Forecasting ice-out in inland lakes

Ice-out on lakes marks the end of winter sports and the start of the open-water recreational season. Thus, the duration of ice cover is economically and culturally important to diverse segments of society. Across Minnesota’s 14,000+ lakes, before 2000, ice-out was fairly predictable from an observed historical ice-out date plus or minus about a week. In the last few decades, however, climate change has led to regional interannual differences of around 70 days between early and late ice-out. This is equivalent to nearly doubling the effective ice-free season. Forecasting ice-out date is important to business and industry because a poor forecast can lead to huge losses in supplies and inventory that spell the difference between profit and catastrophic losses. In this study we used a large historic database of observed ice-out dates coupled with sequential seasonal regression analysis to provide forecasts of ice-out as much as three months in advance of ice loss. The forecasts are based on lake size, latitude as a proxy for albedo, and cumulative freezing and thawing degree days measured at nearby airport weather stations. This is combined with a real-time Leaflet mapping application that uses JavaScript to pull daily climate data from the Applied Climate Information System (ACIS) Web Services JSON feed. This tool provides real-time forecast updates for fixed locations across Minnesota. These estimates serve as increasingly accurate forecasts that are widely used by the public.

Presentation Preference: Oral

Primary Presenter: John Downing, University of Minnesota (downing@d.umn.edu)

Authors:

John Downing, Minnesota Sea Grant, University of Minnesota  (downing@d.umn.edu)