PROTOCOL

ESA CCI LAND COVER

Merging protocol for ESA CCI LC + LUH

1. Aggregation from 300m to 0.25°

The ESA CCI land cover dataset is distributed at 300m (1/360°) resolution (source grid) and each individual grid cell is associated with only one class. Aggregation to 0.25° resolution (target grid) is done by counting the number of source grid cells of each class inside a given target grid cell, accounting by its areas. Each target grid cell precisely envelops a 90x90 block of source grid cells. Thus no partial contribution of source grid cell to several target grid cells occurs and computation is very fast and straightforward. The earlier ESA CCI LC dataset (v2.0.7) is represented in 38 ESA classes, so the aggregation is perfomed with a dedicated user tool which takes into account the cross-walking table defining the correspondance between ESA classes and generic plant functional types (PFTs). The recent ESA CCI LC datasets (v2.0.8 and v2.0.82) are represented directly in generic PFTs, so no transformation of classes is needed for these products.

2. Merging principles

The merging of ESA CCI LC and LUH datasets for each given year is done in the following order:

Impose non-vegetated PFTs from ESA CCI LC (ESA_BARE, ESA_SNOWICE, ESA_WATER-INLAND, ESA_WATER-OCEAN). For historical reconstruction (850-1991) the data from 1992 are used, for future projection (2021-2100) the data from 2020 are used, for other years (1992-2020) the corresponding year data are used.

PFT_BARE = ESA_BARE

PFT_SNOWICE = ESA_SNOWICE

PFT_WATER-INLAND = ESA_WATER-INLAND

PFT_WATER-OCEAN = ESA_WATER-OCEAN

Impose anthropogenic PFTs from LUH (using the corresponding year data):

PFT_CROP-C3 = LUH_C3ANN + LUH_C3PER + LUH_C3NFX = LUH_C3

PFT_CROP-C4 = LUH_C4ANN + LUH_C4PER = LUH_C4

PFT_PASTURE = LUH_PASTR

PFT_BUILT = LUH_URBAN

Each added LUH class data is limited so that the resulting total does not exceed 1.

Calculate resulting imbalance:
IMB = 1 – PFT_BARE – PFT_SNOWICE – PFT_WATER-INLAND – PFT_WATER-OCEAN – PFT_CROP-C3 – PFT_CROP-C4 – PFT_PASTURE – PFT_BUILT
and distribute it between natural PFTs using the present-day fractions in ESA CCI LC data calculated from 1992 for historical reconstruction, from 2020 for future projections or from the corresponding year for a year within 1992-2020.

3. C3/C4-partitioning

The fraction of C4 grass (f_C4) is calculated as:

f_C4 = ESA_GRASS-NAT-C4 / (ESA_GRASS-NAT-C3 + ESA_GRASS-NAT-C4), if ESA_GRASS-NAT-C4 > 0

f_C4 = ESA_GRASS-MAN-C4 / (ESA_GRASS-MAN-C3 + ESA_GRASS-MAN-C4), if ESA_GRASS-NAT-C4 = 0 and ESA_GRASS-MAN-C4 > 0

f_C4 = 0, if ESA_GRASS-NAT-C4 = 0 and ESA_GRASS-MAN-C4 = 0

4. Special accounting of grass content

Anthropogenic and natural grass PFTs are only partly represented by crops and pasture, for the residual grass part we calculate the difference ESA_GRASS-NAT+ESA_GRASS-MAN-LUH_C3-LUH_C4-LUH_PASTR and use this portion to define the residual present-day fraction of grass. This residual is calculated in two steps. First, we caclulate anthropogenic part as ANT=LUH_C3+LUH_C4+LUH_PASTR-ESA_GRASS-MAN and limit it to zero if the ESA managed grasses amount is larger than the total anthropogenic part of LUH. Second, we calculate natural part as NAT=ESA_GRASS-NAT-ANT and limit it to zero if resulting ANT is larger than grass amount of ESA:

LUH_ANT = LUH_C3 + LUH_C4 + LUH_PASTR

GRASS_ANT = max(0, LUH_ANT – ESA_GRASS-MAN-C3 – ESA_GRASS-MAN-C4)

GRASS_NAT = max(0, ESA_GRASS-NAT-C3 + ESA_GRASS-NAT-C4 – GRASS_ANT)

GRASS_C3 = GRASS_NAT · (1 – f_C4)

GRASS_C4 = GRASS_NAT · f_C4

5. Present-day fractions

The calculation of present-day fractions of natural part between generic PFTs is done in the following way:

TOTAL_NAT = ESA_TREE-BE + ESA_TREE-BD + ESA_TREE-NE + ESA_TREE-ND + ESA_SHRUB-BE + ESA_SHRUB-BD + ESA_SHRUB-NE + ESA_SHRUB-ND + GRASS_C3 + GRASS_C4

f_TREE-BE = ESA_TREE-BE / TOTAL_NAT

f_TREE-BD = ESA_TREE-BD / TOTAL_NAT

f_TREE-NE = ESA_TREE-NE / TOTAL_NAT

f_TREE-ND = ESA_TREE-ND / TOTAL_NAT

f_SHRUB-BE = ESA_SHRUB-BE / TOTAL_NAT

f_SHRUB-BD = ESA_SHRUB-BD / TOTAL_NAT

f_SHRUB-NE = ESA_SHRUB-NE / TOTAL_NAT

f_SHRUB-ND = ESA_SHRUB-ND / TOTAL_NAT

f_GRASS-C3 = GRASS_C3 / TOTAL_NAT

f_GRASS-C4 = GRASS_C4 / TOTAL_NAT

Note that for the grass fractions the residual grass amount calculated in point 4 is used (highlighted with bold font).

6. PFT reconstruction

Finally, the resulting generic PFTs are obtained by distributing the imbalance (as defined in point 2) with respect to the pre-calculated fractions:

PFT_TREE-BE = f_TREE-BE · IMB

PFT_TREE-BD = f_TREE-BD · IMB

PFT_TREE-NE = f_TREE-NE · IMB

PFT_TREE-ND = f_TREE-ND · IMB

PFT_SHRUB-BE = f_SHRUB-BE · IMB

PFT_SHRUB-BD = f_SHRUB-BD · IMB

PFT_SHRUB-NE = f_SHRUB-NE · IMB

PFT_SHRUB-ND = f_SHRUB-ND · IMB

PFT_GRASS-C3 = f_GRASS-C3 · IMB

PFT_GRASS-C4 = f_GRASS-C4 · IMB

PFT_CROP-C3 = LUH_C3

PFT_CROP-C4 = LUH_C4

PFT_BARE = ESA_BARE

PFT_BUILT = LUH_URBAN

PFT_SNOWICE = ESA_SNOWICE

PFT_WATER-INLAND = ESA_WATER-INLAND

PFT_WATER-OCEAN = ESA_WATER-OCEAN

PFT_PASTURE = LUH_PASTR

Note that crops, pasture and urban data (highlighted with bold font in above equations) are imposed from LUH data of a corresponding year, whereas bare, snow, ice and water areas are imposed from ESA CCI LC data of a reference year.

7. Finalization

By the end of the merge certain pixels could have unclosed balance. It happens only if the present-day fractions are all-zero, which is possible if in the merged ESA CCI LC map there are no trees and shrubs and if the residual grass part is calculated to be zero. In this case the remained imbalance is put to the natural grass (PFT_GRASS-C3 and PFT_GRASS-C4 proportionally to C3/C4-fractions).