Hurricanes Working Group: mapping the temporal evolution of land-use/land-cover along the historical tracks of land-falling hurricanes and tropical cyclones in the eastern United States

Research Summary

Hurricanes and tropical storms (collectively known as tropical cyclones, hereafter referred to as TCs) are regular events of varying magnitude (category dependent) and moderate frequency (years-1) that provide a significant influx of freshwater resources to surface and subsurface reservoirs in the warm season. Beyond the popular association with devastating disasters, hurricanes and tropical storms with varying degrees of intensity, structural organization, and terrestrial tracks are an essential element of the hydroclimatic regime of vast regions of the planet, including the Asian Monsoon Region, the Caribbean, and of particular interest to this proposal the eastern and southern United States which will be the test-bed of our work. Despite a great deal of fundamental and applied research of the societal impacts of extreme events and climate change over the last 15 years, an end-to-end concerted effort to identify and characterize the linkages and feedbacks that determine the response of the terrestrial environment to extreme events including water quality and quantity and ecosystem changes has not been undertaken. In the context of global climate change, the prognostic of increased frequency of large-scale storm systems demands research to elucidate linkages and feedbacks among pathways of disruption (flooding and landslides, wetland erosion, forest damage), resilience (freshwater recharge and resetting of biogeochemical cycles), and adaptation, which may be used subsequently to develop mitigation and adaptation strategies (recovery of ecosystems and human agency).

Whereas a historical data base of hurricane and tropical cyclone tracks already exists [], systematic forensic studies to inventory and quantify changes along the path of land-falling storms have yet to be conducted but for isolated place-based assessments. The specific research objective of the proposed project is to develop and test a framework for remote sensing analysis that will systematize a comprehensive analysis of historical satellite data (Landsat, AVHRR and SSMI) that will lead to the characterization of continental-scale changes of terrestrial land-use and land-cover along the tracks of land-falling hurricanes and tropical cyclones. We are both interested in detecting changes caused by individual storms, and in characterizing the time-rates and strategies of landscape recovery. The domain of study will encompass the eastern United States east of the 100o meridian and will be used for track-scale studies (1000's kms after TCs landfall), where more than 50% of the US population lives. This research is proposed in the context of developing the infrastructure and basic data to support the resubmission of a multi-investigator Pratt/NSEES proposal to the NSF on the long-term response of terrestrial ecosystems to hurricanes, and extrapolation of climate change impacts.

Research Methodology

The historical tracks of hurricanes will be mapped onto LANDSAT, AVHRR, TMI and more recently MODIS data. The time-series of storm tracks will be analyzed using Ecognition software, and the time-histories of land-use/land-cover (LULC) change from census and from local and state government records will be matched against the track based time-series of satellite data. For each track, the multi-sensor time-histories of satellite data will be saved before the storm, so that a baseline of historical variability of LULC in satellite data can be acquired to distinguish from what happens after the storm. Following careful consultation, it seems that the Ecognition software produced by Definiens is the most effective way of conducting this analysis. The software is based object-oriented and context-enabled image analysis and can be used in a desktop. It supports all relevant applications for this project such as mapping, change detection, target recognition and object detection for natural resource management, and thus it should allow us to focus on the results in a timely manner.


Three distinct types of deliverables are expected: (1) a data base of LULC properties along the tracks of land-falling hurricanes and tropical cyclones using multisensor satellite data, the first such data base; (2) the methodology and computational infrastructure to conduct similar analysis in other regions of the globe and using other data sets; and (3) a synthesis manuscript. Another deliverable will be the submission of a proposal to NSF and one to NOAA.

Research Team:
Ana P. Barros (Pratt, Duke), Robert Jackson (Biology/NSEES, Duke), Jenni Evans and Chris Duffy (Penn State), Douglas Miller (UNC-Asheville), Hans Pearl and Larry Band (UNC-Chapel Hill), Sethu Raman (NC State University) and other Duke colleagues including Paul Baker, Miguel Medina, Ken Reckow and Gabi Hergel.

©2008 Center on Global Change
Box 90658, Duke University, Durham NC 27708-0658
A150 Levine Science Research Center (LSRC), Research Drive
Tel: 919-681-7180 Fax: 919-681-7176