Gdal_rasterize nodata_value does not work

Question

I want to rasterize a source layer made of polygons. No matter what value I assign to “NoData_value”, the outcome in my array is always = 0.
I am using python 3.4.
Can anyone help me understand please ?

source_srs = source_layer.GetSpatialRef()
x_min, x_max, y_min, y_max = source_layer.GetExtent()

# parameters of output file
xSize = math.ceil((x_max - x_min) / pixel_size)  # number of pixels in the x direction given pixel_size, rounded up to the nearest pixel
ySize = math.ceil((y_max - y_min) / pixel_size)  # number of pixels in the y direction given pixel_size, rounded up to the nearest pixel
x_res = int((x_max - x_min) / xSize)  # size of pixel in meters rounded to fit bbox
y_res = int((y_max - y_min) / ySize)  # size of pixel in meters rounded to fit bbox

NoData_value = -9999

# Create output dataset as memory
target_ds = gdal.GetDriverByName('MEM').Create('', xSize, ySize, gdal.GDT_Byte)
target_ds.SetGeoTransform((x_min, x_res, 0, y_max, 0, -y_res))
wkt_projection = source_srs.ExportToWkt()
target_ds.SetProjection(wkt_projection)

band = target_ds.GetRasterBand(1)
band.SetNoDataValue(NoData_value)

# rasterize
gdal.RasterizeLayer(target_ds, [1], source_layer, options=["ATTRIBUTE=expo" ])

# Read as numpy array
array = band.ReadAsArray()

Asked By: Marianne

||

Source

Answer 1

Do you want the values outside the polygons to have NoData_value?

Then, add

band.Fill(NoData_value)

before calling gdal.RasterizeLayer. You need this because gdal.RasterizeLayer only modifies (burn) the values inside the polygons.

You will need to change the format gdal.GDT_Byte to a format that can handle -9999 such as for example GDT_Float32:

target_ds = gdal.GetDriverByName('MEM').Create('', int(xSize), int(ySize), 1, gdal.GDT_Float32)

Note : data type is the 5th argument, not the 4th as in your code. The 4th argument should be used for the number of bands:
Link

I tested and it works on my size.

Answered By: seb007

Gdal_rasterize nodata_value does not work

Question:

Answers: