Best practices for Clipping data from Living atlas

461
9
Jump to solution
09-14-2021 08:30 PM
DanielPhillipsTTU
New Contributor II

Hi All, 

I'm trying to do something that I think seems straightforward and simple.  I want to clip a portion of a living atlas layer (which I've loaded onto my map in arcgis Pro), using a study area boundary polygon (that I've custom defined and also added to the map). I usually use the "extract by mask" tool to do this to a raster dataset, or "clip raster".  However, when I attempt to run either of these to a living atlas layer, it seems to take forever.  I've tried letting it run, but for some reason it yields a huge file (8gb?) for a 2000 acre study area. Obviously I'm doing something wrong?  I know the living atlas layer often represents an entire continent/planet of imagery so they must be huge.  But what is the best practice for actually extracting useful pieces from it for analysis?  

Please advise. 

0 Kudos
1 Solution

Accepted Solutions
RajinderNagi
Esri Contributor

Hi @DanielPhillipsTTU ,

From your screenshot, it seems you are leaving the cell size as service default (0.25 m), which is causing the large size (as noted by Noman also). The elevation services are comprised of multiple sources and resolutions. The default cell size of the service is based on the finest source available at someplace on the earth, but it doesn't mean the whole earth is covered at that resolution. So, defining the appropraite cell size is very important when doing analysis or extracting. You can check the elevation coverage map https://esriurl.com/elevationcoveragemap to see which resolutions are available in your area. You need to use 'Make Image Server Layer' tool and specify the cell size (based on your need) and processing extent. There is a limit of 5000 rows x 5000 cols and if you specify a cell size of 10m, you would be able to extract ~ 50000 x 50000 sq m. I have covered elevation services for analysis in UC presentation available at https://uc2021.esri.com/live-stream/19775209/Living-Atlas-Imagery-and-Elevation. The elevation section starts at the ~ 20 min mark. Let me know if you have further questions.

View solution in original post

0 Kudos
9 Replies
Brooks_SummitGeo
Regular Contributor

@DanielPhillipsTTU are you trying to extract a raster dataset from an image service or from a raster tile cache? If the latter, to my knowledge, the raster toolsets wouldn't work, and I'm surprised theyd even allow you to execute the tools on them. 

If the former, then have you tried restricting the processing extent to match your clip area extent in the environment variables to see if that helps?


David
..Maps with no limits..
DanielPhillipsTTU
New Contributor II

Hi David, thanks for your response.  Yes, I am trying to extract a raster dataset from an image service (assuming that living atlas is an image servie?) I restrictied processing extent to the clipping polygon, still getting the same issue.  

0 Kudos
MervynLotter
Regular Contributor

When working with image services, I like using the Make Image Server Layer GP tool to create a temporary layer using a projection and raster size suitable for my project. This 'layer' can then be used as input into other tools, or exported.

DanielPhillipsTTU
New Contributor II

Thanks Mervyn,  I am not familiar with this tool.  I created an image server layer, but even that is showing an "uncompressed file size of 28 gb?  I'm wondering if the slowness is related to the size of the file (and why would the clipped area be so huge?).  Still taking too long with this route. Any further insight would be greatly appreciated.

0 Kudos
NawajishNoman
Esri Contributor

Hi Daniel, 

You should be able to extract a raster using the Extract by Mask tool. You are accepting the default cell size which is probably resulting in a large raster dataset.

I understand you have specified the Processing Extent Geoprocessing Environment while running the tool. Can you also please try to specify a larger Cell Size in the environment to see if the tool runs faster and create a smaller output? Please specify both the Processing extent and the Cell Size. 

If the tool extracts the raster correctly, you can calculate the size of the raster for the larger cell size by multiplying the number of rows with the number of columns. Once you know the number of rows and columns for the larger cell size, you can also estimate the number of rows, columns, and size of the raster as well as the download time for the default or smaller cell size. 

Please let us know if you still notice a discrepancy.

Thanks

Noman

Nawajish Noman, Esri 

0 Kudos
DanielPhillipsTTU
New Contributor II

Yes, adjusting cell size seemed to work much faster with smaller outputs.

0 Kudos
RajinderNagi
Esri Contributor

Hi @DanielPhillipsTTU ,

From your screenshot, it seems you are leaving the cell size as service default (0.25 m), which is causing the large size (as noted by Noman also). The elevation services are comprised of multiple sources and resolutions. The default cell size of the service is based on the finest source available at someplace on the earth, but it doesn't mean the whole earth is covered at that resolution. So, defining the appropraite cell size is very important when doing analysis or extracting. You can check the elevation coverage map https://esriurl.com/elevationcoveragemap to see which resolutions are available in your area. You need to use 'Make Image Server Layer' tool and specify the cell size (based on your need) and processing extent. There is a limit of 5000 rows x 5000 cols and if you specify a cell size of 10m, you would be able to extract ~ 50000 x 50000 sq m. I have covered elevation services for analysis in UC presentation available at https://uc2021.esri.com/live-stream/19775209/Living-Atlas-Imagery-and-Elevation. The elevation section starts at the ~ 20 min mark. Let me know if you have further questions.

View solution in original post

0 Kudos
DanielPhillipsTTU
New Contributor II

Thanks @RajinderNagi that is great and I'll take a look at the video. One last question after getting it successfully clipped: I'm noticing that when I clip raster/mosaic data (extract by mask) from a living atlas layer, the clipped result doesn't always retain original symbology (IE all the values get randomly assigned to different colors).  I have tried using the GP tool "Apply symbology from layer" also without luck.  I also notice that all the processing templates are lost after clipping.  Is there a way to preserve processing templates and symbology when clipping, or importing from the unclipped layer that is still loaded into my map?  Screenshot attached.

0 Kudos
RajinderNagi
Esri Contributor

@DanielPhillipsTTU , I tried clip raster and extract by mask tool at my end on elevation aspect map layer and in both tools, symbology was preserved (Pro 2.7.1). Not sure which version of Pro you are using. Regarding processing templates, those are server-side raster functions and will no longer be preserved once you clipped a portion out (saved raster on your disk). Hope it helps!

0 Kudos