Quantifying Melt Ponds in the Beaufort MIZ using Linear Support Vector Machines from High Resolution Panchromatic Images
Abstract
Much work has been done on determining changes in summer ice albedo and morphological properties of melt ponds such as depth, shape and distribution using in-situ measurements and satellite-based sensors. Although these studies have dedicated much pioneering work in this area, there still lacks sufficient spatial and temporal scales. We present a prototype algorithm using Linear Support Vector Machines (LSVMs) designed to quantify the evolution of melt pond fraction from a recently government-declassified high-resolution panchromatic optical dataset. The study area of interest lies within the Beaufort marginal ice zone (MIZ), where several in-situ instruments were deployed by the British Antarctic Survey in joint with the MIZ Program, from April-September, 2014. The LSVM uses four dimensional feature data from the intensity image itself, and from various textures calculated from a modified first-order histogram technique using probability density of occurrences. We explore both the temporal evolution of melt ponds and spatial statistics such as pond fraction, pond area, and number pond density, to name a few. We also introduce a linear regression model that can potentially be used to estimate average pond area by ingesting several melt pond statistics and shape parameters.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2017
- Bibcode:
- 2017AGUFM.H31G1597O
- Keywords:
-
- 3399 General or miscellaneous;
- ATMOSPHERIC PROCESSES;
- 0430 Computational methods and data processing;
- BIOGEOSCIENCES;
- 1622 Earth system modeling;
- GLOBAL CHANGE;
- 1855 Remote sensing;
- HYDROLOGY