North-Central Section - 57th Annual Meeting - 2023

Paper No. 8-7
Presentation Time: 8:00 AM-12:00 PM

EVALUATING METHODS FOR DETERMINING THERMOKARST LAKE EXTENT CHANGES OVER THE PERIOD 2001 - 2020


KLAIS, Danielle and DUCHESNE, Rocio, University of Wisconsin - Whitewater, Whitewater, WI 53190

Thermokarst lakes are one of the most evident repercussions of permafrost degradation, which can cause a host of problems, including ecosystem damage, coastal erosion, and acceleration of climate change. Permafrost is the thick mass of frozen sediment and debris held together by ice over the span of decades (even millenia) that is located in arctic climates, and it is buried beneath the active layer of soil just below the earth’s surface. Unlike the active layer of soil, permafrost does not melt in the summer and refreeze during the winter because it is consistently frozen solid. Recently, we have seen more permafrost degradation, which creates craters on the terrain and forms thermokarst lakes. This research project aims to compare multiple methods for estimating Alaskan thermokarst lake extent over the past few decades by using Landsat imagery. The focus of the project is on Deadhorse in Prudhoe Bay, Alaska, which is located in a continuous permafrost zone. One Landsat 7 image was selected for the years 2001, 2005, 2009, and 2015, while the Landsat 8 satellite imagery provided 1 image for the year 2020. The images corresponded to the peak of the growing season, July and August, all were cloudless. Images were downloaded from Earth Explorer and an area of 25 square kilometers was selected close to Deadhorse. Eleven large lakes, all of them greater than 120,000 square kilometers, were chosen. The multispectral bands 1, 2, 3, 4, 5, and 7 were combined in Erdas Imagine and pansharpened to 15 meters using the panchromatic band. The pansharpened images were then clipped to the study area. From there, three methods were used to calculate lake extent. The first method was to digitize the 11 lakes in each image using ArcGIS Pro. The second method was to perform supervised classification with ERDAS Imagine. The third method was using spectral indices to calculate lake area. The effectiveness of the three methods will be assessed and a trend analysis will follow to determine changes in the extent of thermokarst lakes during the period 2001 - 2020.