Date of Award
Master of Science (MS) in Geographic Information Science & Technology
Chemistry & Geosciences
M. Sean Chenoweth
This thesis aimed to evaluate three methods of analyzing blue roofing tarpaulin (tarp) placed on homes in post natural disaster zones with remote sensing techniques by assessing the different methods- image segmentation, machine learning (ML), and supervised classification. One can determine which is the most efficient and accurate way of detecting blue tarps. The concept here was that using the most efficient and accurate way to locate blue tarps can aid federal, state, and local emergency management (EM) operations and homeowners. In the wake of a natural disaster such as a tornado, hurricane, thunderstorm, or similar weather events, roofs are the most likely to be damaged (Esri Events., 2019). Severe roof damage needs to be mitigated as fast as possible: which in the United States is often done at no cost by the Federal Emergency Management Agency (FEMA).
This research aimed to find the most efficient and accurate way of detecting blue tarps with three different remote sensing practices. The first method, image segmentation, separates parts of a whole image into smaller areas or categories that correspond to distinct items or parts of objects. Each pixel in a remotely sensed image is then classified into categories set by the user. A successful segmentation will result when pixels in the same category have comparable multivariate, grayscale values and form a linked area, whereas nearby pixels in other categories have distinct values. Machine Learning, ML, a second method, is a technique that processes data depending on many layers for feature v identification and pattern recognition. ArcGIS Pro mapping software processes data with ML classification methods to classify remote sensing imagery. Deep learning models may be used to recognize objects, classify images, and in this example, classify pixels. The resultant model definition file or deep learning software package is used to run the inference geoprocessing tools to extract particular item positions, categorize or label the objects, or classify the pixels in the picture. Finally, supervised classification is based on a system in which a user picks sample pixel in an image that are indicative of certain classes and then tells image-processing software to categorize the other pixels in the picture using these training sites as references. To group pixels together, the user also specifies the limits for how similar they must be. The number of classifications into which the image is categorized is likewise determined by the user.
The importance of tracking blue roofs is multifaceted. Structures with roof damage from natural disasters face many immediate dangers, such as further water and wind damage. These communities are at a critical moment as responding to the damage efficiently and effectively should occur in the immediate aftermath of a disaster. In part due to strategies such as FEMA and the United States Army Corps of Engineers’ (USACE) Operation Blue Roof, most often blue tarpaulins are installed on structures to prevent further damage caused by wind and rain. From a Unmanned Arial Vehicles (UAV) perspective, these blue tarps stand out amid the downed trees, devastated infrastructure, and other debris that will populate the area. Understanding that recovery can be one of the most important stages of Emergency Management, testing techniques vi for speed, accuracy, and effectiveness will assist in creating more effective Emergency Management (EM) specialists.