Single image summary of time-varying Earth-features

Date
2018-07-18
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The Earth's surface is live and dynamic due to natural and human-made events. Tracking and visualizing Earth-features (e.g., water, snow, and vegetation) is a significant problem. In this thesis, we explore a single image summary approach to detecting changes in Earth-features. Earth observation satellite imagery like Landsat 8 makes the tracking feasible by providing detailed multispectral imagery at regular intervals. Each Landsat scene for a region consists of 11 bands where each band belongs to a fixed range of wavelength and is useful in different remote sensing disciplines. Landsat 8 dataset offers the ability to identify more than what can be observed through the naked eye (using conventional RGB images). In our method, we use spectral indices prepared by combinations of two or more bands, to identify features. Appropriate thresholds for spectral indices help in feature identification. Next, we reference datasets and combine multiple images to generate recurrence-maps (represents the recurrence of a particular feature in the region of interest). Application of predefined color scales to recurrence-maps creates a single image summary of features for a region. We illustrate the benefit of our method with case-studies for the Lake Urmia, the Amazon Rainforest, the Bering Glacier, and the California Rocky Fire. These case studies decipher change-trends such as droughts, deforestation, glacial melting, and burn-area recovery.
Description
Keywords
Citation
Tripathi, G. (2018). Single image summary of time-varying Earth-features (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/32650