BSRLC

Baltic Sea Region Land Cover scripts

Requirements

• FORCE ARD tiled data (https://github.com/davidfrantz/force) in Fold-by-week format (FBW) must be available in the FORCE_datacube_fbm. Each tile muss contain the following bands:
  • {year}-{year}_001-365_HL_TSA_LNDLG_BLU_FBW.tif
  • {year}-{year}_001-365_HL_TSA_LNDLG_GRN_FBW.tif
  • {year}-{year}_001-365_HL_TSA_LNDLG_NDV_FBW.tif
  • {year}-{year}_001-365_HL_TSA_LNDLG_NIR_FBW.tif
  • {year}-{year}_001-365_HL_TSA_LNDLG_RED_FBW.tif
  • {year}-{year}_001-365_HL_TSA_LNDLG_SAV_FBW.tif
  • {year}-{year}_001-365_HL_TSA_LNDLG_SW1_FBW.tif
  • {year}-{year}_001-365_HL_TSA_LNDLG_SW2_FBW.tif
• {year} must be singular year (e.g 2015-2015) not two different years (e.g 2014-2015)

1. Creat the python environment via conda

conda create --name BSRLC --file req.txt

2. Activate environment

conda activate BSRLC

3. Run the mapping from pretrained models

For example, to predict maps for tile X0068_Y0035 in year 2015, run:

python mapping.py --tile X0068_Y0035 --year 2015

The results are stored in the Maps directory with the same datacube structure as the FORCE_datacube_fbm.