Skip to Content.

mget-help - [mget-help] Canny Edge detection with SSS

Please Wait...

Subject: Marine Geospatial Ecology Tools (MGET) help

Text archives


[mget-help] Canny Edge detection with SSS


Chronological Thread 
  • From: Liz Atwood <>
  • To:
  • Subject: [mget-help] Canny Edge detection with SSS
  • Date: Wed, 18 May 2016 17:43:52 +0200

Hi Jason,

I am now trying to also calculate sea surface salinity fronts using the Canny Edge algorithm like you suggested back in March (the Cayula-Cornillon seems to be working fine, but I'd like to check both methods). The process runs through no problem and I enabled the verbose option to be able to check what thresholds are being selected. Using the non-transformed 32-bit floating point NetCDF files, the high threshold was automatically set at 0.03125 and the lo at 0.0125. But I know that the SSS data have an error of ±0.14 and I'd rather be focusing on fronts that have a difference of at least 0.3 psu. Unfortunately changing the high threshold to something over 0.1 results in only zeros.

I also tried scaling up the data by 100 (and transforming to 16-bit unsigned int in the process) and let the algorithm go without specifying the thresholds, which again resulted in values of 0.03125 and 0.0125 (which is strange since everything is multiplied by 100?). Setting the thresholds to lo 14 (since my SSS error is 0.14 psu) and high 35 (just dividing by the standard 0.4 listed in the mget algorithm help page) again results in only zeros.

My question: Am I correct in understanding that the first run listed above found fronts with an above 0.0125 psu difference? It seems like the process will only detect fronts if thresholds below the dataset error level are implemented, would I be right in concluding that this method is not useable with SSS data?

I've uploaded some data in case you want to play around somewhat:
https://www.dropbox.com/sh/kh05294ecvs1ntd/AAA8k8HNGn29VKRfLYWszA4ya?dl=0

Thanks for any help you might be able to provide!

Best,
Liz
-- 
Elizabeth C. Atwood
M.Sc. Quantitative Ecology & Resource Management

RSS Remote Sensing Solutions GmbH
Isarstr. 3
D - 82065 Baierbrunn

Tel.: +49 (0)89/ 4895-4765
Email: 

www.rssgmbh.de/en.html


  • [mget-help] Canny Edge detection with SSS, Liz Atwood, 05/18/2016

Archive powered by MHonArc 2.6.19.

Top of Page