How to polygonize raster with range of values?

Question

I have a raster with range 0 - 1. I want to polygonize them in QGIS according to their value to be later classified from "Very Low" to "Very High".

raster_value	possible_polygon_value	class
0-0.2	0	Very Low
0.2-0.4	1	Low
0.4-0.6	2	Medium
0.6-0.8	3	High
0.8-0.1	4	Very High

I am guessing there is a way to do it by using Raster calculator first to assign values to each pixel first. However, how? There is no "IF" and "ELSE" statement in raster calculator.

What I think:

IF ("raster@1" <= 0.2) * 0,
ELSE IF ("raster@1" > 0.2 AND "raster@1" <= 0.4) * 1,
ELSE IF ("raster@1" > 0.4 AND "raster@1" <= 0.6) * 2,
ELSE IF ("raster@1" > 0.6 AND "raster@1" <= 0.8) * 3,
ELSE IF ("raster@1" > 0.8) * 4

Answer 1

What you want to do is the re-classify the raster before polygonizing. You can GRASS tool r.recode for that (available in QGIS).

For a solution with raster calculator, see below.

Using GRASS r.recode

1. You need a simple text-file defining the classes. Just copy the text below in a file and save it as .txt file (you can use a text editor for that, as e.g. the simple Editor tool in Windows).

2. Run Menu Processing / Toolbox / r.recode, set the raster as input and the saved txt-file as File containing recode rules.

3. You get a new raster as output, containing just five values (categories): 1 to 5. Than polygonize this raster to get your polygons.

Paste this text in an otherwise empty txt file. The first two numbers, devidid by : define the range of values you want to group together to a new category, the last value is the numbering of the category:

0:0.2:1 
0.2:0.4:2
0.4:0.6:3
0.6:0.8:4
0.8:1:5

Screenshot: A simple raster with random-values form 0.009 to 0.997 (see histogram on the right side). Recoded and vectorized. In the resulting vector layer, category 5 = very high is highlighted with red outlines and labeled accordingly: as you can see (especially in the cases marked with arrows), the red outlined polygons contain pixels with different values (shades of gray), but in a certain range (0.8 to 1):

---

Edit: to add another solution that correpsonds more to your initial idea:

Using raster calculator

Run Menu Raster / Raster calculator, paste this expression and run it:

("raster@1" * 1000 <= 200 ) *1 +
("raster@1" * 1000 <= 400 ) *1 +
("raster@1" * 1000 <= 600 ) *1 +
("raster@1" * 1000 <= 800 ) *1 +
("raster@1" * 1000 <= 1000 )*1

Explanation: There is no direct conditional implemented, but you can imitate it. For that evaluate if the pixel-value in case is smaller/equal than 0.2: "raster@1" <= 0.2. This returns 1 for true, 0 for false. Multiply this by 1 and add further conditions in steps of 0.2: each time it is fulfilled, it adds 1. For pixel value 0.9, you thus have: 0+0+0+0+1= 1. Like this, you get an output-raster with 5 categories, inverse to the r.recode solution, category of smallest values getting highest category-code and the other way round.

However, raster calculator seems not to work properly with decimal numbers. Because of that, first multiply the values by 1000. Finally, polygonize as above.