the CLMS has been producing a new dataset from 2018 onward entitled the CORINE Land Cover (CLC)+ Back Bone raster which also include a sealed area thematic class. When comparing the IMD and CLC+ Backbone Raster datasets with sampled reference data, it appears that the imperviousness dataset substantially underestimates sealed areas at European level see Sannier et al., 2024 ( https://doi.org/10.3390/land13071087 ) and ETC/DI, 2024.

The CLC+ dataset only started to be available from 2018 and currently does not include any change layer. To address these issues, a combined, harmonized and bias-corrected soil sealing dataset for Europe was produced by the EEA for the entire observation period.

For the geometric harmonization of the CLMS IMD time series for 2006, 2009, 2012, 2015 and 2018 the following steps were used:

1. Create Sealed 2018 status by extracting class 1 from CLC+ Backbone 2018 (10m) (set all other classes to 0).

2. Create the Sealed 2015 accounting status layer by combining Sealed 2018 status with IMCC1518accounting:

a. Resample 20m resolution IMCC1518accounting to 10m resolution.

b. Combine IMCC1518accounting sealing increase with Sealed 2018 status: IF Sealed 2018 status is sealed (value=1) and IMCC1518accounting is class “new cover” THEN write new Sealed 2015 accounting status layer with value “unsealed” (unsealed=0) at corresponding grid cell.

c. Combine IMCC1518accounting sealing decrease with Sealed 2018 status: IF Sealed 2018 status is not sealed (value=0) and change is “loss of cover” THEN write new Sealed 2015 accounting status layer with value “sealed” (sealed=1) at corresponding grid cell.

d. For the rest of the pixel copy the values (sealed/unsealed, i.e. 1/0) from the Sealed 2018 status layer

3. Create the Sealed 2012 accounting status layer by combining Sealed 2015 status with IMCC1215

a. Resample 20m resolution IMCC1215 to 10m resolution

b. Combine CLMS IMCC1215 sealing increase with Sealed 2015 accounting status: IF Sealed 2015 accounting status is sealed (value=1) and CLMS IMCC1215 is class “new cover” THEN write new Sealed 2012 accounting status layer with value “unsealed” (unsealed = 0) at corresponding grid cell.

c. Combine CLMS IMCC1215 sealing decrease with Sealed 2015 status: IF Sealed 2015 status is not sealed (value =0) and change is “loss of cover” THEN write new Sealed 2012 status layer with value “sealed” (sealed=1) at corresponding grid cell.

d. For the rest of the pixel copy the values (sealed/unsealed, i.e. 1/0) from the Sealed 2015 status layer

4. Create the Sealed 2009 accounting status layer by combining Sealed 2012 status with IMCC0912

a. Resample 20m resolution IMCC0912 to 10m resolution

b. Combine CLMS IMCC0912 sealing increase with Sealed 2012 accounting status: IF Sealed 2012 accounting status is sealed (value=1) and CLMS IMCC0912 is class “new cover” THEN write new Sealed 2009 accounting status layer with value “unsealed” (unsealed = 0) at corresponding grid cell.

c. Combine CLMS IMCC0912 sealing decrease with Sealed 2012 status: IF Sealed 2012 status is not sealed (value =0) and change is “loss of cover” THEN write new Sealed 2009 status layer with value “sealed” (sealed=1) at corresponding grid cell.

d. For the rest of the pixel copy the values (sealed/unsealed, i.e. 1/0) from the Sealed 2012 status layer

5. Create the Sealed 2006 accounting status layer by combining Sealed 2009 status with IMCC0609

a. Resample 20m resolution IMCC0609 to 10m resolution

b. Combine CLMS IMCC0609 sealing increase with Sealed 2009 accounting status: IF Sealed 2009 accounting status is sealed (value=1) and CLMS IMCC0609 is class “new cover” THEN write new Sealed 2006 accounting status layer with value “unsealed” (unsealed = 0) at corresponding grid cell.

c. Combine CLMS IMCC0609 sealing decrease with Sealed 2009 status: IF Sealed 2009 status is not sealed (value =0) and change is “loss of cover” THEN write new Sealed 2006 status layer with value “sealed” (sealed=1) at corresponding grid cell.

d. For the rest of the pixel copy the values (sealed/unsealed, i.e. 1/0) from the Sealed 2009 status layer

For further processing the time series was converted to 1km grid size which was the best compromise in terms of reporting on sealed soil surface and available resources. In the next step a visually interpreted reference database was created, where within 1kmx1km primary sampling units (PSUs) 100 secondary sampling units (SSUs) were selected on a regular grid. With a non-linear regression estimate a relationship was established between the visually interpreted ground truth and the 1km harmonized layers. The regression estimate was applied to the harmonized layers and a new 1km harmonized and corrected time series was created.