seppo_nisar_gcov_convert¶
Dual-pol Ratio (-dpratio / --dualpol_ratio)¶
Computes a polarimetric ratio between the like-pol and cross-pol channels. Requires a DH or DV dual-pol acquisition; QP and single-pol modes are rejected with a warning.
| Polarization | Like-pol | Cross-pol | Ratio band name |
|---|---|---|---|
| DH (freq A) | HHHH | HVHV | HHHH/HVHV |
| DV (freq A) | VVVV | VHVH | VVVV/VHVH |
The individual like-pol and cross-pol bands are always written alongside the ratio.
Output files (single-bands, DH example)¶
| File | Content |
|---|---|
...-EBD_A_hh_<MODE>.tif |
Like-pol (HHHH) |
...-EBD_A_hv_<MODE>.tif |
Cross-pol (HVHV) |
...-EBD_A_hhhvra_<MODE>.tif |
Ratio band |
...-EBD_A_hhhvra_<MODE>.vrt |
3-band snapshot VRT |
With --no_single_bands the output is a single 3-band COG (band 1 = like-pol, band 2 = cross-pol, band 3 = ratio), useful for browse images (e.g. -DN -d 20 --no_single_bands -dpratio).
Ratio formula per output mode¶
| Mode | Formula | Output dtype | Nodata | Clamp |
|---|---|---|---|---|
-pwr (default) |
likepol / crosspol |
float32 | nan | -- |
-amp |
amp(likepol) / amp(crosspol) x 1000 |
uint16 | 0 | [1, 65535] |
-DN |
DN(likepol) - DN(crosspol) + 127 |
uint8 | 0 | [1, 255] |
-dB |
dB(likepol) - dB(crosspol) |
float32 | nan | -- |
For -amp and -DN the nodata mask is determined before clamping (pixels where the denominator or either input is 0 or non-finite).
Clamping is then applied to valid pixels and nodata pixels are set to 0 last, so clamping never accidentally promotes a valid 0 to 1.