I am having a bit of a problem post calibration with a S1 SLC scene processing chain.
The first graph takes one SLC dual pol scene, calibrates it and writes the complex output.
Due to memory limitations I am dividing VV and VH processing separately.
The second graph takes the previous input, applies orbit file, topsar splits (three times), debursts (for each swath), merges (all 3 swaths), multilooks and writes.
Graph one works fine for both VV and VH output. Data load in SNAP GUI, no problems.
Graph two works fine for VV, but I receive the following error for the VH channel:
Error in graph: [Node:TOPSAR-Split] node
There are no discernible differences between the two outputs from the first graph, and the SNAP tool (i.e. without using the graph builder) version of TOPSAR-split reads the VH calibrated output easily. It is just the gpt that fails to read it, even within the GUI if the scene is loaded (see below).
No problems for VV:
Problem for VH:
SNAP tool works fine for VH:
I have looked at the metadata, and the only difference I can see is a slightly different order in the bands (see below) - the VH lists them in the order IW2;IW3;IW1, whereas VV shows IW2;IW1;IW3. I have no idea whether that makes the slightest difference.