nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Zonal Statistics as Table on Model Builder!!!

ShikoNjuno

So I have a model that utilizes the Zonal Statistics as a Table tool and it runs beautifully. However, the output is incredibly false.

The input feature zone data is a shapefile comprised of a select few polygons.
The zone field selected is FID.
The input value raster is a .tiff file which stores the correct values for the output.
The output table is saved to the workspace chosen by user.
And the statistic performed is the MEAN. Which means the output for every FID row is an average of all pixel values in the raster file within that polygon shape.

What's happening when the model runs is that the output is just incorrect....almost like there is no zonal recognition happening.
Is anyone familiar with this and perhaps what I can do to make the output correct??
Thanks

Najděte více příspěvků označených jako

Accepted answers

ShikoNjuno

Interesting things is why you are using MEAN value for landuse code(?)? Want to see the changes after year by year?

This got me thinking....there's no reason to look at the MEAN of the raster files, really what I want is the MAX because in the event that land were to be fallowed in the future, for instance, that statistic tells me the maximum value "lost". And then if I wanted to perform a MEAN statistic, it would be for several of the polygons from the shapefile; this could be done via a Field Statistics analysis!!!!
Problem solved 😄
Thanks for leading me down that thought process!
Cheers.

All comments

ShikoNjuno

Well I'm getting closer to the right output, but in going back and checking ALL the values, I still have one last issue.
Some of the values that I'm getting after running the Zonal Stats tool are slightly off. I'm not getting the raster value that I expect for that shapefile polygon, instead, I'm getting the raster value of the neighboring polygon!!! This isn't happening for all polygons though, just some of them....so it's not very consistent which makes it harder to narrow in on the issue. I'm not too familiar with either one of these but I'm guessing it has something to do with pyramids or resampling methods.
Can anyone help me figure out what might be going on?
Below are my current Zonal Statistics as Table Environment Settings:
[ATTACH=CONFIG]26225[/ATTACH]
Any thoughts?!!

Env_Settings.PNG.zip

ShikoNjuno

Interesting things is why you are using MEAN value for landuse code(?)? Want to see the changes after year by year?

TOMKONG

No...the MEAN is supposed to take the average of all the raster pixels within the corresponding polygon. Here's a snapshot of what I'm working on (Note: has been revised):
[ATTACH=CONFIG]26185[/ATTACH]

At this point I'm pretty sure my problem lies in the environment settings but I can't for the life of me figure out what to change.
Resort: HELP!!!

Did you try using 1*1 cell size?

In another way you can check the result and error reason:
1. Convert your raster to the points (shapfile or geodabase feature class) with code attribute value;
2. Intesect point layer and zonal polygon layer: the intesected point layer will hold both zonal code and raster code values;
3. Using summarized founction (click the ID field name from polygon layer, select "summzrized") to get MEAN value (you can select MEAN, MIN, MAX, MAJ...) for each polygon ID.
----you can do zonal data statistics based on all shapfile also, not always deal with raster.

ShikoNjuno

Interesting things is why you are using MEAN value for landuse code(?)? Want to see the changes after year by year?

No...the MEAN is supposed to take the average of all the raster pixels within the corresponding polygon. Here's a snapshot of what I'm working on (Note: has been revised):
[ATTACH=CONFIG]26185[/ATTACH]

At this point I'm pretty sure my problem lies in the environment settings but I can't for the life of me figure out what to change.
Input: Shapefiles with polygons and GRID raster files that hold the correct output info.
Tool: Zonal Stats as Table
Output: Table showing the MEAN results for all polygons in the shapefile.
Problem: The results do not match the raster info.....close, but not quite.
Solution: STILL RACKING MY BRAIN :mad:
Resort: HELP!!!

Model_1.jpg

TOMKONG

You know you were right about the smaller cell size in this case.
I tried 30 *30 and my output was closer to what it should be. I'm working on a 10 * 10 raster at the moment. It's taking quite a while to create the layer but it's definitely worth a shot.....and the wait 🙂

Interesting things is why you are using MEAN value for landuse code(?)? Want to see the changes after year by year?

ShikoNjuno

The cell size 60 * 60--it is 3600 sq. I don't know your purpose, but it is so big.
You may set it as 1*1, then the return may fit your expectation.
The Cell Assignment method doesn't affect your result but cell size.
If the cell size is too small, the tool doesn't work sometimes. You need to try.

You know you were right about the smaller cell size in this case.
I tried 30 *30 and my output was closer to what it should be. I'm working on a 10 * 10 raster at the moment. It's taking quite a while to create the layer but it's definitely worth a shot.....and the wait 🙂

ShikoNjuno

Usually sampling to a smaller cell size is inappropriate because you will make assumptions from data that is not meant to be used at these scales. I would not resample the raster unless you want your results to be tainted.
How large are the polygons? Realize that the app will first convert your polys to rasters. So as I first said, please set up your environment setting inside the zonal stats tool. The enivronmental settings button is at the bottom of the tool.
Set your snap raster to the raster. The extent to your polys and the cell size to the raster, and see what happens.
Realize that the center of a raster cell has to fall inside the poly before it will be included in the stats.
Hope that helps.

Thanks for you input Alicia,my polygons are all different sizes but the thing about that file is this.... It was originally a raster TIF which I converted to a shapefile of polygons. This created really rough edges around the polygon and it still doesn't look quite right. The reason I had to do this is because I needed to create another raster TIF of one of the attribute fields.Then, as per Tom's earlier suggestion, I converted it to a raster GRID as spatial analyst tools are said to work better with GRID types(which I did find to be true). So far the values for this particular element (landuse) is the most inaccurate of any of my other data. I wonder if it has something to do with all the conversions I've had to make?!?

Anyway I changed the settings as you suggested and below is a snapshot of what my current settings are. Please take a look and let me know if I should make any changes:

[ATTACH=CONFIG]26176[/ATTACH]

And if you can think of anything else I might try let me know too.
Cheers.

Env_Settings.PNG

AliciaMein

The cell size 60 * 60--it is 3600 sq. I don't know your purpose, but it is so big.
You may set it as 1*1, then the return may fit your expectation.
The Cell Assignment method doesn't affect your result but cell size.
If the cell size is too small, the tool doesn't work sometimes. You need to try.

Usually sampling to a smaller cell size is inappropriate because you will make assumptions from data that is not meant to be used at these scales. I would not resample the raster unless you want your results to be tainted.
How large are the polygons? Realize that the app will first convert your polys to rasters. So as I first said, please set up your environment setting inside the zonal stats tool. The enivronmental settings button is at the bottom of the tool.
Set your snap raster to the raster. The extent to your polys and the cell size to the raster, and see what happens.
Realize that the center of a raster cell has to fall inside the poly before it will be included in the stats.
Hope that helps.

TOMKONG

Thanks for the clarification Tom. I actually just realized that all that information is in the Properties menu of each file :rolleyes:
And I have confirmed that they all have matching Spatial Reference.
As for the Cell Size....I'm already pretty tight at 60 by 60, but I'll go ahead and try going lower.
Also, do you know if the Cell Assignment is better as CENTER or MAX??

The cell size 60 * 60--it is 3600 sq. I don't know your purpose, but it is so big.
You may set it as 1*1, then the return may fit your expectation.
The Cell Assignment method doesn't affect your result but cell size.
If the cell size is too small, the tool doesn't work sometimes. You need to try.

ShikoNjuno

Thanks for the clarification Tom. I actually just realized that all that information is in the Properties menu of each file :rolleyes:
And I have confirmed that they all have matching Spatial Reference.
As for the Cell Size....I'm already pretty tight at 60 by 60, but I'll go ahead and try going lower.
Also, do you know if the Cell Assignment is better as CENTER or MAX??

TOMKONG

Hey thanks Tom,
This seems to have made quite a difference. Better!!
However, the values are still not quite right.
Do you know for raster analysis if using the Double or Float Type is more advantageous than the other??
Maybe it is a Spatial Referencing issue as Shitij mentioned?!!

For raster data analyze, the data type in integer will make calculation process quickly(some ESRI analyze tool can only handle integer raster).In your case, there is no difference between two data type.

Spatial reference includes datum, projection, tolerence (or cell size for raster).
You may test if cell size will affect your result by using different cell size.

ShikoNjuno

Export it as GRID (from drop down window to select).

Hey thanks Tom,
This seems to have made quite a difference. Better!!
However, the values are still not quite right.
Do you know for raster analysis if using the Double or Float Type is more advantageous than the other??
Maybe it is a Spatial Referencing issue as Shitij mentioned?!!

ShikoNjuno

Did you check if the spatial references for raster and shapefile match?

How do I check that. I know I checked to make sure the Projection is the same for both, but how exactly do I check spatial reference?

ShitijMehta

Did you check if the spatial references for raster and shapefile match?

ShikoNjuno

Here it is.....let me know if you have any questions.
I'll be working on the TIFF to GRID conversion.

[ATTACH=CONFIG]26160[/ATTACH]

Model_1.jpg

ShikoNjuno

I realize the image is unclear.....let me try again.

ShikoNjuno

Thanks Shitij. Take a look.
Hope you can see this okay.....

[ATTACH=CONFIG]26159[/ATTACH]

The selection is based on a shapefile which has about 15 polygons, which should be zonally linked to the rasters via the Zonal Statistics as Table tool......right?!!?!
So the output is incorrect in that it just doesn't match what the corresponding raster files have as stored values, which is what I was expecting. I understand that the statistic is performed as an average of the pixels within each polygon, but the output for some of the rows is assigned a value even when I know for a fact there is no value i.e. the output should be zero!!!

I hope this isn't too confusing.
Let me know what you think.
Thanks.

Model_1.jpg

ShitijMehta

Print screen > Copy to powerpoint >crop the image and right click the image to save.
Use free version of image capturing softwares> snagit

or

http://www.faststone.org/download.htm

TOMKONG

How do I go from TIFF to GRID?

Export it as GRID (from drop down window to select).

ShikoNjuno

Esri analyze tool only works well under the projected GRID data format (ESRI raste format).
So convert it to grid format first.

How do I go from TIFF to GRID?

TOMKONG

So I have a model that utilizes the Zonal Statistics as a Table tool and it runs beautifully. However, the output is incredibly false.

The input feature zone data is a shapefile comprised of a select few polygons.
The zone field selected is FID.
The input value raster is a .tiff file which stores the correct values for the output.
The output table is saved to the workspace chosen by user.
And the statistic performed is the MEAN. Which means the output for every FID row is an average of all pixel values in the raster file within that polygon shape.

What's happening when the model runs is that the output is just incorrect....almost like there is no zonal recognition happening.
Is anyone familiar with this and perhaps what I can do to make the output correct??
Thanks

Esri analyze tool only works well under the projected GRID data format (ESRI raste format).
So convert it to grid format first.

ShikoNjuno

Hi Shiko,

How are you selecting the polygons?
Are you sure the selection is being passed to the next tool?
How is your output incorrect? Are the values wrong or all records Null?

Always consider attaching a snapshot of the model with the forum post. It makes it easier to see the upstream and downstream processes.

Shitij I've actaully been struggling with that, which is why I haven't. Maybe you could walk me through the process.
What I was trying to do is copy the model by doing a PrintScreen and then converting the image to a JPEG or other image file but to no success. How do you usually do it?

ShitijMehta

Hi Shiko,

How are you selecting the polygons?
Are you sure the selection is being passed to the next tool?
How is your output incorrect? Are the values wrong or all records Null?

Always consider attaching a snapshot of the model with the forum post. It makes it easier to see the upstream and downstream processes.

ShikoNjuno

I dont know if the tiff input is a problem? I usually use grids.
Also, what about cell size? What is the cell size of the tiff, then have you set the environment up for your analysis?

Well, the cells are pretty tight.....60 by 60.
What environment settings in particular should I adjust? Right now I only have the Workspace environment setting checked...I'm not quite sure about the rest.

AliciaMein

I dont know if the tiff input is a problem? I usually use grids.
Also, what about cell size? What is the cell size of the tiff, then have you set the environment up for your analysis?

Připojte se k diskusi

Přihlášení členové mohou přispívat, sledovat aktualizace a další. Jste tu noví? Zaregistrujte si bezplatný účet.

Rychlé odkazy

Community Help
Find useful guides, FAQs, and documents to help you navigate and make the most of Esri Community.