Automatic Calculation of Ice Drift Based on AMSR Data

The results of applying a new method for the automatic calculation of ice drift based on the sequence of Advanced Microwave Scanning Radiometer (AMSR) images are considered. The method is analogous to the cross-correlation method but uses other ways to track template similarity and a new vector reje...

Full description

Saved in:
Bibliographic Details
Published in:Izvestiya. Atmospheric and oceanic physics 2017-12, Vol.53 (9), p.1123-1131
Main Authors: Alexanin, A. I., Stopkin, M. V., Kachur, V. A.
Format: Article
Language:English
Subjects:
Algorithms
Climatology
Correlation coefficient
Correlation coefficients
Drift
Earth and Environmental Science
Earth Sciences
Geophysics/Geodesy
Ice drift
Mathematical analysis
Methods and Means of Processing and Interpretation of Space Information
Microwave radiometers
Radiometers
Rejection
Sea ice
Similarity
Templates
Vectors
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The results of applying a new method for the automatic calculation of ice drift based on the sequence of Advanced Microwave Scanning Radiometer (AMSR) images are considered. The method is analogous to the cross-correlation method but uses other ways to track template similarity and a new vector rejection criterion. Ice-concentration maps constructed using the ARTIST Sea Ice (ASI) algorithm (University of Bremen) based on data from 89 GHz spectral channels are used as images. While highly accurate at calculating ice drift, the new method has made it possible to reduce the template size, which makes it possible to obtain more detailed drift maps. Even in the case of a template linear size of 50–70 km, the ice-drift calculation accuracy is within 5 cm/s and method informativeness (the proportion of constructed vectors and vectors after rejection) is 60–70%. The method shows higher efficiency if the brightness mismatch is used as a measure of similarity between two templates instead of the correlation coefficient.
ISSN:0001-4338
1555-628X
DOI:10.1134/S0001433817090389