5. L1 Post-Processing#

5.1 L1a/b Data Formatting#

Algorithm overview and rationale

The scope of this step is to “transcode” the L1a and L1b products generated by the L1 Core Processing step from the L1 Processor internal format to the official BPS L1 Products format.

This format is defined and described in details in [RD26]. The main components of a L1 Product are:

the SAR image file(s), including the SAR image binary raster data for all the four BIOMASS polarisations, organised in two Cloud Optimized GeoTIFF (COG) files, one for amplitude and one for phase (only for L1a products) components. These data are those provided in output by the Autofocus (see sec. 4.11) or by the Ground Projection (see sec. 4.14) step respectively in case of L1a or L1b product;
the main annotation file, including the properties and characteristics of the measurement data or how the measurement data was generated, stored in a single XML file;
the Look-Up Table (LUT) file, including those metadata being too voluminous to be coded efficiently in an XML file (i.e., Look-Up Tables) and requiring a binary NetCDF file. These additional information are provided in output by various L1 Processor steps, in particular:
- RFI Mask, generated by RFI Mitigation step (see sec. 4.1);
- Sigma-Nought and Gamma-Nought Maps, generated by Radiometric Corrections and Calibrations step (see sec. 4.8);
- Faraday Rotation, Phase Screen, TEC, Range and Azimuth Shifts Maps, generated by Ionospheric Calibration step (see sec. 4.10);
- Autofocus Phase Screen Map, generated by Autofocus step (see sec. 4.11);
- Denoising Map, generated by Denoising step (see sec. 4.13);
- Latitude, Longitude, Height, Incidence and Elevation Angles Maps, generated by Back-geocoding step (see sec. 3.5);

Note

LUTs annotated in output product are a decimated version of output data of the various L1 Processor steps. Decimation factors (range and azimuth) are configurable. Different decimation factors for the different LUT groups can be used;

the orbit and attitude files, including orbit and attitude information in dedicated XML files. In principle these are the same auxiliary orbit and attitude files taken input by the L1 Processor, unless it performs any interpolation, extrapolation, … operation. In this case the modified orbit and attitude information are annotated in these two files;
the quick-look and overlay files, including an averaged and decimated preview of the SAR image data set and information about its geographic coverage (see next section).

Inputs

Symbol	Name	Origin	Size	Variability	Default	Description
-	SLC autofocused or GRD	Autofocus step or Ground projection step	(Nrg-Nchirp+1) x N’az array of complex values for L1a or N’’’rg x N’’az array of real values for L1b	-	-	Azimuth compressed data after autofocus or Multi-look detected, denoised and ground-projected data Intermediate product: iSLC_af_corrected / iGRD
-	LUT	Various L1 core processing step (see previous sections)	(Nrg-Nchirp+1) x N’az arrays of real values	-	-	Various side products (RFI masks, ionosphere phase screen, …) added to output product as annotation LUT
-	`pixelRepresentation`	`AUX_PP1` (`AUX_FMT`)	string	-	Abs Phase or Abs	Representation of the image pixels within the image MDS (I Q, Abs Phase, Abs, Pow Phase, Pow)
-	`pixelType`	`AUX_PP1` (`AUX_FMT`)	String	-	32 bit Float	Data type of output pixels within the image MDS
-	`pixelQuantity`	`AUX_PP1` (`AUX_FMT`)	string	-	Beta-Nought	Physical quantity stored in output data (Beta-Nought, Sigma-Nought or Gamma-Nought)
-	`absCompressionMethod`	`AUX_PP1` (`AUX_FMT`)	string	-	LERC	Abs values TIFF compression method
-	`absMaxZError`	`AUX_PP1` (`AUX_FMT`)	scalar value	-	0.0001	Maximum error threshold on abs values. Used only in case `absCompressionMethod` is set to LERC, `LERC_DEFLATE` or `LERC_ZSTD`. If set to 0, compression is lossless. If set to -1, value is computed dynamically
-	`phaseCompressionMethod`	`AUX_PP1` (`AUX_FMT`)	string	-	LERC	Phase values TIFF compression method
-	`phaseMaxZError`	`AUX_PP1` (`AUX_FMT`)	scalar value	-	0.001	Maximum error threshold on phase values. Used only in case `phaseCompressionMethod` is set to LERC, `LERC_DEFLATE` or `LERC_ZSTD`. If set to 0, compression is lossless. If set to -1, value is computed dynamically
-	`noPixelValue`	`AUX_PP1` (`AUX_FMT`)	scalar value	-	-9999	Pixel value in case of invalid data
-	`blockSize`	`AUX_PP1` (`AUX_FMT`)	scalar value	-	512	Size of block written during TIFF files export
-	`lutRangeDecimationFactorList`, `lutAzimuthDecimationFactorList`	`AUX_PP1` (`AUX_FMT`)	2 sets of scalar values	-	2, 12 for geometry LUTs 20, 120 for other LUTs	Look-Up Tables (LUT) ADS range and azimuth decimation factors w.r.t. output L1 product sampling grid, divided per LUT groups
-	`lutBlockSize`	`AUX_PP1` (`AUX_FMT`)	scalar value	-	512	Size of block written during NetCDF files export
-	`qlRangeDecimationFactor`, `qlAzimuthDecimationFactor`	`AUX_PP1` (`AUX_FMT`)	2 scalar values	-	20, 120	Quick-look ADS range and azimuth decimation factors w.r.t. output L1 product sampling grid
-	`qlRangeAveragingFactor`, `qlAzimuthAveragingFactor`	`AUX_PP1` (`AUX_FMT`)	2 scalar values	-	20, 120	Quick-look ADS range and azimuth averaging factors
-	`qlAbsoluteScalingFactorList`	`AUX_PP1` (`AUX_FMT`)	3 scalar values	-	-	Absolute scaling factors (one value per RGB channel) to be applied to quick-look ADS

Outputs

Symbol	Name	Destination	Size	Variability	Default	Description
-	L1S	Stack Processor (`L1_PFD`)	(Nrg-Nchirp+1) x N’az array of complex values for L1a or N’’’rg x N’’az array of real values for L1b	-	-	L1 (a or b) Standard Product
-	L1M	Stack Processor (`L1_PFD`)	- (only annotations)	-	-	L1 (a or b) Monitoring Product

5.1.1 Quick-look Generation#

The quick-look is a lower resolution version of the SAR image included in each L1 product. The following rules and properties apply:

for L1a products, quick-looks are a power detected, averaged and decimated version of the full SAR image;
for L1b products, quick-looks are a squared, averaged and decimated version of the full SAR image.

BIOMASS products, being full-polarisation ones, contain only one quick-look image which is a combination of the imagery from all the four polarisation channels. This is done exploiting the three Red-Green-Blue (RGB) channels of the PNG image and the Pauli decomposition colour coding scheme for L1a products, i.e.:

Red channel: \(\left\vert HH-VV \right\vert\) (double bounce);
Green channel: \(\left\vert HV+VH \right\vert\) (volume scattering);
Blue channel: \(\left\vert HH+VV \right\vert\) (surface scattering).

and the Lexicographic one for L1b products, i.e.:

Red channel: \(\left\vert HH \right\vert\);
Green channel: \(\frac{1}{2}\sqrt{{HV}^2+{VH}^2}\);
Blue channel: \(\left\vert VV \right\vert\).

The quick-look images are produced by basic sample averaging and decimation, in both azimuth and range directions. The number of averaging samples in range and azimuth as well as the range and azimuth decimation factors are configurable parameters.

After this operation, each RGB channel values are normalized by a configurable absolute scaling factor, in order to have comparable results even on different scenes.