Hi Rui,

 

We have not seen a MemoryError from the Cayula-Cornillon tool before, although the tool does use a lot of memory. The main factors are the dimensions of the input raster (larger == more memory), its pixel type (more bits == more memory), and the histogram window size (smaller == more memory), and histogram window stride (smaller == more memory). The parameters in your screen capture seem reasonable (they are the defaults), except that you should provide the original MODIS integers rather than using the map algebra _expression_ shown. What are the dimensions of the input raster, and what pixel type is used?

 

Also, if it is not too much trouble, could you repro this with verbose logging enabled and send me the complete output? Instructions at http://code.env.duke.edu/projects/mget/wiki/Configuring%20MGET%20logging. This will show exactly where the error occurs.

 

Best,

 

Jason

 

 

Hi Nicholas,

 

Sorry to trouble you with this question, that probably is basic. I am trying to use the Cayula-Cornillon edge detector on a MODIS SST raster (created with HDF converter in MGET) and keep getting a memory error.

 

I have MGET 0.7a15 installed in ARCgis 9.3 (with Spatial Analyst enabled), R 2.8.0, Pythonwin 2.5.1, Windows XP SP3. I’ve changed the ArcMap cache size to 1000 MB. The machine is a Toshiba Tecra, Intel CoreDuo, 2.53 Mhz, 2.99 GB ram.        

 

I am using a rectangle with 20º by 20º over the N Atlantic, from an 8day composite MODIS SST raster, created with the HDF converter in MGET; I’ve tried to run the Cayula-Cornillon detector both with the original integers (range 7-25779) and by applying an algebra _expression_ [int((inputRaster) * 0.000717 – 2)] to get the temperature range. The remaining parameters are the default and can be seen in the image.

 

Every time I run the tool, I get the following message:

 

Start Time: Tue Sep 29 16:00:36 2009

Running script CayulaCornillonEdgeDetectionDetectEdgesInArcGISRaster...

Finding fronts in C:\DATA\Environment\SST\SST\NA SST Aquamodis 8 day files\2008\extract_1372 using parameters: medianFilter=3, histogramWindow=32, histogramStride=16, minPropCellsWithData=0.65000000000000002, minPopProp=0.25, minPopMeanDiff=3.0, minTheta=0.76000000000000001, minSinglePopCohesion=0.90000000000000002, minGlobalPopCohesion=0.92000000000000004, threads=1.

Output rasters: fronts=C:\DATA\Environment\SST\SST\NA SST Aquamodis 8 day files\2008\test, filteredImage=None, candidateCounts=None, frontCounts=None, windowStatusCodes=None, windowStatusValues=None

MemoryError:

<type 'exceptions.MemoryError'>:

Failed to execute (CayulaCornillonEdgeDetectionDetectEdgesInArcGISRaster).

End Time: Tue Sep 29 16:00:39 2009 (Elapsed Time: 3.00 seconds)

 

I am a bit skeptical that I am running low on memory to run the tool. After trying to find a similar problem in the questions archive in the mget-help, I see no other solution than to bother you. Is this problem known to you?

 

Thanks for any help you can give,

 

Rui Prieto