This repository is a fork of https://github.com/langnico/global-canopy-height-model/ and contains the code used to create the results presented in the paper: A high-resolution canopy height model of the Earth. Here, we developed a model to estimate canopy top height anywhere on Earth. The model estimates canopy top height for every Sentinel-2 image pixel and was trained using sparse GEDI LIDAR data as a reference. In this fork with we fixed some small bugs, added some automation for canopy estimation in Brazil biomes. Now you can choose the AOI of the place you want to predict the canopy height. Also we added GPU parallelization support for the inference.
canopy-height-low.mp4
- Data availability
- Installation and credentials
- Loading the model
- Deploying
- Training
- ALS preprocessing for independent comparison
- Citation
This is a summary of all the published data:
- Global canopy top height map for 2020 (ETH Research Collection)
- Train-val dataset (ETH Research Collection)
- Rasterized canopy top height models from airborne lidar (Zenodo)
- Trained model weights (SciCrop release)
- Demo data for example scripts (Zenodo)
- Sparse GEDI canopy top height data (Zenodo)
- ESA WorldCover 10 m 2020 v100 reprojected to Sentinel-2 tiles (Zenodo)
Please follow the instructions in INSTALL.md.
from gchm.models.xception_sentinel2 import xceptionS2_08blocks_256
# load the model with random initialization
model = xceptionS2_08blocks_256()Please see the example notebook on how to load the model with the trained weights.
This is a demo how to run the trained ensemble to compute the canopy height map for a Sentinel-2 tile (approx. 100 km x 100 km).
-
With Brazilian CAR code:
cd gchm python3 app.py --car {car-code} -
With Geojson:
cd gchm python3 app.py --aoi {geojson path} -
Just preparation:
cd gchm python3 app.py --prepareThis creates the following directories:
deploy_example/ ├── ESAworldcover │ └── 2020 │ └── sentinel2_tiles │ └── ESA_WorldCover_10m_2020_v100_32TMT.tif ├── image_paths │ └── 2020 │ └── 32TMT.txt ├── image_paths_logs │ ├── predictions_provided │ ├── sentinel2 │ └── sentinel2_aws trained_models/ └── GLOBAL_GEDI_2019_2020 ├── model_0 │ ├── FT_Lm_SRCB │ │ ├── args.json │ │ ├── checkpoint.pt │ │ ├── train_input_mean.npy │ │ ├── train_input_std.npy │ │ ├── train_target_mean.npy │ │ └── train_target_std.npy │ ├── args.json │ ├── checkpoint.pt │ ├── train_input_mean.npy │ ├── train_input_std.npy │ ├── train_target_mean.npy │ └── train_target_std.npy ├── model_1 │ ├── ... ├── model_2 │ ├── ... ├── model_3 │ ├── ... ├── model_4 │ ├── ...The checkpoint.pt files contain the model weights. The subdirectories
FT_Lm_SRCBcontain the models finetuned with a re-weighted loss function.
This demo script processes 10 images (from the year 2020) for the tile "32TMT" in Switzerland and aggregates the individual per-image maps to a final annual map.
Provide a text file with the image filenames per tile saved as ${TILE_NAME}.txt. The demo data contains the following file:
cat ./deploy_example/image_paths/2020/32TMT.txt
S2A_MSIL2A_20200623T103031_N0214_R108_T32TMT_20200623T142851.zip
S2A_MSIL2A_20200723T103031_N0214_R108_T32TMT_20200723T142801.zip
S2A_MSIL2A_20200812T103031_N0214_R108_T32TMT_20200812T131334.zip
...
The corresponding images are stored in ./deploy_example/sentinel2/2020/.
- Set the paths in
gchm/bash/config.sh - Set the tile_name in
gchm/bash/run_tile_deploy_merge.sh - Run the script:
bash gchm/bash/run_tile_deploy_merge.sh
The ESA WorldCover 10 m 2020 v100 reprojected to Sentinel-2 tiles is available on Zenodo. We apply minimal post-processing and mask out built-up areas, snow, ice and permanent water bodies, setting their canopy height to ”no data” (value: 255). See the script here.
Sentinel-2 images can be downloaded on the fly from AWS S3 by setting GCHM_DOWNLOAD_FROM_AWS="True"
and providing the AWS credentials as described above.
This was tested for 2020 data, but might need some update in the sentinelhub routine to handle newer versions.
- Download the train-val h5 datasets from here.
- Merge the parts file to a single
train.h5andval.h5by running this script. Before running it, set the variablesin_h5_dir_partsandout_h5_dirto your paths. Then run:bash gchm/preprocess/run_merge_h5_files_per_split.sh`
A slurm training script is provided and submitted as follows.
Before submitting, set the variable CODE_PATH at the top of the script and set the paths in gchm/bash/config.sh. Then run:
sbatch < gchm/bash/run_training.sh
In cases where rastered high-resolution canopy height models are available (e.g. from airborne LIDAR campaigns) for independent evaluation, some preprocessing steps are required to make the data comparable to GEDI canopy top height estimates corresponding to the canopy top within a 25 meter footprint.
- A rastered canopy height model with a 1m GSD should be created (E.g. using
gdalwarp). - The 1m canopy height model can then be processed with a circular max pooling operation to approximate "GEDI-like" canopy top heights. This step is provided as a pytorch implementation.
Example: Download the example CHM at 1m GSD from here. Then run:
python3 gchm/preprocess/ALS_maxpool_GEDI_footprint.py "path/to/input/tif" "path/to/output/tif"
Please cite the first authors of the paper if you use this code or any of the provided data.
Lang, N., Jetz, W., Schindler, K., & Wegner, J. D. (2023). A high-resolution canopy height model of the Earth. Nature Ecology & Evolution, 1-12.
@article{lang2023high,
title={A high-resolution canopy height model of the Earth},
author={Lang, Nico and Jetz, Walter and Schindler, Konrad and Wegner, Jan Dirk},
journal={Nature Ecology \& Evolution},
pages={1--12},
year={2023},
publisher={Nature Publishing Group UK London}
}