Polar Science Center, Applied Physics Laboratory
University of Washingon, Seattle, Washington

 PSC logo
 APL logoUW logo

Predictions of Summer Ice Conditions near Barrow Alaska

2012 Prediction Season

Here we attempt to predict the ice conditions near Barrow Alaska as listed in Table 4 of the Seasonal Outlook For North American Arctic Waters Summer 2011 prepared by the North American Ice Service in a collaboration with the Canadian Ice Service, 02 June 2011 (http://www.natice.noaa.gov/products/nais_forecasts.html ).  The table is not available in the 2012 outlook, but NAIS provied us with the data we needed for 2011 to update the prediction model. The parameters in this table (listed below) are various measures of the amount or duration of open water near Barrow.

The predictions are based on the output of a coupled ice-ocean model which provides us with retrospective estimates of the ice and ocean conditions in the Arctic.  The model is the PIOMAS model developed and operated by Dr. Jinlun Zhang (http://psc.apl.washington.edu/zhang/IDAO/seasonal_outlook.html).   The model uses the air temperature, winds, clouds, and precipitation from the NCEP Reanalysis to estimate maps of the ice motion, ice thickness distribution, and ocean temperatures and currents for past years, up to and including the most recent month.  The observed ice concentration is assimilated so that the model ice extent is close to the observed ice extent.  Statistical relationships between the model parameters in May (or any other month) and the various measures of the ice conditions are found from past years using a method developed by Dr. Sheldon Drobot (Drobot et al, 2006).  This relationship is then used with the current month model output to predict either the pan-Arctic ice extent or any other scalar quantity such as those listed in the Ice Service outlook.   Updates are posted at http://psc.apl.washington.edu/lindsay/prediction.html

The Alaskan ice parameters predicted here are
1. Distance from Point Barrow northward to ice edge on 10 Aug (nautical miles).
2. Distance from Point Barrow northward to ice edge 15 Sept (nautical miles).
3. Distance from Point Barrow northward to boundary of five-tenths ice concentration on 10 Aug (nautical miles).
4. Distance from Point Barrow northward to boundary of five-tenths ice concentration on 15 Sept (nautical miles).
5. Initial date entire sea route to Prudhoe Bay less than/equal to five tenths ice concentration (year day).
6. Date that combined ice concentration and thickness dictate end of prudent navigation (year day).
7. Number of days entire sea route to Prudhoe Bay ice free.
8. Number of days entire sea route to Prudhoe Bay less than/equal to five-tenths ice concentration.
9. Number of days between initial opening date and 1 Oct
10. Barnett Ice Severity Index, high numbers indicate large expanses of open water.
11. Rank of the BIS index from 1953 to 2009 (1 is the highest BIS Index, 57 the lowest)

We have determined the single most effective predictor for each of these variables for each predictor month and the amount of the variance explained by the predictor.   The candidate predictors include fields of the ice thickness (H), the ice concentration (IC), the ice extent (IX, 0 or 1 for each grid cell), the fraction of the area with open water or ice less than 0.4m (G0.4m), less than 1.0 m (G1.0m), or less than 1.9 m (G1.9m).   Table 1 lists each of the Alaskan ice parameters and for each month the percent of the variance explained and the most effective predictor.  The method uses monthly averaged model output, so for predictions using the month of May, the model is run through the end of May and the fields of the average values for each predictor variable are used to make the prediction.


Predictions for 2012 from the end of May:

Parameter Predictor R2 Prediction Error
Ice_Dist_15Sep H
341 74
Ice_05_10Aug G1.0m
205 37
Ice_05_15Sep G1.9m
388 78
Date_Start H 0.38 192 (10 July) 9
Date_End IC
289 (15 Oct) 8
Ndays_Ice_Free G1.0m
81 17
Ndays_Ice_05 IC
97 18
Ndays_Start_to_1 H
81 9
BIS_Index G1.9m
913 168

These figures below show the correlation of the predictor variable with the dependent variable (what we try to predict) on the left map, the anomally of the predictore this year (middle), and the product of the first two which shows where the predictor is influential in making this years prediction. The prediction (orange star) and the errors are shown in the top figure along with the historic observations (black line) and the model fit (cyan diamonds). The trend line (dashed) and the prediction of the trend line (black star) are also shown.

Start Date

The start date for navigation is perhaps the most interesting parameter, but the uncertainty in the prediction is quite large. We predict a start date of 10 July, significantly earlier than the trend line. It is ased on the mean thickness and the region most influntial ist he eastern Beaufort Sea (right map) where the ice is quite thin (middle map).

start date june 2011



Barnett Ice Severity Index

The Barnett Ice Severity Index is predicted to be quite high (low ice conditions) at 914, but probably not a record. The area with ice more than 1.9 m thick is the best predictor, explaining 64% of the variance. Aboad region stretching from the eastern Beaufort Sea to the Kara Sea (right map) is most influential in making the prediction.

BIS from June 2011


Drobot, S. D., J. A. Maslanik, and C. F. Fowler (2006), A long-range forecast of Arctic summer sea-ice minimum extent, Geophys. Res. Lett., 33, L10501, doi:10.1029/2006GL026216

Lindsay, R. W., J. Zhang, A. J. Schweiger, and M. A. Steele, 2008a: Seasonal predictions of ice extent in the Arctic Ocean, J. Geophys. Res., 113, C02023, doi:10.1029/2007JC004259.