GRDTILER

Python package for processing, normalising, and tiling Synthetic Aperture Radar (SAR) datasets. Supports Sentinel-1 (S1), Radarsat-2 (RS2), and RCM formats.

---

Features

• Grid tiling — sliding-window tiles with configurable size, overlap, and centering
• Point-based tiling — tiles centred around geographic coordinates
• Wave-mode tiling — single-tile extraction for Sentinel-1 WV products
• Sigma0 detrending — GMF-based detrending via xsarsea (CMOD5N, S1_V2, …)
• Multi-mission support — S1, RS2, RCM with per-mission GMF configuration
• Footprint computation — WKT polygon + centroid added to every tile
• NetCDF output — hierarchical directory structure keyed by mode / size / resolution / date

---

Installation

Conda (recommended)

conda install grdtiler -c oceanscope

From source

git clone https://github.com/jean2262/grdtiler.git
cd grdtiler
pip install -e .          # runtime dependencies only
pip install -e ".[dev]"   # + pytest / pytest-cov

---

Usage

1. Grid tiling — tiling_prod

Slides a window across the full dataset and returns all complete tiles.

from grdtiler import tiling_prod

dataset, tiles = tiling_prod(
    path="/path/to/S1A_IW_GRDH.SAFE",
    tile_size=17600,          # metres; or dict {"line": 17600, "sample": 17600}
    resolution="400m",
    detrend=True,
    noverlap=0,               # pixels; or dict {"line": 0, "sample": 0}
    centering=True,
    side="left",
    save=True,
    save_dir="./tiles",
    config_file="grdtiler/config.yaml",
)

Returns (dataset, tiles) — both are xr.Dataset.

---

2. Point-based tiling — tiling_by_point

Extracts one tile centred on each supplied geographic point.

from shapely.geometry import Point
from grdtiler import tiling_by_point

locations = [Point(-3.704, 40.417), Point(2.352, 48.857)]

dataset, tiles = tiling_by_point(
    path="/path/to/S1A_IW_GRDH.SAFE",
    posting_loc=locations,
    tile_size=20000,          # metres
    resolution="10m",
    detrend=True,
    save=True,
    save_dir="./point_tiles",
    config_file="grdtiler/config.yaml",
)

Points outside the product footprint are skipped with a warning.

---

3. Wave-mode tiling — tiling_wv

Extracts a single centred tile from a Sentinel-1 WV product. Antimeridian crossing is handled automatically.

from grdtiler import tiling_wv

dataset, tile = tiling_wv(
    path="/path/to/S1A_WV_SLC.SAFE",
    tile_size=20000,
    resolution="10m",
    detrend=True,
    config_file="grdtiler/config.yaml",
)

---

API reference

tiling_prod

Parameter	Type	Default	Description
path	str | xr.Dataset	—	Path to SAR product or pre-loaded dataset
tile_size	int | dict	—	Tile size in metres ({"line": …, "sample": …} for non-square)
resolution	str	None	Target resolution, e.g. "400m"
detrend	bool	True	Apply GMF-based sigma0 detrending
noverlap	int | dict	0	Overlap in pixels between adjacent tiles
centering	bool	False	Centre the tiling grid inside the dataset
side	str	"left"	Alignment when centering ("left" or "right")
add_footprint	bool	True	Compute and attach tile footprint + centroid
save	bool	False	Save tiles to NetCDF
save_dir	str	"."	Root directory for saved tiles
to_keep_var	list	None	Variables to retain (default set if None)
config_file	str	"config.yaml"	Path to GMF model configuration

tiling_by_point

Parameter	Type	Default	Description
path	str | xr.Dataset	—	Path to SAR product or pre-loaded dataset
posting_loc	list[Point]	—	Shapely Point objects (lon, lat)
tile_size	int	—	Tile size in metres
resolution	str	None	Target resolution
detrend	bool	True	Apply sigma0 detrending
scat_info	dict	None	Scatterometer wind data to annotate tiles
save	bool	False	Save tiles to NetCDF
save_dir	str	"."	Root directory for saved tiles
to_keep_var	list	None	Variables to retain
config_file	str	"config.yaml"	Path to GMF model configuration

tiling_wv

Parameter	Type	Default	Description
path	str	—	Path to Sentinel-1 WV SAFE product
tile_size	int	—	Tile size in metres
resolution	str	None	Target resolution
detrend	bool	True	Apply sigma0 detrending
to_keep_var	list	None	Variables to retain
config_file	str	"config.yaml"	Path to GMF model configuration

---

GMF configuration

The config.yaml file maps each mission to its GMF model names:

gmf_base_path: "/path/to/gmf/nc_luts"

gmf_models:
    S1:
        GMF_VV_NAME: "gmf_cmod5n"
        GMF_VH_NAME: "gmf_s1_v2"
        GMF_HH_NAME: "nc_lut_gmf_cmod7_Rhigh_hh_zhangA"
        GMF_HV_NAME: "nc_lut_gmf_cmod7_Rhigh_hh_zhangA"
    RS2:
        GMF_VV_NAME: "gmf_cmod5n"
        GMF_VH_NAME: "gmf_rs2_v2"
    RCM:
        GMF_VV_NAME: "gmf_cmod5n"
        GMF_VH_NAME: "gmf_rcm_noaa"

---

Running tests

pip install -e ".[dev]"
pytest tests/ -v

All 29 tests are self-contained — no SAR product download required.

---

License

MIT — see LICENSE.