For research use only. Not for clinical decisions. Copyright (c) 2026, Payton Jachec

S2Down

A pipeline for downloading tiles, cleaning, and computing standard geospatial analyses (NDVI, NDWI, SAVI, MNDWI, DEM) over large extents without API keys or access credentials.

S2Down Contents

• download_rgb.py
  • Downloads spectral bands (all except DEM) from planetary-computer for the specified extent.
• download_dem.py
  • Downloads DEM tiles for specified extent.
• clean_tiles.py
  • Removes or moves unfinished or broken tiles (allows dry-runs).
• median_composite.py
  • Collects tiles and computes median mosaic in chunks for resource management.
• compute_indices.py
  • collects mosaics from median_composite.py and computes analysis indices.
• stitch_dem.py
  • Stitches DEM tiles together.
• environment.yml
  • Dependencies list for virtual environment.

Required Packages and Environment

Known Working Configuration

Package	Version
arrow	1.4.0
bokeh	3.9.0
dask	2026.3.0
distributed	2026.3.0
mgrs	1.5.4
numpy	2.4.6
pandas	3.0.3
planetary-computer	1.0.0
pystac-client	0.9.0
pystac	1.14.3
python	3.11.15
rasterio	1.4.4
scipy	1.17.1
stackstac	0.5.0
xarray	2026.4.0

Required Channels

• conda-forge

Environment Setup

It is recommended to use Anaconda or Mamba to create a venv for this program suite. If running on an HPC, default configurations will almost certainly break the code.

Suggested Environment

~] $ micromamba create -n ENV_NAME python=3.11.15 -c conda-forge bokeh dask distributed mgrs numpy pandas>=2.2.3 planetary-computer pystac-client rasterio scipy stackstac xarray

or using the included environment.yml

~] $ micromamba create -n ENV_NAME environment.yml

Job Submission

download_rgb.py

To submit a job using this script, open the script and modify the following fields:

Field	Line	Use
OUTPUT_DIR	22	Specifies desired output directory
BANDS	23	Selects bands to download over the ROI
TARGET_RES	24	Resolution to resample to (set to 10 to turn off)
START_DATE	25	First Date to search for tiles inside the ROI
END_DATE	26	Last date to search for tiles inside the ROI
MAX_CLOUD_COVER	27	Sets cloud cover max. percentage
ROI_BOX	28	Modifies the bounds of the extent
WORKERS	35	Number of workers for Dask array
THREADS	36	Number of threads/worker
MEMLIMIT	37	Working memory limit for workers

Then, simply:

~] $ python3 download_rgb.py

download_dem.py

This is done similarly to download_rgb.py, but notably the DEM does not have any cloud cover, bands, or timescale, so there are less fields to change in download_dem.py:

Field	Line	Use
BASE_DIR	14	Sets base directory for current location of all tiles
DEM_DIR	15	Specifies subdirectory name for DEM tiles
WEST	17	Sets western extent
SOUTH	18	Sets southern extent
EAST	19	Sets eastern extent
NORTH	20	Sets northern extent

Then, simply:

~] $ python3 download_dem.py

clean_tiles.py

Similarly to the download scripts, this script requires you to change some values within it to run properly.

Field	Line	Use
DEFAULT_TILES_DIR	17	Changes default master imagery directory
REQUIRED_BAND_PATTERNS	21	Add, remove, or change entries to change required image patterns

Then, simply:

~] $ python3 clean_tiles.py

Flags

1. --tiles-dir /PATH/TO/TILES/DIR

   • Specifies alternative filepath to master tiles folder default is "./sentinel2_tiles"

2. --dry-run

   • Validates code before moving all unfinished folders to "incomplete"

median_composite.py

This script does not have a config section and should be submitted with the following flags:

Flag	Req./Opt.	Use
--tiles_dir	Req.	Sets path to parent tile directory
--out_dir	Req.	Sets output directory
--chunk_size	Opt.	Sets chunk side length in pixels
--n_workers	Opt.	Sets number of workers for parallel processing
--memory_limit	Opt.	Sets RAM limit per worker
--nodata	Opt.	Sets default value for pixels with no data
--scale	Opt.	Sets scale for input images to scale down to 0-1 range
--keep_chunks	Opt.	If used, chunks will not be deleted

Example submission:

~] $ python3 median_composite.py --tiles_dir /PATH/TO/TILES/DIRECTORY/ --out_dir /PATH/TO/OUTPUT/DIRECTORY/
--chunk_size 4096 --n_workers 4 --memory_limit 12GB --nodata 0 --scale 10000 --keep_chunks

compute_indices.py

This script computes the analysis indices from the mosaiced images created by median_composite.py

Field	Line	Use
BAND_PATHS	14-19	Sets filepaths to individual band images
OUT_DIR	23	Sets output directory
COMPUTE	26-32	Selects indices to compute
MNDVI_C1	36	Sets multiplier for MNDVI numerator
MNDVI_C2	37	Sets multiplier for MNDVI denominator
SAVI_L	40	Sets SAVI L value
SABI_L	43	Sets SABI L value
CHUNK_ROWS	46	Sets chunk (row) height
NODATA_OUT	47	Sets value to fill for nodata pixels
CLAMP	48	Toggles clamping on/off
COMPRESS	49	Sets compression method
OVERWRITE	50	Toggles overwriting existing output files

Then,

~] $ python3 compute_indicies.py

stitch_dem.py

This script stitches the DEM tiles downloaded with download_dem.py together into a composite image.

Field	Line	Use
DEM_DIR	22	Sets DEM tile source directory
OUT_PATH	25	Sets output path
NODATA_IN	28	Sets default value to fill for NaN pixels on input
NODATA_OUT	29	Sets default value to fill for NaN pixels on output
OUT_DTYPE	33	Sets output data format
COMPRESS	36	Sets compression method
OVERWRITE	39	Toggles overwriting of existing files in output directory

Then,

~] $ python3 stitch_dem.py

For research use only. Not for clinical decisions. Copyright (c) 2026, Payton Jachec