Data Format and Structure¶
Data Format¶
Images are stored in the Cloud Optimized GeoTIFF (COG) format specification which enables users to effeciently read these data directly off S3 while still maintaining all the benefits of distributing imagery as a GeoTIFF. These COG images contain internal overviews that GDAL can use when reading a low resolution version of the image. Each image stores a single polarization as a single band. Data are stored in linear units of power instead of logarithmic units of decibels, and saved as Float32. The NODATA value is 0, as in the original GRD product.
The data are gridded into tiles based on the Sentinel-2 20m gridding scheme. Each tile is 5490x5490 in size in a UTM zone based on the grid definition. This Sentinel-2 grid is based on the Military grid system (latest KML file). For this dataset extent (Contiguous United States), there are 993 grid tiles.
You may download the vector data we use to define the dataset extent and grid here:
Grid (based off Sentinel-2 MGRS grid)
Both vector files are stored in the GeoJSON format. The grid GeoJSON file has attributes that you may use to identify the grid tiles and know ahead of time the geospatial attributes of the tiled raster data (e.g., the Affine transform and projection).
Each backscatter image has the following attributes,
Attribute |
Value |
Description |
|---|---|---|
Projection |
UTM |
UTM zone varies by scene |
Resolution |
20m |
20x20m spatial resolution |
Bands |
1 |
Each polarization is stored as a separate 1-band image |
Shape |
5490x5490 |
The shape of a Sentinel-2 MGRS grid tile |
No Data Value |
0 |
Image “no data value” |
Data Type |
Float32 |
Float32 used to store backscatter as linear power ratio |
Compression |
Deflate |
Deflate compression with predictor=3 |
Block Sizes |
512x512 |
512x512 pixels per block |
We also provide the local incident angle used by the Sentinel-1 Toolbox when performing terrain flattening and correction. These angle images are formatted slightly differently, and differ in the following attributes:
Attribute |
Value |
Description |
|---|---|---|
Data Type |
UInt16 |
Unsigned 16-bit integer |
Scaling |
100 |
Angle (degrees) is scaled and converted to UInt16 to save storage space. For example, 4500 is 45 degrees. |
Data Structure¶
We organize the dataset on S3 in a similar manner to the layout of the Sentinel-2 L1C and L2A datasets on AWS. Data are initially split according to the grid tile and then are organized according to the year of the acquisition. All images for a given tile and date are organized within their own subdirectory, making it relatively easy to locate radar backscatter imagery and metadata if you know the grid tile and date range you are searching for.
The layout of the bucket is as follows,
s3://sentinel-s1-rtc-indigo/tiles/RTC/1/[MODE]/[MGRS UTM zone]/[MGRS latitude label]/[MGRS Grid Square ID]/[YEAR]/[SATELLITE]_[DATE]_[TILE ID]_[ORBIT DIRECTION]/[ASSET]
Where,
Path Component |
Values |
Description |
|---|---|---|
MODE |
|
Only Interferometric Wide (IW) are processed |
MGRS UTM Zone |
|
UTM Zone component of tile ID |
MGRS Latitude |
|
Latitude band (CONUS spans from R to U) |
MGRS Grid Square ID |
|
The 100x100km Grid Square ID |
YEAR |
|
Year of the acquisition |
SATELLITE |
|
Satellite mission ID |
DATE |
|
Date of the acquisition starting time,
formatted as |
TILE ID |
|
Sentinel-2 20m MGRS tile grid ID |
ORBIT DIRECTION |
|
Ascending or descending orbit direction |
ASSET |
|
Tiled RTC Sentinel-1 Cloud Optimized GeoTIFF |
For example,
s3://sentinel-s1-rtc-indigo/tiles/RTC/1/IW/14/T/PN/2020/S1A_20200801_14TPN_ASC/
s3://sentinel-s1-rtc-indigo/tiles/RTC/1/IW/14/T/PN/2020/S1A_20200801_14TPN_ASC/Gamma0_VV.tif
s3://sentinel-s1-rtc-indigo/tiles/RTC/1/IW/14/T/PN/2020/S1A_20200801_14TPN_ASC/Gamma0_VH.tif
s3://sentinel-s1-rtc-indigo/tiles/RTC/1/IW/14/T/PN/2020/S1A_20200801_14TPN_ASC/local_incident_angle.tif
These data are acquired during an ascending orbit in dual-vertical polarization for August 1st, 2020 for the tile 14TPN.