Dear Juliano,

Thanks for your interest in MGET. I should be able to help you with this problem.

There are two main methods for applying the Cayula-Cornillon algorithm with MGET in ArcGIS:

1.    Use Cayula-Cornillon tools listed under the Data Products node of the MGET toolbox. These tools adapt the algorithm to specific remote sensing datasets. For example, if you navigate to MGET à Data Products à NOAA National Oceanographic Data Center à 4km AVHRR Pathfinder Version 5.0 and 5.1 SST, you will see a Cayula-Cornillon tool in there. For that tool, I recommend the following parameters:

a.    Temporal Resolution: Daily

b.    Observation Time: Nighttime

c.    Front Detection Threshold: depends on what fronts you want to detect. Try 0.5 or 1.0 to start with.

d.    Output Workspace: pick a folder to store the images.

e.    Masking method: for now, I recommend picking Standard. I recently discovered a problem with the Optimized method that causes it to not mask all of the clouds in certain situations. Depending on the time of year and weather, the Optimized method might still work for you. If it does, it will perform better than Standard. But if it does not, many false fronts could be detected.

f.     Spatiotemporal Extent options: in here, put in a spatial extent corresponding to your area of interest. Then put in a relatively short spatial extent, such as 1 month, just to try out the tool. If you do not specify a temporal extent, the tool will try to download 30 years of data, which will take a very long time! Try to pick a time when the weather is not cloudy. The problem with AVHRR (and all satellite thermal sensors) is that they cannot see through clouds. Depending on your location and season, it can take many days to obtain a cloud-free image.

2.    If you have your own SST images already prepared with software such as Bilko, then use the tools under MGET à Oceanographic Analysis à Find Fronts with Cayula-Cornillon Single Image Edge Detection à In ArcGIS Raster. Make sure your images are in a format that ArcGIS can read. Then try the tool called Cayula-Cornillon Fronts in ArcGIS Raster. This will find fronts in a single image. It is very important that pixels in your image be set to “no data” wherever there is land or clouds. When I looked at your attached image, they were not set to “no data”. That will cause a problem with the fronts tool. It will detect fronts all around land and clouds, because it believes there is very strong temperature gradients there. In order to use this tool, your image must be integers, not floating point numbers such as degrees C. If you have floating point numbers, you can convert to integer using the Map Algebra _expression_ under the Input Raster Options. There is an example there. Please read the documentation for it carefully. Then, you must set the Front Detection Threshold to the integer difference that will constitute a front. For example, if your integers each represent 0.1 degrees C, then to find a front of 0.5 degrees C you would use a threshold of 5.

Good luck. Let me know if you have any questions. Sorry there is no better tutorial yet.

Best regards,

Jason