Recover Password  New User
Undergraduate Research Project Management System

Ice Breakup Prediction

Status Complete
Seeking Researchers No
Start Date 11/01/2009
End Date 05/01/2010
Funding Source
Funding Amount
Community Partner
Related Course CS A498
Last Updated 08/17/2014 10:34PM
Keywords ice, nenana, machine learning

People

Faculty
  Kenrick Mock

Student Researchers
  Olya Korobova

Abstract

Ice breakup dates are useful to understand climate, its processes and their impact. Accurate forecasting of ice breakup is of great importance as it helps to prevent and minimize many hazards and losses caused by ice flood or ice jams. The purpose of this project was an attempt to predict Tanana River's ice breakup date. Tanana River is located in the east-central Alaska near the village of Tanana. The date when the ice breaks on this river is the event that commemorates both the start of spring in Alaska, as well as transportation in Alaska before paved roads, trains, and planes. The Nenana Ice Classic is a popular competition to guess the date and time of the ice breakup that began in 1917 and held annually since then.

The weather data used for the research is the archive for the 60 years, beginning from 1949. The attributes that were used include the temperature, deviation point, humidity, wind speed and ice thickness. The data prior to the Nenana Ice Classic submission deadline was taken.

The explored algorithms include nearest neighbor and a collection of machine learning algorithms provided by the Weka software. During the history of the Ice Classic, the earliest calendar date the ice broke was April 20; the latest date was May 20. As of to date, the results from different algorithms do not exhibit a clear advantage in the prediction of when the ice is going to break up. However, some techniques do better than guessing the median.

Shared Project Files (e.g. papers, presentations)

File name Description Uploaded by