Date of Award
2018
Document Type
Thesis
Degree Name
Master of Science (MS)
Department
Atmospheric Science
Committee Chair
Udaysankar Nair
Committee Member
Rahul Ramachandran
Committee Member
John Mecikalski
Subject(s)
Satellite meteorology., Cirrus clouds., Machine learning., Neural networks (Computer science)
Abstract
Transverse cirrus bands (TCBs) occur in conjunction with meteorological features such as jet streaks and hurricanes and are often used as a proxy for clear-air turbulence. This study examines the viability of using a convolutional neural network to detect TCBs in satellite imagery. The Visual Geometry Group-16 (VGG-16) network architecture developed for general-purpose image classication was adapted for TCB detection using the transfer-learning approach. The modied network successfully detected (94% accuracy) the presence of TCBs in NASA MODIS and VIIRS true-color satellite browse imagery and outperformed a random forest classier (84% accuracy) trained on the same dataset. The CNN was then used to create a multi-year (2013-2015) global climatology of TCB occurrence, and this was used to explore seasonal and inter-annual variability and associations between TCBs and parent phenomena. The climatology analysis showed that TCBs co-occur most commonly with extratropical cyclones in two regions, eastern North America and eastern South America. This study suggests deep learning as a promising approach for detection of meteorological phenomena in satellite imagery, with potential for use in meteorological research and operations.
Recommended Citation
Miller, Jeffrey, "A satellite-based climatology of transverse cirrus band occurrence using deep learning" (2018). Theses. 234.
https://louis.uah.edu/uah-theses/234