###	Marine predator Areas of Ecological Significance (AES) under two global climate change scenarios
    
###	Data downloaded from the SCAR DistAnt Ecological Model Output Repository https://zenodo.org/records/14792295/files/hindell_et_al-2020.zip on 2026-08-29
    Hindell M.A., Reisinger R.R., Ropert-Coudert Y., et al (2020) Tracking of marine predators to protect Southern Ocean ecosystems. Australian Antarctic Data Centre. https://data.aad.gov.au/metadata/records/fulldisplay/SCAR_RAATD
    
### Processing steps:

    1.  Download Hi2023-aes_colony_weighted_cog, Hi2023-aes_colony_weighted-RCP45_cog.tif and Hi2023-aes_colony_weighted-RCP85_cog.tif from https://zenodo.org/records/14792295/files/hindell_et_al-2020.zip and file in cwd/input
        
    2.  Run 02_create_stack_AES.py
        * Add a cyclic overlap at +/-180° longitude boundary to all three source rasters before reprojection
        * Reproject the continuous present-day colony-weighted habitat importance raster from EPSG:4326 to EPSG:3031 using bilinear resampling
        * Derive present-day AES from the upper 10% of valid reprojected habitat importance values
        * Reproject the RCP4.5 and RCP8.5 model-count rasters from EPSG:4326 to the same EPSG:3031 grid using nearest-neighbour resampling
        * Convert all NoData values to -999
        * Create gain/loss bands showing the number of climate models predicting AES gain or retention
        * Stack results into five-band raster RAATD_marine_predators_AES_GCM_predictions.tif:

            Band 1: present_AES
                    0 = non-AES
                    1 = AES

            Band 2: RCP45_AES_modelCount
            Band 3: RCP85_AES_modelCount
                    0-8 = number of the eight climate models predicting AES-like conditions
                    0 = no models predict AES-like conditions
                    8 = all models predict AES-like conditions

            Band 4: RCP45_gainloss
            Band 5: RCP85_gainloss

                    For cells that are not present-day AES:
                        0-8 = number of models predicting gain of AES-like conditions
                        0 = no models predict gain
                        8 = all models predict gain

                    For cells that are present-day AES:
                        10-18 = 10 + number of models predicting retention of AES
                        10 = no models predict retention (8/8 predict loss)
                        18 = all models predict retention (0/8 predict loss)

                    Value 9 is unused to separate gain and retention/loss classes.