1

I am a bit confused, as during my studies I learnt that rasters have only one cell value and for some it might be true, as the attribute table is greyed out in ArcGIS. But for others, multiple fields are provided, so the assumption seems to be wrong.

Can anyone explain me the difference of rasters that can contain one or multiple fields?

In ArcGIS an attribute table can be created for raster, but only for integer raster, not for float raster (see the "Build Raster Attribute Table" function).

  • If you have consulted the Build Raster Attribute Table help, then I think that would be worth mentioning in your question. – PolyGeo Mar 17 '16 at 7:10
1

The difference you are observing is due to the raster cell type: integer or float. Integer rasters are the only raster type that can have more columns than just the VALUE column. Rasters of type float do not have attribute tables. So if the button is greyed out it will be due to the fact it is a float raster. Just go to the layer properties > source to prove it to yourself.

  • Thanks for the Information! But is this only implemented like this in ArcGIS or is this a general characteristic of rasters as a data model? – the_chimp Mar 16 '16 at 6:00
  • So if someone asks me, if a raster can have only one attribute value or multiple - can I say usually the raster model expects one cell value, but it depends on the implementation, like in ArcGIS float raster can only have one field, but integer raster multiple fields ? – the_chimp Mar 17 '16 at 10:48
  • 1
    The attribute table for an integer raster is a sort of lookup table, it does not have a row per pixel, it is a summary of raster value and the number of cells for that value. So a raster with a million pixels but only two values e.g. 1 or 0 can have an attribute table with only two rows. As you can create a table for an integer grid you can treat it like any other table, including adding more fields, so you could have a text description of what 1 and 0 actually mean (e.g. land or water). I would imagine other GIS systems operating in a similar way? – Hornbydd Mar 17 '16 at 11:26
0

To convert the values you can multiply it with 100 or 1000 depending upon the decimal places of the pixel values, using raster calculation. Then you can build attribute using DataManagement->Raster->RasterProperties->BuildRasterAttributeTable. Once you get the raster with attributes then again divide the values with 100 or 1000 using raster calculation or FieldCalculator.

0

I am a bit confused, as during my studies I learnt that rasters have only one cell value and for some it might be true, as the attribute table is greyed out in ArcGIS. But for others, multiple fields are provided, so the assumption seems to be wrong.

I had similar questions about Raster Attribute Table (RAT)s What is a raster attribute table?.

I think one cause of the confusion is that the raster attribute table for a raster is not really an "attribute table" in the sense of attributes in the vector data model. In the vector model, the attributes are for each individual object/row. So if you have 100 geometric elements in your vector layer, you will have 100 rows in the attribute table.

In contrast, in RATs, you don't have a record (set of attribute) for each pixel/cell. (And indeed, people don't/can't typically store very complex attributes for each pixel.) Instead, you have a record for each cell value. So if you have 100 pixels in a binary (0/1) raster/image, you will have 2 rows for each value. (If the "attribute" in the RAT is a count of pixels of a particular value, e.g. 0 or 1, your RAT is really just a histogram of the image, which is a summary of the pixels.) This is probably why RATs are also called Value Attribute Tables (VAT), which is a much less confusing name IMO.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.