Please use this identifier to cite or link to this item:
|Title: ||A comparison of satellite-derived sea surface temperature front using resent two edge detection methods|
|Authors: ||Yi Chang;P. Cornillon;T. Shimada;D. Ullman;Ming-An Lee;H. Kawamura|
|Contributors: ||NTOU:Department of Environmental Biology and Fisheries Science|
|Issue Date: ||2011-10-20T08:24:23Z
|Publisher: ||Guangzhou, China|
|Abstract: ||abstract:Satellite-derived SST fronts provide a valuable resource for the study of oceanic fronts. In this study, two edge detection algorithms designed specifically for the identification of fronts in satellite-derived SST fields are compared. One of these is the entropy-based algorithm of Shimada et al. (2005) and the other the histogram-based algorithm of Cayula and Cornillon (1992). The algorithms were applied to a one year (2004) series of AVHRR-derived SST fields and the probability of finding a front at each pixel location - the front probability distribution - was calculated for the time series for each algorithm. Although the general characteristics of the distributions are similar, the entropy-based algorithm tends to find more fronts on average than the histogram-based algorithm. This is true everywhere suggesting a lower threshold for noise. The increase in the number of background fronts results from three contributions: (1) fronts found by the entropy-based algorithm tended to be `thicker' resulting in a greater contribution from each front, (2) shorter fronts were accepted by the entropy-based algorithm than by the histogram-based one, and (3) the number of frontal segments in a given region does not have the same constraint for the entropy-based as for the histogram-based algorithm. In light of these observations, the fronts obtained with the entropy-based algorithm were decomposed into segments and the same length and thickness constraints were applied to these segments as to those from the histogram-based algorithm. This post-processing of fronts from the entropy-based algorithm resulted in a distribution that was substantially closer to the histogram-based results.|
|Appears in Collections:||[環境生物與漁業科學學系] 演講及研討會|
Files in This Item:
There are no files associated with this item.
All items in NTOUR are protected by copyright, with all rights reserved.