detemine domes

MathiasHaarmann · ‎03-09-2013

Hi everone,

you guys know how to determine domes in a raster. I thought something of "curvature", but still cannot distinguish them from basins. Maybe something with a specific concave/convex kurtosis... Anyone´s got some experience with this?

Thanks for your time!

curtvprice · ‎03-09-2013

you guys know how to determine domes in a raster. I thought something of "curvature", but still cannot distinguish them from basins.

Bill Huber would probably have a better idea -- but I can't resist suggesting that if you're looking for local maxima you could try selecting areas where

(ingrid == focalmax(ingrid)) and (ingrid > focalmin(ingrid))

Using a larger-than-default neighborhood may help you find broader "domes" that have some flat areas on the top. These areas could then be collapsed to single-cells or points using the Zonal Geometry tool with the CENTROID option.

MathiasHaarmann · ‎03-11-2013

Bill Huber would probably have a better idea -- but I can't resist suggesting that if you're looking for local maxima you could try selecting areas where
(ingrid == focalmax(ingrid)) and (ingrid > focalmin(ingrid))
Using a larger-than-default neighborhood may help you find broader "domes" that have some flat areas on the top. These areas could then be collapsed to single-cells or points using the Zonal Geometry tool with the CENTROID option.

mh... result still looks quite blurry, especially that code. Maybe someone has got a more concrete approach?

curtvprice · ‎03-11-2013

Sorry, I slipped back into 9x map algebra.

Here's a raster calculator syntax

("inraster" == FocalStatistics("inraster","#","MAXIMUM")) and 
("inraster" > FocalStatistics("inraster","#","MINIMUM"))

Using a larger-than-default neighborhood may help you find broader "domes" that have some flat areas on the top. These areas could then be collapsed to single-cells or points using the Zonal Geometry tool with the CENTROID option.

This will select areas where the center cell is the equal to or greater than the maximum of the neighborhood, as long as the whole neighborhood is not the same value (flat).

MathiasHaarmann · ‎03-13-2013

Sorry, I slipped back into 9x map algebra.

Here's a raster calculator syntax
("inraster" == FocalStatistics("inraster","#","MAXIMUM")) and 
("inraster" > FocalStatistics("inraster","#","MINIMUM"))
Using a larger-than-default neighborhood may help you find broader "domes" that have some flat areas on the top. These areas could then be collapsed to single-cells or points using the Zonal Geometry tool with the CENTROID option.

This will select areas where the center cell is the equal to or greater than the maximum of the neighborhood, as long as the whole neighborhood is not the same value (flat).

Okay, I get some idea of what you´re heading for and the syntax. I tried it but it failed:

dtm20=> my raster-data, # = neighboorhood settings?

("dtm20" == FocalStatistics("dtm20","5","MAXIMUM")) and 
("dtm20" > FocalStatistics("dtm20","5","MINIMUM"))

maybe a short explanation how to implement the code? Or maybe running "focal statistics" first, before some "raster calculator"?

Thanks a lot for your help

curtvprice · ‎03-13-2013

Okay, I get some idea of what you´re heading for and the syntax. I tried it but it failed

Okay -- try this:

1. Add "dtm20" to ArcMap as a raster layer
2. Start Raster Calculator.
3. Compose this expression.

"dtm20" == FocalStatistics("dtm20",NbrCircle(5),"MAXIMUM")

This should give you a 1-0 grid where 1 (True) cells are where the cell is at the local maximum for the 5x5 circular neighborhood. This will include flat areas.

You can play with it further once you get that far.

Example of how to specify neighborhood are in the help for Focal Statistics.

MathiasHaarmann · ‎03-15-2013

Okay -- try this:

1. Add "dtm20" to ArcMap as a raster layer
2. Start Raster Calculator.
3. Compose this expression.
"dtm20" == FocalStatistics("dtm20",NbrCircle(5),"MAXIMUM")
This should give you a 1-0 grid where 1 (True) cells are where the cell is at the local maximum for the 5x5 circular neighborhood. This will include flat areas.

You can play with it further once you get that far.

Example of how to specify neighborhood are in the help for Focal Statistics.

Okay, it worked out, thank you very much for your help.

Just two more things: I guess, in this case, exclusion of focalmins (like �??and (ingrid > focalmin(ingrid))�?�) is redundant, is it (just got a look with "swipe" with the effects-toolbar)? I don´t think �??zonal geometry�?� makes sense in case of chosen cells of a raster, too.

FionaStapleton · ‎03-21-2013

Hi all,

I am working on data which consists of many mound features > 50 m high and average 200 m diameter. Some are rounded and others are elongate.

The edges of the mounds are difficult to see on a shaded relief image, impeding the ability to place accurate zone boundaries around the mounds. Thus, using the measurement tool in ArcMap to find diameter is not a solution. Also, I would like to utilize an automated mound detection method.

Does anyone have a method for detecting mound boundaries?

Thank you,

Stapes

curtvprice · ‎03-21-2013

Does anyone have a method for detecting mound boundaries?

You need a better definition of a mound. We got the local maximum nailed, but then you need to define the extent around the point.

For example if you wanted to define the mound as all contiguous areas at or below the peaks we found using the above map algebra, within x amount of elevation, that would be a start.

FionaStapleton · ‎03-25-2013

Thanks,

Well the mound boundary has been defined as the point where the 6 degree slope contour begins. So this could be used in addition to the peak.

So maybe: all contiguous areas between the 6 degree slope contour and the peak height..?

I'm a GIS amature..

Stapes