Geophysical Feature Extraction and Spatiotemporal Analysis of Polar Sea Ice Using High Spatial Resolution Imagery

Abstract

The Arctic sea ice region has become an increasingly important study area since it is not only a key driver of the Earth’s climate, but also a sensitive indicator of climate change. To model and validate sea ice changes, it is crucial to extract geophysical features of sea ice from high-resolution remote sensing data. We collected a large volume of remote sensing images from multiple platforms such as airborne Digital Mapping System (DMS) and Worldview series satellite in the Arctic region during melting season. Processing such a large volume of imagery poses a significant challenge for extracting sea ice spatiotemporal patterns in a timely manner. Additionally, high spatial resolution (HSR) has been largely ignored due to its complex and heterogeneous nature in both space and time, and Arctic operational missions can routinely produce hundreds of gigabytes of data. The advancement of drone technologies keeps adding rapidly to the volume of sea ice aerial-survey-based observations. In summary, processing such big sea ice data includes challenges such as: 1) the big data challenges in HSR image product, e.g., the big data volume and the heterogeneous formats of a variety of sea ice HSR image data collected by different platforms or agencies; 2) the lack of standard sea ice feature extraction procedure from HSR imagery; 3) the ability for managing, visualizing, and processing HSR sea ice image data, and extracting geophysical properties or attributes. I propose a reliable and effective high-accuracy and high-performance approach to extract sea ice geophysical features from a large amount of HSR remote sensing data to support scientists and allow them to gain new insights from the spatiotemporal analysis on big data process. The objectives of this research are to 1) develop an efficient geophysical feature extraction workflow based on object-based image analysis (OBIA) method for HSR image data to classify different sea ice features and extract the relevant geophysical parameters such as sea ice leads, sea ice floe, melt pond and ice ridge; 2) design a practice workflow to analyze spatiotemporal patterns of sea ice geophysical features; and 3) design and develop a prototype of an on-demand web service for the cyberinfrastructure, providing a publicly available portal for various data owners and users. In order to achieve these objectives, an on-demand sea ice HSR imagery management and processing service is developed, and a scientific case study is demonstrated for geophysical feature extraction and spatiotemporal analysis of sea ice leads. This research on geophysical feature extraction and spatiotemporal analysis of sea ice from high spatial resolution data is innovative for: 1) the practical OBIA classification workflow in a distributed environment for large datasets; 2) the extracted geophysical features could serve as ground references in sea ice research; 3) the developed arctic cyberinfrastructure provides a data service prototype for polar community. The results of this research can be helpful for the understanding of sea ice processing and utilization of climate modeling and verification at different scales.