What does 'Area' in Zonal Statistics refer to?

[Willem_Bouw]

Happy new year everyone!

Upon running a Zonal Statistics Table, one of the parameters shown is 'Area'. I assumed that this gives the area quantified in the amount of pixels. However, when I re-calculate the area from pixels to hectares, it does not give a logical number. Which makes me doubt whether 'Area' is in pixels or I may have done something wrong elsewhere. Does someone know more about this?

[DanPatterson]

Zonal Statistics (Spatial Analyst)—ArcGIS Pro | Documentation

How zonal statistics tools work—ArcGIS Pro | Documentation

Has discussions on how the cell size is determined, specifically

When the cell size of the Input Raster or Feature Zone Data and the Input Value Raster is different, the output cell size will be the Maximum Of Inputs value, and the Input Value Raster will be used as the snap raster internally. If the cell size is the same but the cells are not aligned, the Input Value Raster will be used as the snap raster internally. Either of these cases will trigger an internal resampling before the zonal operation is performed.

Hence it is best to use projected raster data for both inputs, ensuring that the cell size and cell alignment are both equal and defined.

Also, use a comformal (eg UTM, state plane) or equal area projection... anything other than Web Mercator

[Willem_Bouw]

Thanks for the elaborate reply! Indeed I used UTM as a projected spatial reference which gave me pixels of 500x500 m (so 250.000 m2 --> 25ha for each pixel).

Now when using Zonal Statistics, the table shows a value in the column 'Area' that, upon calculated into hectares (so times 25), gives an abnormally high value for the actual area. That made me question whether 'Area' is actually indicated in pixels.

[DavidPike]

Your math seems fine.  Might be a good initial check if you sum the Area values in the table that should give the count of all Zone pixels in your extent. that sum value * 25 should give the total extent area (in Hectares) If you don't have any NoData etc.  How far off are your calculations?  Also easy to confuse the value raster with the zone raster.

Would you be able to share any screenshots etc. ?

[Willem_Bouw]

The area of interest is the Bahia Negra basin in Paraguay. The country Paraguay itself is about 40.675.200 ha, with the Bahia Negra basin being a smaller area in the North of the country.

Upon conducting Zonal Statistics and summing all area type 'Area' values in Excel, I get 9.638.255.260 pixels for the entire area. Now an error I found thanks to you is that I used the pixels of the Value Raster (500x500=25ha cells) rather than the Zone Raster (30x30=0.09ha cells). So thank you for making me realize my mistake 🙂 However, changing this, it would give the total area the size of 9.638.255.260pixels * 0.09ha = 867.442.973 ha, which is still too much. So there must be something else I do wrong.

Added are some screenshots of the Area values given in the Zonal Statistics table (value stands for land cover), and the land cover of the regarding map of the area of Bahia Negra

Land cover value 3 (green) and 15 (yellow) should be the only ones with a significant amount of Area. The other land cover values are barely represented in the map so it is odd that they are given this big amount of pixels in the Zonal Statistics Table

EDIT: I just realize that the column 'Count' gives the amount of pixels. So in that case, 'Count' should be considered as the area given in pixels, rather than the 'Area' column. That would also give more logical values. But what does the column 'Area' refer to then?

[DanPatterson]

What was the value?

check the math (Edited... @DavidPike good catch)

500m* 500m = 250,000 m^2 = 25 ha

which is a very large cell size.  What area are you covering?

[Willem_Bouw]

David made me realize a significant error I made: I used Value Raster pixel size (cells of 500x500m) rather than the Zone Raster pixel size (cells of 30x30m=0.09ha). Still though, my outcomes are too large so there should be something else I do wrong. See the comment I made on David's reply