Averaging multiple point data layers

895
7
Jump to solution
02-27-2022 05:49 AM
GraceWood
New Contributor II

Hello,

I've used the merge tool in ArcGIS Pro 2.7 with a 'mean' rule to find the average of several point data layers of monthly sea salinity. I am unsure what I did wrong as the output layer has far more information than any of the input fields. To clarify, each layer had around 500,000 fixed points of data, I merged 12 layers and the output now has around 6,000,000. So I assume the data has all been placed into one layer, but not averaged.

Could anyone help explain where I may have made a mistake? Or if this is actually the wrong tool to use?

I am new to this software so any help would be greatly appreciated!

0 Kudos
1 Solution

Accepted Solutions
DanPatterson
MVP Esteemed Contributor

When you say "gridcode" was the data originally in a "grid" format?  It would be a lot easier to work with rasters than points because you can use

Cell Statistics (Spatial Analyst)—ArcGIS Pro | Documentation

and get the average of a bunch of input rasters on a cell-by-cell basis.  If you need points from that, you can do the conversion after you have the result


... sort of retired...

View solution in original post

7 Replies
KimGarbade
Occasional Contributor III

I'm assuming you mean that one location (having the same ID across all of the input layers) has multiple records in the source layers (maybe different sampling events) and you want to average them?  Can you give us an idea of the result you want to achieve?  I.E. Do you want the average in a separate table, do you want to label each point with the average, generate a report, symbolize based on the average value? 

0 Kudos
DanPatterson
MVP Esteemed Contributor

Are your points from multiple layers sharing the exact same coordinate values?

If not, you may have to aggregate the points into clusters

Aggregate Points (Cartography)—ArcGIS Pro | Documentation

then summarize the point values within the clusters

Summarize Within (Analysis)—ArcGIS Pro | Documentation

further insights into your current data patterns would help with other options


... sort of retired...
0 Kudos
GraceWood
New Contributor II

Thanks very much for taking the time to reply.

To clarify the 12 layers are global monthly mean sea salinity values and I would like to find an annual mean. The datasets are all records from the same point locations. I assumed only an average of these would be presented in the attribute table of the output layer after using the 'mean' rule.

So that would be around 500,000 points, all an average of the 12 monthly mean point datasets. I don't know if the attached screenshot helps but I have included it in case a mistake is noticeable there.

Apologies for being vague, I'm still getting to grips with the software terms and expressions.

0 Kudos
DavidPike
MVP Frequent Contributor

Merge is quite different in a geoprocessing context as opposed to an editing one (whereby 2 or more features become 1).  The operation simply outputs the union of all input features into a new feature class.

You can probably accomplish what you need with the resultant merged feature class as input to Dissolve (Data Management)—ArcGIS Pro | Documentation , specifying your MEAN field map on the required fields.

0 Kudos
GraceWood
New Contributor II

Hi, thanks very much for explaining this issue.

I have tried using Dissolve and several warnings appeared, the table also shows only 13 rows. I'm not sure the merged output is in the right format for this operation as each of the input features are not given in seperate columns, instead there is a single column (titled: grid code) of combined data, hence my original confusion at there being 6 million rows instead of the expected 500,000.

0 Kudos
DanPatterson
MVP Esteemed Contributor

When you say "gridcode" was the data originally in a "grid" format?  It would be a lot easier to work with rasters than points because you can use

Cell Statistics (Spatial Analyst)—ArcGIS Pro | Documentation

and get the average of a bunch of input rasters on a cell-by-cell basis.  If you need points from that, you can do the conversion after you have the result


... sort of retired...
GraceWood
New Contributor II

It was yes. I converted it to point data as I intended to perform a spatial join with a polygon layer later on.

I have used the original raster data and it seems to have worked perfectly.

Thank you for your advice!

0 Kudos