Fusing GEDI shots and annual forest change maps to estimate tropical forest recovery rates across the Amazon basin

This repository contains the code and analysis pipelines used in the manuscript, "Fusing GEDI shots and annual forest change maps to estimate tropical forest recovery rates across the Amazon basin."

Requirements

• Python 3.9+
• PostGIS 14+
• Java

Data availability

This project uses the following data sources:

Data source	Availability	Sensors	Data	Date range	No. observations (used)	Area covered
GEDI Level 4A (v002)	public	Space-borne LiDAR	Full-waveform LiDAR	2019-2020	150 Mio.	7.4 Mio. ha
EU Joint Research Council (JRC) Annual Change dataset	public	Space-borne multispectral imagery (Landsat)	Landcover Raster	1992-2020	-

Project Organization

├── LICENSE
├── Makefile           <- Makefile with commands like `make init` or `make lint-requirements`
├── README.md          <- The top-level README for developers using this project.
|
├── notebooks          <- Jupyter notebooks. Naming convention is a number (for ordering),
|   |                     the creator's initials, and a short `-` delimited description, e.g.
|   |                     `1.0_jqp_initial-data-exploration`.
│   ├── exploratory    <- Notebooks for initial exploration.
│
│
├── requirements       <- Directory containing the requirement files.
│
├── setup.py           <- makes project pip installable (pip install -e .) so src can be imported
├── biomassrecovery    <- Source code for use in this project.
│   ├── __init__.py    <- Makes src a Python module
│   │
│   ├── data           <- Functions to download and parse data
│   │
│   ├── processing     <- Functions to aggregate, overlay, and run data models
|   |
│   ├── spark          <- Spark-based end-to-end processing pipeline scripts
│   │
│   └── utils          <- Common utilities
│
└── setup.cfg          <- setup configuration file for linting rules

General Guide to the project

The above diagram shows an overflow of the steps taken to generate the results. For those looking to reuse this code, note that you will need to take the following steps:

BASIC SETUP

Fork this repository, create a new python environment, and install the project requirements, including python packages in requirements/requirements.txt.

Set appropriate environment variables: Change biomassrecovery/environment.py to contain the database location and password for your database. Set the file paths to appropriate values. Create a file called .env in your home directory with the following format:

# Earthdata access
EARTHDATA_USER="<earthdata-username-here>"
EARTHDATA_PASSWORD="<earthdata-password-here>"
EARTH_DATA_COOKIE_FILE="~/.urs_cookies"

# User path constants
USER_PATH="<path-to-home-dir-containing-repo>"
DATA_PATH="<path-to-store-data>"

# Database constants
DB_HOST="<db-hostname>"
DB_NAME="<db-name>"
DB_USER="<db-user>"
DB_PASSWORD="<db-password>"

DATA DOWNLOAD AND SETUP

GEDI data: The GEDI data is to be stored in a PostGIS database. Set up a new PostGIS database with the schema found in schema.sql. Download the GEDI data for your region of interest and ingest it into PostGIS with the script found at biomassrecovery/spark/gedi_download_pipeline.py. Note that this requires an existing earthdata account.

JRC data: Download the JRC rasters for your area of interest from the publicly available dataset.

DATA PROCESSING

Run the overlay script, found at biomassrecovery/spark/gedi_recovery_analysis_pipeline_monte_carlo.py Run with --help to see all of the arguments required.

DATA ANALYSIS/FIGURES

All results can now be found in the directory you specified using the --save_dir flag for the overlay script. The results include all of the output for a full monte carlo sample, which is probably more data than you are interested in. To get a summary of the data, aggregated by mean AGBD and mode recovery age across the monte carlo sample, use the notebook exploratory/13-ah2174-gedi-monte-carlo-preprocessing.ipynb To reproduce the figures in the paper, use the notebook 28-ah2174-gedi-recovery-figures.ipynb This will also give you an idea of how to load and manipulate the output results to perform other analyses.