Create a Sentinel-1 radar image

Tilebox indexes global Sentinel-1 Ground Range Detected (GRD) scenes and their public AWS assets in the open_data.aws_earth.sentinel1 dataset. In this guide, you query a scene over Venice, read its VV polarization measurement, and create a north-up monochrome radar image.

Prerequisites

You have a Tilebox API key.
You have Python 3.11 or newer.

uv add tilebox shapely numpy pillow rasterio

Select a Sentinel-1 scene

Define an area around Venice, then query a dual-polarization Sentinel-1C scene from October 4, 2025:

Python

from shapely import box
from tilebox.datasets import Client

# west, south, east, north
area = box(12.3, 45.385, 12.49, 45.466)  # Venice

collection = Client().dataset("open_data.aws_earth.sentinel1").collection("GRD")
scenes = collection.query(
    temporal_extent=("2025-10-04", "2025-10-05"),
    spatial_extent=area,
)

datapoint = scenes.isel(time=0)
print(datapoint.stac_id.item())

Resolve the VV measurement

Each datapoint provides measurement COGs for its available polarizations, alongside product, calibration, noise, manifest, and preview assets. Select the VV measurement and resolve its public HTTPS location:

Python

from tilebox.datasets.assets import AssetCollection
from tilebox.storage.aio import AssetAccessPolicy, Client as StorageClient

assets = AssetCollection.from_datapoint(datapoint)
vv = assets["vv"]

storage = StorageClient(
    policy=AssetAccessPolicy(preferred_schemes=("https",)),
)
resolved = storage.resolve(vv)

No AWS credentials or requester-pays configuration is required for this asset.

Read a north-up image window

Sentinel-1 GRD measurement COGs store geolocation as ground control points. Use a WarpedVRT to apply that geolocation, project the image, and read a bounded north-up window without downloading the complete scene:

Python

import numpy as np
import rasterio
from rasterio.enums import Resampling
from rasterio.vrt import WarpedVRT
from rasterio.warp import transform_bounds
from rasterio.windows import from_bounds

with rasterio.open(resolved.href) as source:
    with WarpedVRT(source, crs="EPSG:3857") as geotiff:
        projected_bounds = transform_bounds(
            "EPSG:4326",
            geotiff.crs,
            *area.bounds,
        )
        window = from_bounds(*projected_bounds, transform=geotiff.transform)
        vv_pixels = geotiff.read(
            1,
            window=window,
            out_shape=(675, 1200),
            resampling=Resampling.bilinear,
        ).astype(np.float32)

Render the radar image

Stretch the central 96% of valid pixel values across a gray display range, then save the result:

Python

from PIL import Image

valid = vv_pixels[vv_pixels > 0]
low, high = np.percentile(valid, (2, 98))
grayscale = np.clip((vv_pixels - low) / (high - low), 0, 1)

image = Image.fromarray((grayscale * 255).astype(np.uint8), mode="L")
image.save("sentinel1-venice.png")

Monochrome Sentinel-1 VV radar image of Venice and the surrounding lagoon

Smooth water appears dark because it reflects little radar energy back toward the sensor, while dense buildings appear bright because their geometry produces strong returns.

This percentile stretch creates a visual image from the stored measurement values. Calibrate the measurement and account for acquisition geometry before using pixel values in quantitative SAR analysis.

Tilebox Guides

Datasets

Workflows

Operations

Create a Sentinel-1 radar image

Prerequisites

Select a Sentinel-1 scene

Resolve the VV measurement

Read a north-up image window

Render the radar image

Next steps

Create a Sentinel-2 RGB image

Assets and storage

​Prerequisites

​Select a Sentinel-1 scene

​Resolve the VV measurement

​Read a north-up image window

​Render the radar image

​Next steps

Create a Sentinel-2 RGB image

Assets and storage

Prerequisites

Select a Sentinel-1 scene

Resolve the VV measurement

Read a north-up image window

Render the radar image

Next steps