.photobucket.com/albums/y396/RobDekker/JunePredict_zpsquedrtdc.png

The interesting finding is that the June land snow cover signal is clearly present in the September ice extent numbers.
Statistical
NSIDC monthly June sea ice 'extent' and 'area' numbers : ftp://sidads.colorado.edu/DATASETS/NOAA/G02135/Jun/N_06_area_v2.txt Rutgers Snow Lab Northern Hemisphere monthly land snow cover : http://climate.rutgers.edu/snowcover/table_area.php?ui_set=1&ui_sort=0
rob@verific.com
The concept behind my method pertains to estimating albedo-based Arctic amplification during the melting season.
I use the "whiteness" of the Arctic in June as a predictor for how much ice will melt out between June and September.
Specifically, I set up a formula which reflects how "dark" areas near the Arctic in June would create heat that will melt out ice over the months until the September minimum.
As an educated guess, such a formula could take the following form :
Melt_formula = 0.25 * Snow - 1.0 * (Extent - Area) + 0.5 * Area
With factors explained like this :
For (Extent - Area): 1.0 (assuming that ALL solar radiation onto melting ice and into polynia will cause ice to melt later in the season.
For (Area): 0.5 (assuming that half of the heat absorbed in the ocean OUTSIDE of the main pack will cause ice melt (while the other half would cause the ocean to warm up.
For (snow cover): 0.25 (assuming that half the heat from lack of snow cover will be blown North, and half of that will go to ice melt. Then I set up a regression equation for how much ice will melt out between June and September :
september_extent - june_area = alpha + beta * (Melt_Formula) ;
When I tweek the factors, to obtain the best fit over the 1992-2015 range, the ‘Melt_Formula' that obtains the best correlation (R=0.94) is this one (centered to (extent - area):
Melt_Formula = 0.434*snowcover - 1.0*(extent - area) + 0.65*area
Which is remarkably close to the "educated guess" factors explained above. This suggests that this formula is realistic, and the effect is physically real.
Using this formula, for the period 1992 - 2015, I obtain R=0.94, beta = 0.368, and a prediction for Sept 2017 of 5.4 M km^2 with a standard deviation over the residuals of the (hindcast) prediction of 342 k km^2. Which is considerably better than the (hindcast) SD of a linear trend (550 k km^2).
5.4
342 k km^2 (hindcast) standard deviation over the residuals
13
7/7/2017 9:18:57PettyNASA-GSFC
No do not use my prediction this month in later months
Based on an analysis of June sea ice concentration data provided by the NSIDC (NASA Team), I forecast a 2017 September Arctic sea ice extent of 4.50 +/- 0.35 M km2. This is slightly lower than the May forecast, and instead is very similar to the extent expected from persistence of the long-term linear trend. The forecast does not suggest a new record low September extent will be reached in 2017 (lower than the 3.62 M km2 observed in 2012).

The June forecast appear to be driven primarily by the low SIC in the Beaufort and Laptev seas.

Our June forecast of Alaskan sea ice extent is coming in at 0.37 M km2, so slightly lower than the May forecast (0.41).

We also produced, for the first time, an Antarctic September sea ice forecast using the same methodology. This does show some skill, especially in more recent years (since 2008). The forecast is 18.93 Mkm2
StatisticalNASA Team, June 2017alekpetty@gmail.com
In this forecast we use sea ice concentration (SIC) data (1979-present day), derived from passive microwave brightness temperature using the NASA Team algorithm. The SIC data are detrended spatially using linear trend persistence (from the given forecast year) then averaged, to generate a detrended SIC dataset. A least-squares linear regression model is fit from the mean detrended SIC/SIE data. To produce the SIE forecast, the relevant monthly mean/detrended SIC data are applied to the linear regression model. See my website (http://alekpetty.com/blog/2017ArcticForecasts) for more details.
4.518.930.37
14
7/7/2017 13:50:05
NASA Global Modeling and Assimilation Office (NASA GMAO)
Richard I. Cullather [primary contact; 1,2], Anna Y. Borovikov[1,3], Eric C. Hackert [1], Robin M. Kovach [1,3], Jelena Marshak [1], Andrea M. Molod [1], Steven Pawson [1], Max J. Suarez [1,4], Yury V. Vikhliaev [1,4], and Bin Zhao [1,5]
[1] Global Modeling and Assimilation Office, NASA Goddard Space Flight Center, Greenbelt, MD.
[2] Earth System Science Interdisciplinary Center, University of Maryland at College Park.
[3] Science Systems and Applications, Inc., Greenbelt, MD.
[4] GESTAR, Universities Space Research Association, Columbia, MD.
[5] Science Applications International Corporation, Greenbelt, MD.
No do not use my prediction this month in later months
An experiment of the GMAO seasonal forecasting system using CryoSat-2 derived ice thickness predicts a September average Arctic ice extent of 4.90 ± 0.34 million km2. The test examines the application of ice thickness data in a near-real time setting for the seasonal forecast system.
Dynamic Model
NASA Team for 01-Apr, 16-Apr, 01-May, 16-May, and 31-May 2017
GMAO ODAS ice thickness integrated using CryoSat-2 derived daily ice thickness inserted over the period 1-January to 1-April, obtained from the Goddard Cryospheric Sciences Laboratory (Kurtz et al., 2014).
richard.cullather@nasa.gov
Model Name: Goddard Earth Observing System Model (GEOS).
Atmosphere: GEOS AGCM initialized with MERRA-2 and GMAO forward processing NWP analysis.
Ocean: MOM4 initialized with GMAO Ocean Data Assimilation System (LETKF).
Ice: CICE4 (LETKF).
The GMAO seasonal forecast is produced from coupled model integrations. The main components of the AOGCM are the GEOS atmospheric model, the MOM4 ocean model, and CICE sea ice model. Daily CryoSat-2 derived ice thickness observations from 1-January through 1-April were inserted into the GMAO Ocean Data Assimilation System (ODAS) based on the model background ice thickness distribution. The forecast model was initialized from five restarts dropped from the ODAS system over the April and May period. Forecast fields were re-gridded to the passive microwave grid for averaging.
4.9N/A0.34
The given uncertainty is the standard deviation of the 5 member ensemble.
The model output was re-gridded to the standard Northern Hemisphere passive microwave grid.
0.90± 0.25; reference area: 4.00
15
7/9/2017 20:09:48Reid, Davies, Massom
Phil Reid (Australian Bureau of Meteorology and Antarctic and Climate Ecosystems Cooperative Research Centre), Laura Davies (Australian Antarctic Gateway) and Rob Massom (Australian Antarctic Division and Antarctic and Climate Ecosystems Cooperative Research Centre. Phil Reid is primary contact
Yes automatically include my contributions in July and August 2017
We provide an outlook of the net Antarctic sea-ice extent for September 2017, based on observations up to and including the 8th July 2017. We use a statistical method that matches current observations with those from the past.
We obtain a mean-September extent of 17.7 x 106 kms2 and a annual daily maximum value of 17.8 x 106 kms2 occurring on the 22 September.
Statistical
NASA Team 1979 through current.
p.reid@bom.gov.au
Basic statistical analysis:
1. Match current (past 30-days) of pattern of sea –ice extent and growth rate with past observations (1979-2016) for the same time of the year. Note that this is not matching the net-extent, but the pattern of sea ice and growth rate.
2. Use best match from (1), along with current conditions to provide outlook.
17.7
16
7/10/2017 5:43:42CPOM
CPOM (D. Schroeder, D. Feltham, D. Flocco, M. Tsamados)
Yes automatically include my contributions in July and August 2017
Based on melt pond fraction in May+June we predict a mean 2017 September ice extent of 5.1 (4.6 to 5.6) mill km2 (within the range observed during last 4 years). The likehood for a new record minimum is below 1%. While melt pond fraction has been above 2006-2015 mean values in the western parts of the Arctic, less ponding and melting occurred in the eastern part due to more snow and relatively cold temperatures. In the past the regions - where melt pond fraction is low in 2017 - were more important for September ice extent than e.g. the Beaufort Sea. Consequently, we predict the September ice extent to be quite large in spite of the lowest Arctic ice volume in recent months.

Statisticaln/an/a
D.Schroeder@reading.ac.uk
n/a
This is a statistical prediction based on the correlation between the ice area covered by melt-ponds in May and ice extent in September. The melt pond area is derived from a simulation with the sea ice model CICE in which we incorporated a physically based melt-pond model1. See our publication in Nature Climate Change http://www.nature.com/nclimate/journal/v4/n5/full/nclimate2203.html for details2.

References:
1. Flocco, D., Schröder, D., Feltham, D. L. & Hunke, E. C., 2012: Impact of melt ponds on Arctic sea ice simulations from 1990 to 2007. J. Geophys. Res. 117, C09032.
2. Schröder D., D. L. Feltham, D. Flocco, M. Tsamados, 2014: September Arctic sea-ice minimum predicted by spring melt-pond fraction. Nature Clim. Change 4, 353-357, DOI: 10.1038/NCLIMATE2203.
5.10.5
The given uncertainty is the mean forecast error based on forecasts for the years 1984 to 2016. For all these forecasts only data from previous years were used (forecast mode).
n/a
17
7/10/2017 9:25:51
Navy Earth System Model (NESM)
E. Joseph Metzger 1(primary contact), Neil Barton2, Pamela Posey1, Alan Wallcraft1 and Michael Phelps3
1Naval Research Laboratory, Marine Oceanography Division
2Naval Research Laboratory, Marine Meteorology Division
3Jacobs Technology Inc, Stennis Space Center, MS
No do not use my prediction this month in later months
The projected Arctic minimum sea ice extent from the Navy's global fully coupled atmosphere-ocean-ice modeling system (Navy Earth System Model - NESM) is 4.5 Mkm2. This projection is the average of a 10 member ensemble. The range of the ensemble is 4.2 to 5.0 Mkm2. Note that our ensemble range does not represent a full measure of uncertainty, and the system is currently in a development stage.
Dynamic Model
SSMIS and JAXA AMSR2, June 2017
GOFS 3.1 SIT, June 2017
pamela.posey@nrlssc.navy.mil
Component Name Initialization
Atmosphere NAVGEM DA:NAVDAS-AR
Ocean HYCOM DA:NCODA
Ice CICE DA:NCODA assimilating SIC only
Forecasts were initialized from the pre-operational US Navy Global Ocean Forecasting System (GOFS) 3.1 for the ocean and sea ice using the Navy Coupled Ocean Data Assimilation (NCODA) system. The ice model assimilated SSMIS and AMSR2 ice concentration products. Atmospheric initial conditions were from the operational NAVy Global Environmental Model (NAVGEM) using the Navy Research Laboratory Atmospheric Variational Data Assimilation System (NAVDAS-AR). Ten ensemble members were completed using a time-lagged approach. Model forecasts started at 12Z for the following days: June 1, 2, 3, 4, 6, 7, 8, 9, 10 and 12. Each ensemble member was integrated through the end of September 2017.
4.521.2
Pan-Arctic: 4.2 to 5.0 Mkm2
Pan-Antarctic: 20.0 to 21.6 Mkm2
Alaskan: 0.3 to 0.6 Mkm2
The uncertainty estimate is the range of the 10 member ensemble, and does not represent a full measure of uncertainty. Projections of Sept sea ice extent with this system have not been fully validated.
The only post-processing performed was the calculation of the mean September sea ice extent from sea ice concentrations.

For Sea Ice Probability (SIP): We computed SIP as requested: converted Sept mean SIC into SIE for each ensemble member. Then averaged the ensemble across the Sept mean SIE. Hence, SIP is the probability of sea ice cover in the ensemble and ranges from 0 to 100%.

For Ice-Free Day (IFD): We computed the first ice-free day when SIC falls below 15% for all points where there is at least 15% SIC on the day we initialized the model. If the point is ice free (SIC<15%) at initialization, IFD will be ordinal day 152 (June 1). If the point is always covered in ice (SIC>=15%), the IFD will be ordinal day 273 (Sept 30). We then computed the average and standard deviation of IFD across the ensemble.
0.5 Mkm2. Max possible extent is 3.98 Mkm2 using NSIDC region definition.
18
7/10/2017 12:13:05Modified_CanSIPS
Primary contact: William J. Merryfield, ECCC/CCCma
Arlan Dirkson, UVic/SEOS Woosung Lee, ECCC/CCCma
Michael Sigmond, ECCC/CCCma
Slava Kharin, ECCC/CCCma
Adam Monahan, Uvic/SEOS

ECCC* Environment and Climate Change Canada
CCCma* Canadian Centre for Climate Modelling and Analysis
UVic* University of Victoria
SEOS* School of Earth and Ocean Sciences
Yes automatically include my contributions in July and August 2017
Our forecasts of total Arctic sea ice extent (SIE) and sea ice probability (SIP) were produced using the Canadian Seasonal to Interannual Prediction System (CanSIPS), but in a modified experimental mode intended to test several potential updates to the sea ice forecast methodology. These updates include changes to the data used to initialize both sea ice concentration (SIC) and sea ice thickness (SIT), as well as the methodology to produce probabilistic SIC forecasts.
Dynamic Model
SIC is initialized by nudging model SIC to the Canadian Meteorological Centre (CMC) daily SIC analysis with a 3 day time constant. The initial SIC field provided to the SIO is the ensemble average of these nudged SIC values on June 30, the day the forecasts were initialized.
SIT was estimated using the statistical model ‘SMv3’ described in Dirkson et al., 2017 (doi:10.1175/JCLI-D-16-0437.1). The parameters in SMv3 were fit using PIOMAS SIC and SIT data over the period 2001-2016. The daily CMC SIC field described above for June 30 was then used as the real-time predictor in SMv3 to estimate real-time SIT.
bill.merryfield@canada.ca
Component Name Initialization
assimilation: nudging
CanCM3
Atmosphere CanAM3 6-hourly CMC GDPS analysis
Ocean CanOM4 daily CMC (SST), GIOPS (subsurface)
Ice Cavitating fluid daily CMC SIC, SMv3 SIT

CanCM4
Atmosphere CanAM4 6-hourly CMC GDPS analysis
Ocean CanOM4 daily CMC (SST), GIOPS (subsurface)
Ice Cavitating fluid daily CMC SIC, SMv3 SIT
CanSIPS combines forecasts from two models, CanCM3 and CanCM4, with a total of 20 ensemble members (10 from CanCM3, 10 from CanCM4). The Arctic SIE anomaly was calculated for each individual ensemble member relative to the 1981-2010 climatology for the respective model. These anomalies were then added to the NSIDC climatological value of 6.5 million square kilometers, and then averaged over all 20 ensemble members to yield a total SIE of 4.33 million square kilometers.

For constructing the SIP map, we first fit the 10-member ensemble SIC values from each model (per grid point) to a zero- and one- inflated beta distribution (Ospina and Ferrari, 2010 doi:10.1007/s00362-008-0125-4; Dirkson et al, 2017 in preparation). After calibrating the parametric distribution per grid point and per model (as described below in question 10c), we then calculated the probability that local SIC will exceed 15% (or equivalently SIP) directly from the calibrated parametric distribution. Lastly, the average was taken between CanCM3 and CanCM4 SIP estimates to produce the final SIP map.

Historical skills for prediction of September-mean SIE from the beginning of July by Modified_CanSIPS are:

Anomaly Correlation Coefficient:
0.90 (1979-2016)
0.85 (1981-2010)

Linearly Detrended Anomaly Correlation Coefficient:
0.57 (1979-2016)
0.51 (1981-2010)

Root-Mean Square Error (in million sq km)
0.52 (1979-2016)
0.47 (1981-2010)
4.33
The uncertainty in forecast total SIE is estimated to be ±0.58 million square kilometers. This value was found by calculating the standard deviation of the ensemble of 20 forecast SIE anomalies, and multiplying by 1.96 to estimate the 95% range (between the 2.5 to 97.5 percentiles) of the forecast distribution. The anomalies were calculated relative to each model’s climatology over the period 1981-2010. The uncertainty in spatially-distributed (local) SIE is provided in the map of SIP.
The uncertainty estimate of ±0.58 million square kilometres for total SIE, and that indicated in the spatial map of SIP, reflects the inter-model uncertainty between CanCM3 and CanCM4 after having removed model mean bias, as well as the uncertainty in the climate system simulated by each ensemble member, resulting from imperfect knowledge of initial conditions.
For pan-Arctic SIE, a simple climatological bias correction was performed, as described above in question 6. For SIP, the parametric distributions described in question 6 were calibrated using a modified version of the ‘quantile mapping’ technique that explicitly accounts for SIC trends and is specifically designed for the parametric distribution used to fit SIC ensemble forecasts (Dirkson et al., 2017 in preparation). The calibration relies on previous CanCM3/CanCM4 forecasts of SIC and observations of SIC from HadISST2 (which is most similar to the CMC SIC data used to initialize SIC in real time) over the 1981-2016 period.
19
7/10/2017 14:54:21
Walt Meier, NASA Goddard
NASA Goddard, Walt Meier (sole contributor)
No do not use my prediction this month in later months
This method applies daily ice loss rates to extrapolate from the start date (June 30) through the end of September. Projected September daily extents are averaged to calculate the projected September average extent. Individual years from 2005 to 2016 are used, as well as averages over 1981-2010 and 2005-2016. The 2005-2016 average daily rates are used to estimate the official submitted estimate.
The predicted September average extent for 2017 is 4.52 (±0.63) million square kilometers. The minimum daily extent is predicted to be 4.41 (±0.62) million square kilometers and occur on 17 September. These values are lower than the June submission (September average of 4.82 [±0.70] million square kilometers and minimum daily extent of 4.70 [±0.71] million square kilometers). Though smaller than in June, the large range of July estimates reflects the large variability in ice loss rates over the final 2+ months of the melt season. Based on the last 12 years, there is only small chance that 2017 will be lower than the current record low extent of 2012.
Statistical
NSIDC NRT NASA Team, June 2017
walt.meier@nasa.gov
This method applies daily ice loss rates to extrapolate from the start date (June 30) through the end of September. Projected September daily extents are averaged to calculate the projected September average extent. Individual years from 2005 to 2016 are used, as well as averages over 1981-2010 and 2005-2016. The 2005-2016 average daily rates are used to estimate the official submitted estimate. The method essentially provides the range of September extents that can be expected based on how the ice has declined in past years, though it is possible that record fast or slow daily loss rates may yield a value outside the projected range. It also can provide a probability of a new record by comparing how many years of loss rates yield a record relative to all years. It has the benefit that it can easily and frequently (daily if desired) be updated to provide updated estimates and probabilities and as the minimum approaches the “window” of possible outcomes narrows.
4.52
+/-0.63 million sq km; standard deviation of predictions from 2005-2016
The projections using the 12 most recent years (2005-2016) are averaged for the official submissions. The range of values (st. dev.) over those 12 years is used as an uncertainty estimate.
Daily total extent values are used to calculate daily extent change rates for different years. These are added day-by-day to the initial 2017 extent (June 30) to simulate daily extent through September 2017. The daily values from September are averaged to obtain the monthly September estimate.
20
7/11/2017 0:58:09Nico Sun
Nico Sun
CryosphereComputing
Yes this contribution is from a "Citizen Scientist"
No do not use my prediction this month in later months
The forecast model is based on my own global surface radiation model and uses arctic sea ice albedo and northern hemisphere snow cover to calculate daily sea ice area and volume losses. The albedo values are approximated from concentration data. The average error for the 2007-2016 period is 0.147 million km2 for daily minimum sea ice area. The final average September extent value is calculated over the compaction ratio.
Mixed
NASA Team, 20 March 2017
PIOMAS, 20th March 2017
nicosun91@gmail.com
The feedback system between sea ice area, extent, volume and energy absorption is a dynamic model.
The prediction of extent to area ratios for the remaining melting season is statistical.
4.243
Interdecile range (+-1.28 SD):
3.6 - 4.9 million km2
NoneNone
21
7/11/2017 13:17:18
Brettschneider/Walsh/Thoman, UAF
Brian Brettschneider, University of Alaska, Fairbanks
John Walsh, University of Alaska, Fairbanks
Richard Thoman, National Weather Service/NOAA
Yes automatically include my contributions in July and August 2017
Forecast is based on an analog method. The analogs are based on sea level pressure, air temperature, sea surface temperature and upper-air geopotential height over several domains within (and including the entire) Northern Hemisphere. The September sea ice extent, expressed as a departure from the 1979-2016 trend line, is the predicted quantity. The actual prediction is the mean September pan-Arctic ice extent of the five best analog years.
Statistical
Atmospheric data were all from the NCEP/NCAR Reanalysis R1. Ocean temperatures were from ERSST v4. Historical time series of September sea ice extent was from NSIDC.
N.A.jwalsh@iarc.uaf.eduN.A.
The atmospheric analogs for this forecast are based on June data (sea level pressure, air temperature, sea surface temperature, upper-air geopotential height). The best atmospheric analogs were selected based on four domains, and the corresponding predictions obtained using each domain were weighted according to the skill shown by hindcast forecasts based on each domain. In each case the analog years were selected based on closeness-of-match to June 2017, where the closeness was measured by the root-mean-square difference of the respective fields.
4.52
10% and 90% confidence levels: 3.99 and 5.29 million km2.
A set of 5000 random forecasts (each based on five randomly chosen years) provided a distribution from which the 10th- and 90th-percentile values were obtained.
Predicted quantity is the departure from the linear trend line (1979-2016) of September sea ice extent.
22
7/12/2017 7:42:08
CNRM System 6 (Chevallier et al)
Matthieu Chevallier, Constantin Ardilouze, Lauriane Batté (CNRM, Meteo France, Toulouse, France), Clotilde Dubois (Mercator Ocean, Toulouse, France) ; CNRM PASTEL and IOGA teams (~10 people).
Yes automatically include my contributions in July and August 2017
This second CNRM outlook has been run with Meteo France system 6. This pre-operational system is based on a new version of CNRM global climate model, CNRM-CM6, and on new ocean-sea ice initial conditions produced by Mercator Ocean. The system used to produce this outlook differs with the other ("CNRM") with respect to (i) model horizontal and vertical resolution (ii) model physics in the atmosphere and ocean and (iii) new initial conditions.
Dynamic Model
Initial conditions for the ocean and sea ice (concentration and thickness) are provided by Mercator Océan. Basis is the Mercator Océan operational analysis, run at a 1/4° horizontal resolution using NEMO-LIM2 and the SAM ocean data assimilation system. There is no data assimilation of sea ice concentration in this analysis. The 1/4° analysis is upscaled to the 1° horizontal grid of CNRM-CM. These fields are used to nudge the ocean-sea ice component of CNRM-CM (NEMO3.6-Gelato6, 1° resolution), run in forced mode (forced by ECMWF Op. analysis). A strong restoring is applied towards Mercator SST, which acts as a constraint on sea ice concentration. Sea ice fields (concentration, thickness...) from this 1° run are used to initialize CNRM-CM (as well as ocean fields from this run).
Sea ice thickness information is output from the 1° simulation described above.
matthieu.chevallier@meteo.fr
Météo France System 6
Component Name Initialization
Atmosphere ARPEGE-Climat v6 ECMWF Operational Analysis
Ocean NEMO3.6 Based on Mercator (DA SEEK)
Sea ice GELATO 6 Based on Mercator (no DA)


Basis of System 6 is the global coupled model CNRM-CM-6-1, which will be used for CNRM contribution to CMIP6.
Horizontal resolution of the atmosphere component is T359 (50km). Nominal resolution of the ocean-sea ice component is 1° at the equator (nearly 50km in the Arctic), with a ~1m vertical resolution close to the surface. The sea ice model uses 4 sea ice categories (0-0.3; 0.3-0.8, 0.8-3 and >3m).
The "CNRM System 6" outlook is a model estimate based on a dynamical ensemble forecast. Initial conditions from the weeks before 1 July 2017 are used. We generate a total 51 members by adding statistical perturbations during the run.
4.6717.9
For pan-Arctic SIO
Median : 4.69
25% : 4.52
75% : 4.86
Min : 3.92
Max : 5.32

For pan-Antarctic SIO
Median : 17.92
25% : 17.57
75% : 18.18
Min : 16.94
Max : 18.76

Statistics are based on the 51-member ensemble.
Standard deviation of the ensemble is 0.30 million km2 for Arctic, 0.41 million km2 for Antarctic.
For the sea ice extent, data are corrected for bias and (linear) trend, using only the hindcast (the hindcast of System 6 is run over the period 1993-2016).
23
7/12/2017 8:13:51
Wanqiu Wang, Thomas Collow, Jinlun Zhang
Name and organization: Climate Prediction Center. Primary contact: Wanqiu Wang. Total number of people: 3
No do not use my prediction this month in later months
The outlook is calculated 0.5X0.5 grid output
Dynamic Model
Climate Forecast System Reanalysis
PIOMAS, June 21-25, 2017
Wanqiu.Wang@noaa.gov
Model name: CFSv2pp
Information about components:
Component Name Initialization
Atmosphere: GFS CFSR
Ocean MOM4p0 CFSR
Ice SIS PIOMAS
The outlook is from CPC experimental dynamical ensemble forecast. The ensemble includes 20 forecast members.
4.60.26
Bias has been removed based on 10 year handcast. Uncertainty is based on ensemble spread (standard deviation from ensemble mean).
Bias has been removed based on 10 year handcast. Uncertainty is based on ensemble spread.
0.42. maximum possible extent is 3.97
24
7/12/2017 8:35:51
Xingren Wu and Robert Grumbine
Xingren Wu and Robert Grumbine
MOAA/NCEP/EMC
Primary contact: Xingren Wu
Yes automatically include my contributions in July and August 2017
The projected Arctic minimum sea ice extent from the NCEP CFSv2 model with revised CFSv2 May and June initial conditions (ICs) using 61-member ensemble forecast is 4.21 million square kilometers with a standard deviation (SD) of 0.53 million square kilometers.
Dynamic Model
NCEP Analysis for May and June 2017
CFSv2 model guess w/ bias correction
Xingren.Wu@noaa.gov
Model Name: NCEP CFSv2
Component Name Initialization
Atmosphere NCEP GFS NCEP CDAS
Ocean GFDL MOM4 NCEP GODAS
ICE Modified GFDL SIS SIC nudging
61 ensemble members (May 1-June 30 2017, each day at 00Z cycle)
We ran the NCEP CFSv2 model with 61-case of May and June 2017 revised ICs. The IC was modified from real time CFSv2 of each day at 00Z by thinning the Arctic ice pack (based on test from previous years’ sea ice outlook). If this thinning would have eliminated ice from areas observed to have sea ice, a minimum thickness of 10 cm was left in place for the ice IC. Bias correction was applied to the Antarctic sea ice extent.
4.2119.64
For Pan-Arctic: The range is 3.06-5.22 million square kilometers with a
standard deviation of 0.53
For Pan-Antarctic: The range is 19.00-20.19 million square kilometers with a
standard deviation of 0.23
We ran the NCEP CFSv2 model with 61-case of May and June 2017 revised ICs. The IC was modified from real time CFSv2 of each day at 00Z by thinning the Arctic ice pack (based on test from previous years’ sea ice outlook). If this thinning would have eliminated ice from areas observed to have sea ice, a minimum thickness of 10 cm was left in place for the ice IC. Bias correction was applied to the Antarctic sea ice extent.
25
7/12/2017 18:05:17
Monica Ionita and Klaus Grosfeld
Alfred Wegener Institute for Polar and Marine Research
Yes automatically include my contributions in July and August 2017
Sea ice in both Polar Regions is an important indicator for the expression of global climate change and its polar amplification. Consequently, a broad information interest exists on sea ice, its coverage, variability and long term change. Knowledge on sea ice requires high quality data on ice extent, thickness and its dynamics. As an institute on polar research we collect data on Arctic and Antarctic sea ice, investigate its physics and role in the climate system and provide model simulations on different time scales. All this data is of interest for science and society. In order to provide insights into the potential development of the seasonal signal, we developed a robust statistical model based on ocean heat content, sea surface temperature and atmospheric variables to calculate an estimate of the September minimum sea ice extent
for every year.
StatisticalNASA Team, May 2017Monica.Ionita@awi.de
The forecast scheme for the September sea ice extent is based on a methodology similar to one used for the seasonal prediction of river streamflow. The basic idea of this procedure is to identify regions with stable teleconnections between the predictors and the predictand. The September sea ice extent has been correlated with the potential predictors (e.g. ocean heat content, sea surface temperature, sea level pressure, precipitable water content, surface zonal and meridional wind) from previous months, up to 8 months lag, in a moving window of 21 years.
4.7419.21
pan-Arctic: lower bound 4.23, upper bound 5.25
pan-Antarctic: lower bound 18.74, upper bound 19.68
26
7/12/2017 22:54:44UTokyo (Kimura et al.)
Noriaki Kimura (The University of Tokyo), Hiroyasu Hasumi (The University of Tokyo)
Yes automatically include my contributions in July and August 2017
Monthly mean ice extent in September will be about 4.79 million square kilometers. Our estimate is based on a statistical way using data from satellite microwave sensor. We used the ice thickness in December and ice movement from December to May. Predicted ice concentration map from July to September is available in our website: http://ccsr.aori.u-tokyo.ac.jp/~kimura_n/arctic/2017e.html
Sea ice in the Laptev Sea and Chukchi Sea is expected to be thin and retreat quickly. On the other hand, sea ice in the East Siberian Sea will retreat slowly with nearly same speed as normal years. On the Canadian side, sea ice in the Beaufort Sea is expected to be thick and retreat slowly compared with the last year.
Statistical
SIT dataset distributed by distributed by Arctic Data archive System (ADS, https://ads.nipr.ac.jp/index.html), December 1 of all AMSR-E/AMSR2 years. This SIT is calculated by an algorithm of Krishfield et al. (2014).
kimura_n@aori.u-tokyo.ac.jp
We predicted the Arctic sea-ice cover from coming July 1 to November 1, using the data from satellite microwave sensors, AMSR-E (2002/03-2010/11) and AMSR2 (2012/13-2016/17). The analysis method is based on our recent research (Kimura et al., 2013). First, we expect the ice thickness distribution in May 31 from redistribution (divergence/convergence) of sea ice during December 1 and May 31, based on the daily ice velocity data. Then, we predict the summer ice area depending on the assumption that thick ice remains later and thin ice melts sooner than the average.
For this analysis, we distributed particles homogeneously over the Arctic sea ice on December 1. We traced the trajectories of the particles to the end of May by using the satellite derived daily ice velocity (Kimura Dataset). Based on the relationship between particle density on May 31 and ice concentration in summer, we predicted the summer sea ice cover of this year. We also take it into account that thickness of sea ice on December 1 calculated by an algorithm of Krishfield et al. (2014) .
4.79
27
7/13/2017 10:45:01
AWI CryoSat Team (S. Hendricks)
Stefan Hendricks, Alfred Wegener Institute
Robert Ricker, Alfred Wegener Institute
Christian Haas, Alfred Wegener Institute
Yes automatically include my contributions in July and August 2017
We assess sea ice thickness data from CryoSat observations in the Central Arctic Basin and compare the result to previous years. We find an ongoing decline of sea ice volumen for the past 4 years, however 2017 marks only the 3rd lowest volume of the CryoSat data record in the central Arctic basin. Below average thicknesses in the Chukchi/Beaufort Sea and multiyear ice region as well as above average thicknesses between the north pole and the Chukchi Sea as well north of Spitsbergen contribute the 2017 volume result. Where available, validation data shows little bias between the CryoSat and airborne thickness estimates.
No outlook, assessment of remote sensing data
n/a
No initialization, but SIT data from AWI CryoSat sea ice data product (version 1.2)
stefan.hendricks@awi.de
n/aNo outlookn/an/an/a
28
7/13/2017 10:57:50Slater/Barrett NSIDC
Drew Slater, Andrew Barrett, Matt Savoie, Trey Stafford, National Snow and Ice Data Center
No do not use my prediction this month in later months
This projection was made using the Slater Probabilistic Ice Extent model developed by Drew Slater (http://cires1.colorado.edu/~aslater/SEAICE/). The model computes the probability of sea ice concentration greater than 15% for Arctic Ocean grid cells in the EASE 25 km grid. These probabilities are aggregated over the model domain to arrive at daily ice extents. A September mean ice extent is calculated from daily forecasts issued on July 1. While the model has predictive skill at lead times up to 90 days, NSIDC runs the forecast model with a 50 day lead time. Forecasts issued on July 1 for September have lead times spanning 63 to 92 days. Therefore we consider the mean September ice extent forecast for the July sea ice outlook to have some skill.
StatisticalNASA Team June 2017Noneapbarret@nsidc.org
This is a non-parametric statistical model of Arctic sea ice extent. The model computes the probability of whether ice concentration greater than 15% will exist at a particular location for a particular lead time into the future, given current ice concentration. The only input is sea ice concentration. Probabilities are computed using data from the past 10 years. These probabilities are adjusted using daily near-real-time concentrations to make a forecast. Pan-Arctic Ice extent is the sum of the product of grid-box area the probability of a grid-box containing ice on the forecast date.
While not as sophisticated as a coupled ocean-ice-atmosphere models, this statistical method has the advantage that the forecasts for all points are completely independent in both space and time; that is, the forecast at any given point is not affected by its neighbors, nor its result from the prior day. Therefore, the model can adapt to changing conditions and is not inherently subject to drift.
4.82
29
7/13/2017 11:04:20
AWI consortium (Kauker et al.)
F. Kauker (AWI and OASys, frank.kauker@awi.de), T. Kaminski (ILab, thomas.kaminski@inversion-lab.com), R. Ricker (AWI, Robert.Ricker@awi.de), L. Toudal-Pedersend (EOLab, elmltp@gmail.com), G. Dybkjaerd (DMI, gd@dmi.dk), C. Melsheimer (Univ Bremen, melsheimer@uni-bremen.de), S. Eastwood (The Norwegian Meteorological Institute, s.eastwood@met.nof, H. Sumata (AWI,
hiroshi.sumata@awi.de), M. Karcher (AWI and OASys, michael.karcher@awi.de), R. Gerdes (AWI, ruediger.gerdes@awi.de)
No do not use my prediction this month in later months
Sea ice-ocean model ensemble prediction initialised with assimilation of sea ice and ocean observations.
Dynamic Model
OSI SAF EUMETSAT OSI-401 March and April 2017
CryoSat-2 from Alfred-Wegener Institute of March and April 2017
frank.kauker@awi.de
NAOSIM sea ice - ocean model. Initialized on 11 July 2017 from a hindcast run (started 1.1.1980) with assimilation of SIT (AWI), SIC and SST (OSI SAF) and snow depth (Univ Bremen) in March and April 2017. Model biases are compensated by a correction to SIT (estimated from the years 2012 to 201
For the present outlook the coupled ice-ocean model NAOSIM has been forced with atmospheric surface data from January 1948 to July 11th 2017 ombination of NCEP/NCAR and NCEP-CFSR and NCEP CFSv2). All ensemble model experiments have been started from the same initial conditions on July 11th 2017. The model setup has not changed with respect to the last year. We used atmospheric forcing data from each of the years 2007 to 2016 for the ensemble prediction and thus obtain 10 different realisations of potential sea ice evolution for the summer of 2017. The use of an ensemble allows to estimate probabilities of sea-ice extent predictions for September 2017. A variational assimilation system around NAOSIM has been used to initialize the model using the Alfred Wegener Institute's CryoSat-2 ice thickness product, the University of Bremen's snow depth product, and the OSI SAF ice concentration and sea-surface temperature products. Observations from March and April were used. A bias correction scheme for the CryoSat-2 ice thickness which employs a spatially variable scaling factor could enhance the skill considerably (Kauker et al, 2015, http://www.the-cryosphere-discuss.net/tc-2015-171/).
4.93
Standard deviation of 0.27 million square kilometers
Ensemble spread of the forcing years 2007 to 2016 used by the sea ice - ocean model (from 11 July to end of September).
The September mean sea ice extent is calculated from the simulated sea ice concentration. A tiny correction (-0.02 mill. km2) is applied to the sea ice extent (calculated from a hindcast run).
30
7/13/2017 14:32:33NMEFC (Li & Li )
Chunhua Li, Ming Li /National Marine Environmental Forecasting Center(NMEFC),China
We predict the September monthly average sea ice extent of Arctic by statistic method and based on monthly sea ice concentration and extent from National Snow and Ice Data Center. The result shows that the Sep. ice extent will be less slightly in 2017 than in 2016.
Statistical
Sea Ice Index - Daily and monthly sea ice concentration and extent from National Snow and Ice Data Center.
bitz@uw.edu
A simple statistical model is used to predict September monthly Arctic sea ice extent. We find that the sea ice extent of September is well related with the sea ice extent of Jan. to Apr. in the same year, and the ice decreasing trend during Jan. to Apr. of 2017 is similar to the trend of 2016, so we assume that the ice extent will decreasing with the same rate of 2016 .Combined the multiple regression method and optimal climate normal method, the predicted September sea ice extent in 2017 is 4.69 million square kilometers.
4.69
A simple statistical model is used to predict September monthly Arctic sea ice extent. We find that the sea ice extent of September is well related with the sea ice extent of Jan. to Apr. in the same year, and the ice decreasing trend during Jan. to Apr. of 2017 is similar to the trend of 2016, so we assume that the ice extent will decreasing with the same rate of
2016 .Combined the multiple regression method and optimal climate normal method, the predicted September sea ice extent in 2017 is 4.69 million square kilometers.
31
7/13/2017 17:37:51RASM(Kamal et al.)
Our RASM Team includes the following people:
1. Samy Kamal, primary contact, Naval Postgraduate School
2. Wieslaw Maslowski, Naval Postgraduate School
3. Robert Osinski, Institute of Oceanology, Polish Academy of Sciences
4. Andrew Roberts, Naval Postgraduate School
5. Mark Seefeldt, University of Colorado
6. John Cassano, University of Colorado
Yes automatically include my contributions in August 2017
We used the Regional Arctic System Model (RASM), which is a limited-area, fully coupled climate model consisting of the Weather Research and Forecasting (WRF) model, Los Alamos National Laboratory (LANL) Parallel Ocean Program (POP) and Sea Ice Model (CICE) and the Variable Infiltration Capacity (VIC) land hydrology model (Maslowski et al. 2012; Roberts et al. 2014; DuVivier et al. 2015; Hamman et al. 2016; Hamman et al. 2017; Cassano et al. 2017). WRF and VIC are configured on the same polar stereographic grid at 50-km resolution, and POP and CICE are sharing a rotated spherical grid at 1/12o (~9 km). Reanalysis data are used to force RASM From Sept 1979 to July 2017, after that the NCEP version 2 Coupled Forecast System model (CFSv2) Operational Forecasts are used for forecast forcing.
Dynamic Model
Initial Sea ice conditions are RASM-produced from multi-decadal simulations, from 1979 through June 2017.
Same as #7smkamal@nps.edu
Model name: Regional Arctic System Model (RASM).
Atmospheric model: WRF.
Ocean model: POP.
Ice model: CICE.
Land hydrology: VIC.
We used a 12-member ensemble to determine the Sept 2017 arctic sea ice status. The forecast ensemble utilized three different sea ice initial states produced by three different RASM 1979-2017 simulations (root cases). Root cases are forced with reanalysis data and ran from Sept 1979 to July 2017. After that, each set of initial conditions is used to run 4 forecast cases forced with four different versions of CFSv2 seasonal forecast data and ran from July 1st to Sept 30th 2017. The CFSv2 data used were initialized on Jun 17th at hours: 0000, 0600, 1200 and 1800 but each ensemble was forced at the same time, starting July 1st at 0000.
4.110.191
We estimate our uncertainty to be equal to the spread in September mean extent among all the ensemble members.
Alaskan regional sea ice extent is 0.0880 million square kilometer. Alaskan region total area is 4.012 million square kilometers.
32
7/13/2017 15:28:39Dmitri Kondrashov
University of California, Los Angeles (UCLA)
No do not use my prediction this month in later months
This contribution relies on data-harmonic analysis techniques to predict sea ice conditions over the Pan-Arctic region. The prediction model is obtained by data-adaptive harmonic decomposition and stochastic inverse modeling of Multisensor Analyzed Sea Ice Extent – Northern Hemisphere (MASIE-NH) dataset, as well as regional sea ice extent (SIE) from Sea Ice Index (SII) dataset.
Statistical
dkondras@atmos.ucla.edu
The forecasting methodology relies on Data-adaptive Harmonic Decomposition (DAH) and Multilayer Stuart-Landau Models (MSLM) techniques [Chekroun and Kondrashov, 2017]. This methodology is applied to the Multisensor Analyzed Sea Ice Extent – Northern Hemisphere (MASIE-NH) and regional SIE from SII dataset, combined into several key Arctic regions. The daily MASIE-NH and SIE data were aggregated to provide weekly-sampled dataset. DAH-MSLM predictive model has been derived from SIE anomalies with annual cycle removed. The key features of DAH-MSM model are memory effects conveyed by the non-Markovian model formulation and data-adaptive basis that helps to disentangle complex regional dynamics of by harmonic spatio-temporal patterns. The stochastic DAH-MSLM model is driven from latest initial conditions of SIE by ensemble of white noise realizations to provide probabilistic regional Arctic forecasts, as well as pan-Arctic ones.

[1] Kondrashov, D., M. Chekroun, and M. Ghil, 2015: Data-driven non-Markovian closure models. Physica D., 297, 33–55.

[2] Chekroun, M., and D. Kondrashov, 2017: Data-adaptive Harmonic Spectra and Multilayer Stuart-Landau Models, HAL preprint, https://hal.archives-ouvertes.fr/hal-01537797
4.50.42
33
7/13/2017 17:22:57Lamont (Yuan et al.)
Xiaojun Yuan (primary contact), Cuihua Li and Lei Wang
A Linear Markov model is used to predict monthly Arctic sea ice concentration (SIC) at all grid points in the pan Arctic region. The model is a stochastic linear inverse model that is built in the multi-EOF space and is capable to capture the co-variability in the ocean-sea ice-atmosphere system. September pan Arctic sea ice extent is calculated from predicted sea ice concentration. The model predicts negative SIC anomalies throughout the pan Arctic region. Large negative sea ice concentration anomalies (larger than -40%) will occur in the Chukchi Sea, E. Siberian Sea and Laptev Sea; and moderate negative anomalies (-24% to -40%) in the Kara Sea in September 2017. Negative SIC anomalies in other areas will be smaller than 24%. These anomalies are relative to the 1979-2012 climatology. The September mean pan Arctic sea ice extent (SIE) is predicted 5.14 million squared kilometers with a RMSE of 0.61 million square kilometers. The September mean pan Antarctic SIE is predicted to be 20.41 million square kilometers with RMSE of 1.75 million square kilometers by a similar linear Markov model.
Statistical
We used NASA Team sea ice concentration in June 2917 to initialize the prediction.
N/A
xyuan@ldeo.columbia.edu
N/A
The linear Markov model has been developed to predict sea ice concentrations in the pan Arctic region at the seasonal time scale. The model employs 3 variables: NASA Team sea ice concentration, sea surface temperature (ERSST), and surface air temperature (NCEP/NCAR reanalysis) for the period of 1979 to 2012. It is built in multi-variate EOF space. The model utilizes first 11 mEOF modes and use a Markov process to predict these principal components forward one month at a time. The pan Arctic sea ice extent forecast is calculated by summarizing all cell areas where predicted sea ice concentration exceeds 15%. Bias corrections have been applied to ice concentration predictions at grid points as well as the total sea ice extent prediction. The predictive skill of the model was evaluated by anomaly correlation between predictions and observations, and root-mean-square errors (RMSE) in a (take one-year out) cross-validated fashion. On average, the model is superior to the predictions by anomaly persistence, damped anomaly persistence and climatology (Yuan et al, 2016). For the three-month lead prediction of September sea ice concentrations, the model has higher skill (anomaly correlation) and lower RMSE in the Chukchi Sea and Beaufort Sea than in other regions. The skill of the three-month lead prediction of the pan Arctic sea ice extent in September is 0.78 with a RMSE of 0.61 million squared kilometers.
5.1420.41
The forecast uncertainty is measured by RMSE. The RMSE of the September Arctic SIE 3-month lead prediction is 0.61 million square kilometers. The RMSE of the September Antarctic SIE 3-month lead prediction is 1.75 million square kilometers.
Cross-validated experiments of 3-month lead forecasts for September Arctic SIC were performed over 34 years with bias correction at grid points. The September Arctic SIE was then calculated from the SIC predictions and predicted SIE bias was corrected through a linear regression fit of prediction errors as function of initial conditions. The bias between model grid and observation grid was then corrected. The uncertainty (RMSE) was calculated based on on 34 years of predicted and observed September SIEs. For the Antarctic SIE prediction, the RMSE was estimated from the errors of the last ten years of forward predictions and observations, which is 1.75 million square kilometers.
The predictions of Arctic SIC anomalies were corrected for biases at grid points and combined with climatology to generate SIC forecasts. The SIE forecast was then constructed (area of SIC > 15%) from predicted SIC and SIE biases were corrected to form the final prediction.
34
7/13/2017 20:26:09
Antarctic Gateway Partnership (Dr. Laura Davies), University of Tasmania, Australia
Dr Laura Davies of Antarctic Gateway Partnership, University of Tasmania, Australia
Yes automatically include my contributions in August 2017
I have been investigating Antarctic sea ice growth and I was experimenting with whether I could make a forecast using some of the results.
StatisticalNASA Team, May 2017
laura.davies@utas.edu.au
I used historical sea ice growth rates to predict the September mean.
18.4
35
7/14/2017 9:09:21Met Office
K. Andrew Peterson (primary), C. MacLachlan, E.W. Blockley, A.A. Scaife, Met Office, FitzRoy Road, Exeter, UK
Yes automatically include my contributions in August 2017
Using the Met Office GloSea5 seasonal forecast systems we are issuing a model based mean Northern Hemisphere September sea ice extent outlook of (4.0 +/- 0.8) million sq. km. This has been assembled using startdates between 30 May
and 19 June to generate an ensemble of 42 members. The Southern Hemisphere September sea ice extent outlook using the same start dates is (17.6 +/- 0.7) million sq. km.
Dynamic Model
Met Office Forecast Ocean As-similation Model (FOAM) ocean and sea ice analysis [Blockley et al., 2014] using the SSMIS brightness temperature observations of sea ice concentration product of the EU- METSAT Ocean and Sea Ice Satellite Application Facility (OSI-SAF, www.osi-saf.org) [OSI-SAF].
Met Office FOAM ocean and sea ice analysis [Blockley et al., 2014] as evolved by model dynamics and thermodynamics. No assimilation of thickness observations is performed.
drew.peterson@metoffice.gov.uk
HadGEM3 -- GC2 in use within the GloSea5 seasonal prediction system

Component : Name : Initialization
Sea Ice: CICE -- GSI6.0 : FOAM ocean and sea ice analysis
Ocean: NEMO -- Global Ocean 5.0 : FOAM ocean and sea ice analysis
Atmosphere: Met Office Unified Model, UM -- GA7.0 : NWP 4DVar
Land: JULES -- GL6.0 : Soil temperature and snow - NWP 4DVar, Soil Moisture - climatology
Ensemble coupled model seasonal forecast from the GloSea5 seasonal pre-
diction system [MacLachlan et al., 2015], using the Global Coupled 2 (GC2) version [Williams et al., 2015] of the HadGEM3 coupled model [Hewitt et al., 2011]. Forecast compiled together from forecasts initialized between 30 May and 19 June (2 per day) from an ocean and sea ice analysis (FOAM/NEMOVAR) [Blockley et al., 2014, Peterson et al., 2015] and an atmospheric analysis (MO-NWP/4DVar) [Rawlins et al., 2007] using observations from the previous day. Special Sensor Microwave Imager Sensor (SSMIS) ice concentration observations from EUMETSAF OSI-SAF [OSI-SAF] were assimilated in the ocean and sea ice analysis, along with satellite and in-situ SST, sub-surface temperature and salinity profiles, and sea level anomalies from altimeter data. No assimilation of ice thickness was performed.
417.6
0.8 million sq km for Arctic. 0.7 million sq km for Antarctic. This represents two standard deviations of the (42 member) ensemble spread around the
ensemble mean.
Validation of the forecast was done using our 1993-2015 historical re-forecast (hindcast) using startdates of 01, 09 and 17 June (7 members each). Root mean square error of the hindcast with observations was 0.4 million sq. km for both the Arctic and Antarctic, with a insignificant bias of 400 square km above observed climatology (the Antarctic basis was somewhat larger, but still trivial, at 0.02 million sq. km). This is consistent with the error measurement based on the ensemble spread given above. Over the hindcast period, the correlation between the GloSea5 forecast and NSIDC sea ice extent observations was 0.89 (Arctic, Antarctic 0.54) which reduces to a correlation of 0.65 (Arctic, Antarctic: 0.64) if the trend is removed from the time series.

None, as over the 1993-2015 hindcast, there is a insignificant forecast
bias of 400 sq km above the observed sea ice extent climatology. The Antarctic basis was some what larger, but still trivial, at 0.02 million sq. km below observed climatology.
36
7/13/2017 15:05:15Barthélemy et al.
Antoine Barthélemy (antoine.barthelemy@uclouvain.be)
François Massonnet (francois.massonnet@uclouvain.be)
Hugues Goosse (hugues.goosse@uclouvain.be)
Thierry Fichefet (thierry.fichefet@uclouvain.be)

Georges Lemaître Centre for Earth and Climate Research (TECLIM), Earth and Life Institute (ELI), Université catholique de Louvain (UCL), Louvain-la-Neuve, Belgium
No do not use my prediction this month in later months
Our estimate is based on results from ensemble runs with the global ocean-sea ice coupled model NEMO-LIM3. Each member is initialized from a reference run on June 30, 2017, then forced with the NCEP/NCAR atmospheric reanalysis from one year between 2007 and 2016. Our final estimate is the ensemble median, and the given range corresponds to the lowest and highest extents in the ensemble.
Dynamic Model
Initial sea ice concentrations come from a model free run on June 30.
Initial sea ice thicknesses come from a model free run on June 30.
antoine.barthelemy@uclouvain.be
Ocean - NEMO 3.5 - initialized on June 30 from a free run
Sea ice - LIM3 - initialized on June 30 from a free run
Our estimate is based on results from ensemble runs with the global ocean-sea ice coupled model NEMO-LIM3. The ensemble members are expected to sample the atmospheric variability that may prevail this summer. In practice, the model is forced with NCEP/NCAR atmospheric reanalysis data from 1948 to June 30, 2017. No data are assimilated during this simulation. Ten ensemble members are then started from the obtained model state, each using atmospheric forcing from one year between 2007 and 2016. This choice is a compromise between a sufficiently large ensemble and the rapidly changing Arctic atmospheric conditions in recent decades. The estimate given above corresponds to the ensemble median monthly September extent, corrected by the mean bias between simulated and observed values reported in the NSIDC sea ice index, which equals 0.6 million square kilometers. The model configuration is exactly the same as in our last four years contributions. Additional details can be found in our 2013 reports.
3.618.1
Projection uncertainty range:
- Arctic: from 2.4 to 4.2 million square kilometers
- Antarctic: from 17.5 to 18.6 million square kilometers
The projection uncertainty is given as the range between minimum and maximum extents in the ensemble. Although relatively wide, this neglects potential erroneous initial state and model errors not accounted for through the mean bias correction. It is based solely on the uncertainties arising from atmospheric variability, and on the hypothesis that the 2017 atmospheric summer conditions will be similar to the ones observed during the last decade.
The September sea ice extents from the ensemble members have been corrected by the mean bias between the observed (from the NSIDC sea ice index) and simulated (by the reference NEMO-LIM3 run) extents over 2007-2016, which equals 0.6 million square kilometers.
-
37
7/3/2017 16:54:21FIO-ESM (Qiao et al.)
Fangli Qiao (First Institute of Oceanography, State Oceanic Administration, China)
Qi Shu (First Institute of Oceanography, State Oceanic Administration, China)
Zhenya Song (First Institute of Oceanography, State Oceanic Administration, China)
Xunqiang Yin (First Institute of Oceanography, State Oceanic Administration, China)
Ying Bao (First Institute of Oceanography, State Oceanic Administration, China)
No do not use my prediction this month in later months
Our prediction is based on FIO-ESM (the First Institute of Oceanography-Earth System Model) with data assimilation. The prediction of September pan-Arctic extent in 2017 is 4.45 (+/-0.49) million square kilometers. 4.45 and 0.49 million square kilometers are the average and one standard deviation of 10 ensemble members, respectively.
Dynamic Modelshuqiemail@163.com
Atmosphere CAM3 1992-2017 integration
Ocean POP2 DA – EAKF DA system
Ice CICE4 1992-2017 integration
Wave MASNUM-wave model 1992-2017 integration
This is a model contribution. The initialization is also from the same model (FIO-ESM) but with data assimilation. The data assimilation method is Ensemble Adjustment Kalman Filter (EAKF). The data of SST (sea surface temperature) and SLA (sea level anomaly) from 1 January 1992 to 1 July 2017 are assimilated into FIO-ESM model to get the initial condition for the prediction of the Arctic Sea Ice. There is no sea ice data assimilation.
4.45
Our prediction is 4.45 (+/-0.49) million square kilometers based on 10 ensemble members. 4.45 and 0.49 million square kilometers are the average and one standard deviation of these 10 ensemble members, respectively.
38
7/12/2017 5:52:54CNRM (Chevallier et al)
Matthieu Chevallier, Constantin Ardilouze, Lauriane Batté (CNRM, Meteo France, Toulouse, France)
No do not use my prediction this month in later months
CNRM outlook is based on the operational seasonal forecast issued by Météo France in early July 2017 with the System 5 (component of the European multi-model EUROSIP).
Dynamic Model
Initial conditions for the ocean and sea ice (concentration and thickness) are provided by Mercator Océan. Basis is the Mercator Océan operational analysis, run at a 1/4° horizontal resolution using NEMO-LIM2 and the SAM ocean data assimilation system. There is no data assimilation of sea ice concentration in this analysis. The 1/4° analysis is upscaled to the 1° horizontal grid of CNRM-CM. These fields are used to nudge the ocean-sea ice component of CNRM-CM (NEMO3.2-Gelato5, 1° resolution), run in forced mode (forced by ECMWF Op. analysis). A strong restoring is applied towards Mercator SST, which acts as a constraint on sea ice concentration. Sea ice fields (concentration, thickness…) from this 1° run are used to initialize CNRM-CM (as well as ocean fields from this run).
Sea ice thickness information is output from the 1° simulation described above.
matthieu.chevallier@meteo.fr
Météo France System 5
Component Name Initialization
Atmosphere ARPEGE-Climat v6 ECMWF Operational Analysis
Ocean NEMO3.2 Based on Mercator (DA SEEK)
Sea ice GELATO 5 Based on Mercator (no DA)

Basis of System 5 is the global coupled model CNRM-CM5 (Voldoire et al., 2014).

Horizontal resolution of the atmosphere component is T255 (70km). Nominal resolution of the ocean-sea ice component is 1° at the equator (nearly 50km in the Arctic). The sea ice model uses 4 sea ice categories (0-0.3; 0.3-0.8, 0.8-3 and >3m).
The CNRM outlook is a model estimate based on a dynamical ensemble forecast. Initial conditions from the week before 1 July 2017 are used. We generate a total 51 members by adding statistical perturbations during the run.
3.11 (STD 0.55)18.37 (STD 0.29)
For pan-Arctic SIO
Median : 2.99
25% : 2.68
75% : 3.48
Min : 2.29
Max : 4.42

For pan-Antarctic SIO
Median : 18.41
25% : 18.23
75% : 18.57
Min : 17.42
Max : 18.85

Statistics are based on the 51-member ensemble.
Standard deviation of the ensemble is 0.55 million km2 for Arctic, 0.29 million km2 for Antarctic.
Post-processing includes bias correction and correction of the (linear) trend based on the hindcast only (1993-2014).
39
7/10/2017 15:44:30
GFDL/NOAA, Bushuk et al.
Mitch Bushuk (GFDL/NOAA)
Rym Msadek (CNRS/CERFACS)
Mike Winton (GFDL/NOAA)
Gabe Vecchi (Princeton University)
Rich Gudgel (GFDL/NOAA)
Tony Rosati (GFDL/NOAA)
Xiaosong Yang (GFDL/NOAA)
Matt Harrison (GFDL/NOAA)
Tom Delworth (GFDL/NOAA)
No do not use my prediction this month in later months
Our July 1 prediction for the September-averaged Arctic sea-ice extent is 4.21 million square kilometers, with an uncertainty range of 3.58-4.61 million square kilometers. Our prediction is based on the GFDL-FLOR ensemble forecast system, which is a fully-coupled atmosphere-land-ocean-sea ice model initialized using a coupled data assimilation system. Our prediction is the bias-corrected ensemble mean, and the uncertainty range reflects the lowest and highest sea ice extents in the 12-member ensemble.
Dynamic ModelNoneNone
mitchell.bushuk@noaa.gov
Model: GFDL-FLOR

Component Name Initialization (e.g., describe Data Assimilation)
Atmosphere AM2.5 AMIP run forced with observed SST/sea ice
Ocean MOM4 EnKF coupled data assimilation
Sea Ice SIS1 EnKF coupled data assimilation (no ice data)
Our forecast is based on the GFDL Forecast-oriented Low Ocean Resolution (FLOR) model (Vecchi et al., 2014), which is a coupled atmosphere-land-ocean-sea ice model. The model is initialized from an Ensemble Kalman Filter coupled data assimilation system (ECDA; Zhang et al., 2007), which assimilates observational surface and subsurface ocean data and atmospheric reanalysis data. The system does not assimilate any sea ice concentration or thickness data. The FLOR atmospheric initial conditions are produced from an AMIP run forced by observed SST and sea ice. Historical radiative forcing is used prior to 2005 and the RCP4.5 scenario is used for predictions after 2005. For the predictions initialized after 2004, the aerosols are fixed at the RCP4.5 scenario year of 2004. The performance of this model in seasonal prediction of Arctic sea ice extent has been documented in Msadek et al. (2014) and Bushuk et al. (2017). For an evaluation of the model's September sea ice extent prediction skill from a July 1 initialization, see Section 2 below.

References:

Bushuk, M., R. Msadek, M. Winton, G. Vecchi, R. Gudgel, A. Rosati, and X. Yang, 2017: Skillful regional prediction of Arctic sea ice on seasonal timescales. Geophys. Res. Lett., 44.

Msadek, R., G. Vecchi, M. Winton, and R. Gudgel, 2014: Importance of initial conditions in seasonal predictions of Arctic sea ice extent. Geophys. Res. Lett., 41 (14), 5208-5215.

Vecchi, G. A., et al., 2014: On the seasonal forecasting of regional tropical cyclone activity. J. Climate, 27 (21), 7994-8016.

Zhang, S., M. Harrison, A. Rosati, and A. Wittenberg, 2007: System design and evaluation of coupled ensemble data assimilation for global oceanic climate studies. Mon. Wea. Rev., 135 (10), 3541-3564.
4.21
Median: 4.21
Range: 3.58-4.61
St. dev.: 0.26
These statistics are computed using our 12 member prediction ensemble.
These forecasts are bias corrected based on an additive correction using a
suite of retrospective forecasts spanning 1980-2016. The bias is defined as the
September sea ice extent difference between NSIDC NASA team observations
and forecasts initialized on July 1.
0.10

Maximum possible extent: 3.80 million square kilometers
40
7/13/2017 13:00:05
MPAS-CESM (Cavallo, Szapiro, and Skamarock)
Steven Cavallo, University of Oklahoma. Nicholas Szapiro, University of Oklahoma. Bill Skamarock, NCAR
Yes automatically include my contributions in July and August 2017
Our July 1 predictions are the means and standard deviations from an ensemble of forecasts from June 30th using MPAS-CESM as a fully-coupled atmosphere-land-ocean-sea ice model. The control member is initialized from GFS analysis for the atmosphere and a restart from member 20 of the CESM Large Ensemble for the other components. An 11 member initial condition ensemble is constructed in 2 sets: a perturbed atmosphere (GEFS members 1-5) and perturbed other components (with Large Ensemble members 5-9).
Dynamic Model
No external SIC is used. CESM Large Ensemble members are used for initial conditions.
No external SIT is used. CESM Large Ensemble members are used for initial conditions.
nick.szapiro@ou.edu
Component | Name | ICs
Atmosphere | CAM5-MPAS | GFS FNL (and GEFS) for 2017-06-30 00Z
Ocean | POP2 | CESM Large Ensemble restart
Sea ice | CICE4 | CESM Large Ensemble restart
Land | CLM4 | CESM Large Ensemble restart
River | RTM | CESM Large Ensemble restart
For our experimental forecast, we use CAM-MPAS on an Arctic-refined (92-25 km) atmospheric mesh coupled to 1 degree land, ocean, and sea ice in CESM. For the control member, the atmosphere is cold-started with GFS analysis and the other components are initialized from a spun-up analog member of the CESM Large Ensemble. An 11 member initial condition ensemble is constructed in 2 sets: a perturbed atmosphere (GEFS members 1-5) and perturbed other components (with Large Ensemble members 5-9). Forecasts are integrated as if in 2021 under an RCP8.5 scenario to match the Large Ensemble's climatological trend for September SIE for 2017. September time mean SIC is used directly with a 15% threshold for extent.
4.518.1
Standard deviations:
Arctic: .5
Antarctic: .3
Alaska: .1
Standard deviation across the individual ensemble members
0.4 +/- .1
The September mean concentration has been nearest-neighbor interpolated to the NSIDC masked grid and converted to extent with a 15% threshold. Maximum possible coverage is 4 M km^2.
41
7/18/2017 12:12:51Canadian Ice Service
Adrian Tivy - Canadian Ice Service, Environment Canada
Yes automatically include my contributions in later months 2017
Environment Canada’s Canadian Ice Service (CIS) is predicting the 2017 minimum Arctic sea extent at 4.0* 106 km2. As with previous CIS contributions, the 2017 forecast was derived by considering a combination of methods: 1) a qualitative heuristic method based on observed end-of-winter Arctic ice thickness/extent, as well as winter surface air temperature, spring ice conditions and the summer temperature forecast; 2) two simple statistical methods based on an Optimal Filtering Based Model (OFBM), that uses an optimal linear data filter to extrapolate the September sea ice extent time-series into the future and 3) a Multiple Linear Regression (MLR) prediction system that tests ocean, atmosphere and sea ice predictors.
Mixedbitz@uw.edu
Based on winter air temperatures and sea ice extents and thickness, a September 2015 minimum ice extent value of 3.8*10^6 km2 is heuristically predicted. The CIS OFB models predict 4.24*10^6 km2 and 3.78*10^6 km2 and the CIS MLR model predicts 4.2 *10^6 km2. The average forecast value of the four methods combined is 4.0*10^6 km2
4
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
Loading...
 
 
 
Form Responses 1