Using time series of Sentinel-1 SAR observations to evaluate passive microwave melt detection methods over Antarctic Peninsula Ice Shelves
Abstract
Satellite passive microwave datasets provide high repeat surface emission measurements of the cryosphere at coarse resolutions from 1978 onward. Over ice sheets and ice shelves, melt detection algorithms are applied to the time series of passive microwave brightness temperatures to determine if meltwater is present for each day. Unfortunately the results of these melt detection algorithms can vary greatly from one method to another and few ground truth datasets are available. Today however, Sentinel-1 SAR provides high resolution C-band observations over these regions with a repeat of 1-3 days, and the presence of surface meltwater can be detected in SAR time series. This study evaluates the effectiveness of passive microwave melt datasets by comparing them to SAR melt datasets over Antarctic Peninsula Ice Shelves. Four passive microwave melt detection algorithms are applied, including a new proposed method using a KMean Clustering Algorithm to detect melt. Time series of Sentinel-1 SAR data is acquired and prepared using Google Earth Engine. Comparison is done over flat, uniformly melting ice shelf regions. High-resolution SAR imagery also illuminates the difficulties of applying passive microwave techniques in regions with mixed melting and non-melting, such as where topography is present.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFM.C31A1482J
- Keywords:
-
- 0555 Neural networks;
- fuzzy logic;
- machine learning;
- COMPUTATIONAL GEOPHYSICS;
- 0758 Remote sensing;
- CRYOSPHERE;
- 0794 Instruments and techniques;
- CRYOSPHERE