PRISMATIC-Initialization

Generating initial conditions for FATES

Install

conda env create -f environment.yml

Troubleshoots

For MacOS M1/M2 users:

• It does not work with R arch arm64, so you will need to reinstall R arch x86_64, follow this guide here.
• Library not loaded: /opt/X11/lib/libX11.6.dylib: run this command: brew install xquartz --cask
  • If you need to install brew, run this command: $ /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

GDAL, OSR:

• To import GDAL, use: from osgeo import gdal, osr

Cannot install leafR:

• sudo aptitude install libgdal-dev

Usage

conf/paths/paths.yaml: you may need to update the path.

• data_raw_inv_path: where to save inventory data.
• data_raw_aop_path: where to save raw NEON Airborne Observatory Platform (AOP) data.
• data_int_path: where to save processed data.
• data_final_path: where to save final-formatted data.

conf/sites/sites.yaml: you may need to add other sites if you want to process those.

Data procesing workflow: Workflow diagram of functions

• For each site, download NEON data:
  • download_lidar: discrete return lidar
  • download_hyperspectral: tiles or flightlines
  • download_veg_structure_data: woody veg data
  • download_polygons: polygons for woody veg plots
  • download_trait_table: table of trait values for species
• For all sites at once, generate_pft_reference: generates table linking species
• For one site at a time, process intermediate data products:
  • prep_veg_structure: organize woody veg data into year and sampling effort
  • prep_polygons: identify subplot associated with woody veg data
  • normalize_laz: normalize laz tiles
  • clip_lidar_by_plots: clip the laz/tif files given plots generated from prep_polygons
  • prep_lad: generate leaf area density profile stratified by size class
  • prep_biomass: calculate biomass for all invidivuals in NEON data
  • if using hyperspectral flightlines rather than tiles
    • correct_flightlines: apply BRDF and topographic corrections and convert corrected flightlines into tiles
  • prep_manual_training_data: clean manually delineated tree crowns with plant functional type (PFT) labels
  • prep_aop_imagery: prepare NEON AOP-derived rasters for PFT classifier
  • extract_spectra_from_polygon: Extract features (remote sensing data) for each pixel within the specified shapefile (tree crown for training, plot for prediction)
• For all sites at once, train_pft_classifier: train PFT classifier
• For a single target site and year, generate_initial_conditions: generate FATES initial conditions (cohort and patch files) by combining forest structure and composition

We generate FATES intital conditions in three types:

• ic_type == field_inv_plots: initialization from NEON forest inventory plots
• ic_type == rs_inv_plots: initialization from remote sensing data over NEON forest inventory plots
• ic_type == rs_random_plots: initialization from remote sensing data over plots randomly generated across entire NEON site

The final result is at data_final_path/site/year/ic_type.

# run all sites with default params
python main.py

# force prep_biomass to rerun for all sites/years
python main.py sites.global_run_params.force_rerun.prep_biomass=True

# run only SJER
python main.py sites.global_run_params.run=SJER

# run SJER and SOAP
python main.py sites.global_run_params.run='[SJER, SOAP]'

# run only SJER, force to rerun prep_biomass on SJER
python main.py sites.global_run_params.run=SJER sites.SJER.2019.force_rerun.prep_biomass=True