# Buffer raster and estimate values

295
6
03-30-2021 01:24 PM
Labels (2)
New Contributor

I have a raster of worldwide marine fish species counts, and want to apply those values to nearby coast (e.g., estimate the amount of diversity available to people on the coasts). I have a general sense that the thing to do would be to place a buffer on the marine raster -- I'm defining "coast" as land within 50 km of the ocean -- and then assign values to the cells in the buffer based on an average of nearby marine cells. I want the coastal buffer to end up being its own layer. Pretty new to GIS, so I'd appreciate it if suggestions could be fairly explicit about what to do. Thanks!

1 Solution

Accepted Solutions
New Contributor II

Not a problem--it seemed an analytically attractive problem set to me.  With the additional info about what you're trying to do, I think you might be able to use part of my droning answer. ;)

Specifically, approach "B" involved you deciding how often you wanted unique data points (for comparing available species) along the coastline.  You could pick 10, 20, 50, 100 km or whatever you'd like.  You'd then use those points to get the average number of species within a "fishing range" you pick (Buffer, then Zonal Stats-Average) for each point.  Once you have those points on the coast with values, you can use them together with your other data to calculate a weighted value for polygons of land on along the coasts.

Sorry, I probably missed again, but I thought I'd try ...

6 Replies
New Contributor II

Could you provide some more details about your fish species raster?  Does each cell of the raster provide a value (a "count") of how many different species exist at that location?  If you look at a group of, say, 3 adjacent cells with different values, do you have any way to tell how many different species are represented in those combined results?  For example,  if the cell values for the 3 cells were: 2; 3;  and 2,  do you know if you have just 3 unique species or possibly 4, 5, 6 or 7?

Sorry if that's confusing, but what your data looks like impacts how you approach your analysis.

New Contributor

Hi! Each cell is just a count value; there's no way to tell about overlap.

New Contributor II

Okay, so let's walk down the analytical path a bit on this.  You said that your analytical goal was to display the "amount of diversity available to people on the coasts."  Let me lay out the way I'd think about the problem.

Focusing only on direct distribution (ignoring national / regional distribution systems), most people within 50 miles of the coast are going to get their fresh fish from the people that caught them.  Most of the people catching fish in bulk are going to be working out of ports.  Fisherman working "local waters" (i.e. same day out & back) out of those ports have some rough working range (might vary some by their boat, but you could use some midpoint).  Now comes the data availability question -- can you find a data source that gives you the ports you'd like to use as point data?

A.  If you can, you could buffer each of those points using the fishing radius, then use the "Zonal Statistics" tool to grab the average species count for each buffer.  You could then use the "Join Field" tool to merge the output of the Zonal Statistics back into your port data.  With that data column in the ports layer, you could apply symbology to adjust either the color or size of each port based upon how many species it has available.

B. If you can't, you could use the "Generate Points Along Lines" tool to turn polygons of coasts into point (you'd pick the distance between points) then use those points instead of actual ports to do the analysis.

I hope this helps.

New Contributor

Wow, I appreciate your taking the time to examine the whole analytical picture! I was writing in brief in the original ask; this is an archaeological project, and I'm using marine fish diversity, along with marine net primary productivity, land primary productivity, effective temperature, and a slew of other variables to estimate the suitability of different landscape for prehistoric hunter-gatherers. I'm not interested in ports, per se, so much as the productivity of a landscape -- and granting coastal land weight based on marine resources.

New Contributor II

Not a problem--it seemed an analytically attractive problem set to me.  With the additional info about what you're trying to do, I think you might be able to use part of my droning answer. ;)

Specifically, approach "B" involved you deciding how often you wanted unique data points (for comparing available species) along the coastline.  You could pick 10, 20, 50, 100 km or whatever you'd like.  You'd then use those points to get the average number of species within a "fishing range" you pick (Buffer, then Zonal Stats-Average) for each point.  Once you have those points on the coast with values, you can use them together with your other data to calculate a weighted value for polygons of land on along the coasts.

Sorry, I probably missed again, but I thought I'd try ...

New Contributor
This is a great lead -- thanks!