Date of Award


Level of Access Assigned by Author

Open-Access Thesis

Degree Name

Master of Science (MS)


Forest Resources


William D. Ostrofsky

Second Committee Member

Steven D. Sader

Third Committee Member

Kate Beard-Tisdale


In January 1998, a severe ice storm struck the northeastern United States, causing severe injury to forested areas. Forest damage from ice storms is a result of glaze formation on twigs and branches. Ice storms are recognized as severe disturbances due to their highly destructive nature as a result of ice glaze. Researchers and landowners have been concerned that thinned stands are more susceptible to ice injury than their unthinned counterparts. Thinned stands have fewer trees per area and thus less inter-tree support. In addition, the effects of wind maybe greater in thinned stands. The objectives of this study were to investigate injury and recovery from the 1998 ice storm in thinned and unthinned hardwood stands. Four field sites were chosen and individuals in both thinned and unthinned areas were measured to determine damage and recovery values. Damage variables measured included pre- and post-storm crown class, percent crown loss and number and size of broken branches. Recovery variables included transparency rating, tree height, number and location of sprouts as well as shigometer readings for each individual. Aerial photography was used to determine ice injury using a computer-automated approach. This method consisted of rectifying and mosaicking four, digitally-scanned aerial photographs, performing an edge detection enhancement process, and using the results of this enhancement as a guide for creating computer training sites for an automatic detection of ice damage classes. The results of this method were compared to ice injury maps created from the same air photos that were analyzed using a traditional, manual approach. Results indicate that thinned stands did not suffer the effects of the 1998 ice storm greater than the unthinned stands for all four study sites combined. However, at individual sites large differences between thinned and unthinned stands were detected. Percent crown loss at site 3, a heavily thinned area, was significantly different between the two treatments. While recovery variables suggested there was no significant difference in recovery between the treatments for the four sites combined, shigometer measurements at sites 3 and 4 suggested that the thinned stands are recovering better than unthinned stands. At site 1, the unthinned area was recovering better than the thinned areas. At site 2, the thinned area was more vigorous than the unthinned area, but a significant difference did not exist between the two. In addition, when data was analyzed according to site and species, the thinned stands were more vigorous than the unthinned stand. Core data suggests that trees were growing significantly better in the thinned stands after the ice storm, although for the two years prior to the storm, the thinned stands were doing significantly better as well. The aerial approach for detecting ice injury was not comparable to the ice injury maps created using the more traditional manual approach. The overall accuracy of the new method correctly identifying the ice injury damage classes was 60%. Even if the overall accuracy met with accepted accuracy standards (85% accuracy), time and cost limitations would prevent the digital classification approach from becoming an appropriate method for using remotely sensed data to detect and assess ice injury.