Date of Award

Spring 1-1-2011

Document Type


Degree Name

Master of Science (MS)


Aerospace Engineering Sciences

First Advisor

William J. Emery

Second Advisor

James Maslanik

Third Advisor

Cora Randall


Supraglacial melt ponds are common features of ice sheets and valuable parameters in the mass budget of the cryosphere. In addition, melt ponds are a useful proxy for monitoring global climate change as they are influenced by both the temperature of the surrounding ice and the incident radiation, which itself is influenced by the atmosphere. This document will describe an investigation of supraglacial melt ponds in a small region of the southwestern coast of the Greenland Ice Sheet, which was surveyed using an unmanned aerial vehicle in July of 2008. The data gathered during this expedition will be mined for melt ponds using Iterative Self-Organizing Data Analysis Technique, Adaptive Boosting, and Maximum Likelihood methods, and this information will be used to estimate the size and volume of the melt ponds using the known attenuation properties of water and the Beer-Lambert-Bouguer Law. Comparisons of the lake location data from UAV and satellite observations indicates that the results of the Adaptive Boosting and Maximum Likelihood algorithms are accurate to within 300 meters, or approximately ten pixels in the satellite data. The results of the lake depth analysis were inconclusive due to disagreements in the outcome when the calculations were made with different observing wavelengths and because of a lack of ground truth data. The most likely error source is the presence of suspended sediment in the lake, floating ice crystals on the lake, either of which would affect the attenuation coefficient of the water, or settled sediment on the lake bottom, which would affect the lake bottom reflectivity. Finally, attempts to develop methods to detect drained supraglacial lakes led to the promising possibility that texture analysis or observation band ratio analysis could reveal drained lake locations without the advantage of change detection. However, texture analysis proved useful only in the UAV data, which has an extremely high spatial resolution, and no correlation between lake depth and observation band ratio was observed.