Sum values within one raster using another raster (1km) ArcGIS Desktop 10.3

[RingPang]

Well, I just want to match population raster(see appended "ahpopprj2015w.rar") with land use raster("ah151k.rar").

pop raster information:

spatial references:GCS_WGS_1984;Datum:D_WGS_1984;size is (0.0008333, 0.0008333);unit:degree.

land use raster info:

spatial references:Albers_Conic_Equal_Area;Datum:D_Krasovsky_1940;size is (1000,1000);unit:meter.

I tried to use Project Raster to transform pop's projection to land's, but just don't know how to set "Geographic Transformation". Further, "resampling techique" has no "sum" option, so how to set "output cell size"?

Regards

[DanPatterson_Retired]

You can specify the cell size and snap raster to match your other raster according to the help topic (see the Environments setting. As for the transformation, did it offer you a transformation?  There is documentation on appropriate transformations in ....

geographic_transformations.pdf located in \ArcGIS\Desktop\10.\Documentation.

Code samples are there since you didn't indicate whether you were using the tool or coding a solution

[DanPatterson_Retired]

And are you looking to Tabulate Area when you get everything set

[RingPang]

Hi Dan, I am trying to project pop raster(spatial references:GCS_WGS_1984;Datum:D_WGS_1984) to match land raster(spatial references:Albers_Conic_Equal_Area;Datum:D_Krasovsky_1940) using Project Raster tool in Arcmap 10.3 as follows:

After reprojection, using sum option in Zonal Statistics tool to get total population for check. Result showed that after reprojection, sum of all cell value is only 48,650,000, much less than 61,140,000 before reprojection. Output cell size is default. The most probable geographic transformation is set. I tried different Resampling  Technique options, yet nothing different.

What might cause such great error? 

I have appended materials in question, could you have a try?

Thank you！

[DanPatterson_Retired]

cubic is going to cause some issues because of potential overshoot and undershoot bilinear should suffice.  Check the Project Raster docs for further discussion on interpolation issues. This is a classic case where trying to reproduce an areal sum in raster world would be best served if the vector data could be obtained and converted to raster using the desired projection in the first place, rather than a reprojection of a raster.

[RingPang]

Thks for your patience, Dan.

I agree with your advice that the best way to solve this problem is to obtain vector data.

The original pop data is from:http://www.worldpop.org.uk/data/files/index.php?dataset=298&action=group

It's pity having no vector data. I am trying "people per hectare" data rather than current "people per pixel" data , do you think will it help?

And as for current pixel data, I heard using "Raster to point" and "point to raster", do you think it'l help?

--

Thanks a lot

Li Jun

祝好

李俊

在 2017-09-23 10:05:35，"Dan Patterson" <geonet@esri.com> 写道：

|

|

GeoNet

|

|

Re: Sum values within one raster using another raster (1km) ArcGIS Desktop 10.3

reply from Dan Patterson in GeoNet Help - View the full discussion

cubic is going to cause some issues because of potential overshoot and undershoot bilinear should suffice. Check the Project Rasterdocs for further discussion on interpolation issues. This is a classic case where trying to reproduce an areal sum in raster world would be best served if the vector data could be obtained and converted to raster using the desired projection in the first place, rather than a reprojection of a raster.

Reply to this message by replying to this email, or go to the message on GeoNet

Start a new discussion in GeoNet Help by email or at GeoNet

Following Re: Sum values within one raster using another raster (1km) ArcGIS Desktop 10.3 in these streams: Inbox

This email was sent by GeoNet because you are a registered user.

You may unsubscribe instantly from GeoNet, or adjust email frequency in your email preferences

|

|

[DanPatterson_Retired]

Yes some have had success with raster to point, then interpolation from that ( I think, but not sure, that Natural Neighbor is used). 

Population density should be based on people per land unit, hence hectare would be more appropriate.