Offset between WMS & vector data

[johnwbryant]

I'm looking at the Fire Scars by Month 2020 layer, and noticing an offset between the WMS and the vector data for the same layer, looking at an area in the Kimberley here:

The offset jumps around in QGIS on small panning movements (notice the hotspots WMS layer shifts correctly, while the fire scars layer doesn't).

There is a CRS difference between the WMS (ESPG:4326) and vector (EPSG:4283) data, not sure if that has any relevance.