Skip to content

Correct rdr_geometry argument for gpu-enabled ionosphere screen geocoding - #245

Open
CBurton90 wants to merge 1 commit into
isce-framework:developfrom
CBurton90:correct_geocode_insar.py
Open

CBurton90 wants to merge 1 commit into
isce-framework:developfrom
CBurton90:correct_geocode_insar.py

Conversation

@CBurton90

Copy link
Copy Markdown

When using the main_diff_ms_band method and geocoding the resultant ionospheric correction there is a dimension/shape mismatch as the radar geometry argument is not multilooked (rdr_geometry). Using the multilooked geometry (rdr_geometry_iono) corrects this.

ERROR 5: HDF5:/opt/outputs/ionosphere/main_diff_ms_band/RUNW.h5://science/LSAR/RUNW/swaths/frequencyB/interferogram/HH/ionospherePhaseScreen, band 1: Access window out of range in RasterIO().  Requested
(0,0) of size 3781x2280 on raster of 945x2280.

lijun99 added a commit to lijun99/isce3 that referenced this pull request Aug 23, 2026
gpu_geocode_rasters() for main_diff_ms_band/main_side_band was called
with rdr_geometry (stale, non-multilooked) instead of the correctly
computed rdr_geometry_iono, producing a shape mismatch that geocoded
the ionosphere phase screen as all zeros. Same fix as upstream PR
isce-framework#245 by CBurton90 (open, not yet merged) --
independently found via the all-zero geocoded layer, then confirmed
against that PR's description.
lijun99 added a commit to lijun99/isce3 that referenced this pull request Sep 16, 2026
gpu_geocode_rasters() for main_diff_ms_band/main_side_band was called
with rdr_geometry (stale, non-multilooked) instead of the correctly
computed rdr_geometry_iono, producing a shape mismatch that geocoded
the ionosphere phase screen as all zeros. Same fix as upstream PR
isce-framework#245 by CBurton90 (open, not yet merged) --
independently found via the all-zero geocoded layer, then confirmed
against that PR's description.
lijun99 added a commit to lijun99/isce3 that referenced this pull request Sep 18, 2026
gpu_geocode_rasters() for main_diff_ms_band/main_side_band was called
with rdr_geometry (stale, non-multilooked) instead of the correctly
computed rdr_geometry_iono, producing a shape mismatch that geocoded
the ionosphere phase screen as all zeros. Same fix as upstream PR
isce-framework#245 by CBurton90 (open, not yet merged) --
independently found via the all-zero geocoded layer, then confirmed
against that PR's description.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant