Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Long-Range Statistical Forecasting of Ice Severity in the Beaufort–Chukchi Sea

Long-Range Statistical Forecasting of Ice Severity in the Beaufort–Chukchi Sea Interannual variations in Beaufort Sea summer ice conditions influence a wide range of socioeconomic activities, including merchant shipping in the Beaufort Sea and subsistence lifestyles on the Alaskan North Slope. Each year, the National Ice Center quantifies Beaufort Sea summer ice conditions based on the Barnett severity index (BSI), which is based on distances from Point Barrow, Alaska, to the sea ice edge, as well as characteristics of the shipping season from Prudhoe Bay to the Bering Sea. Long-range forecasts (monthly to seasonal) of the BSI would be valuable for the above-mentioned users, provided that the forecasts are communicated effectively and used properly. Utilizing mean monthly sea ice and atmospheric data from 1979 through 2000, multiple linear regression models are developed here to forecast the BSI at monthly intervals from October of the previous year through to July of the prediction year. The final models retain between three and five variables, with decreased multiyear sea ice (MYI) and total sea ice (CT) concentration in the Beaufort Sea, and increased MYI in the transpolar drift stream leading to less severe summer ice conditions. Variations in antecedent autumn and spring wind patterns associated with the October east Atlantic index and the March North Atlantic Oscillation index also play a key role in defining the ensuing summer's ice severity, with fluctuations in July heating degree days being somewhat valuable. Monte Carlo simulations suggest that the final models are not adversely influenced by artificial skill, while Durbin–Watson and variance inflation factor (VIF) statistics indicate the final models are statistically valid. Model accuracy, as defined by the coefficient of determination, ranges from 0.74 with October data to 0.92 with July data. Categorical forecasts (e.g., forecasted ice conditions are ranked from heaviest to lightest) are provided as an example of effectively communicating the model output for all users, while a probability of exceedance curve is shown as an example of communicating uncertainty information to more advanced users. It is important to note that this method does not show good skill on historical (1953–78) data, likely due to a regime shift in the mid-1970s, and that if the Arctic climate changes, the methods described here will need to be altered. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Weather and Forecasting American Meteorological Society

Long-Range Statistical Forecasting of Ice Severity in the Beaufort–Chukchi Sea

Drobot, Sheldon
Weather and Forecasting , Volume 18 (6) – Jul 30, 2002
Download PDF
16 pages

Loading next page...
 
/lp/american-meteorological-society/long-range-statistical-forecasting-of-ice-severity-in-the-beaufort-MLT5UH7WdX

References (50)

Publisher
American Meteorological Society
Copyright
Copyright © 2002 American Meteorological Society
ISSN
1520-0434
DOI
10.1175/1520-0434(2003)018<1161:LSFOIS>2.0.CO;2
Publisher site
See Article on Publisher Site

Abstract

Interannual variations in Beaufort Sea summer ice conditions influence a wide range of socioeconomic activities, including merchant shipping in the Beaufort Sea and subsistence lifestyles on the Alaskan North Slope. Each year, the National Ice Center quantifies Beaufort Sea summer ice conditions based on the Barnett severity index (BSI), which is based on distances from Point Barrow, Alaska, to the sea ice edge, as well as characteristics of the shipping season from Prudhoe Bay to the Bering Sea. Long-range forecasts (monthly to seasonal) of the BSI would be valuable for the above-mentioned users, provided that the forecasts are communicated effectively and used properly. Utilizing mean monthly sea ice and atmospheric data from 1979 through 2000, multiple linear regression models are developed here to forecast the BSI at monthly intervals from October of the previous year through to July of the prediction year. The final models retain between three and five variables, with decreased multiyear sea ice (MYI) and total sea ice (CT) concentration in the Beaufort Sea, and increased MYI in the transpolar drift stream leading to less severe summer ice conditions. Variations in antecedent autumn and spring wind patterns associated with the October east Atlantic index and the March North Atlantic Oscillation index also play a key role in defining the ensuing summer's ice severity, with fluctuations in July heating degree days being somewhat valuable. Monte Carlo simulations suggest that the final models are not adversely influenced by artificial skill, while Durbin–Watson and variance inflation factor (VIF) statistics indicate the final models are statistically valid. Model accuracy, as defined by the coefficient of determination, ranges from 0.74 with October data to 0.92 with July data. Categorical forecasts (e.g., forecasted ice conditions are ranked from heaviest to lightest) are provided as an example of effectively communicating the model output for all users, while a probability of exceedance curve is shown as an example of communicating uncertainty information to more advanced users. It is important to note that this method does not show good skill on historical (1953–78) data, likely due to a regime shift in the mid-1970s, and that if the Arctic climate changes, the methods described here will need to be altered.

Journal

Weather and Forecasting – American Meteorological Society

Published: Jul 30, 2002

There are no references for this article.