Hello @andalo!

This might be happening because lineage for MERGE operations is not fully captured by default. Could you try setting the Spark property spark.databricks.dataLineage.mergeIntoV2Enabled to true?
Just a heads-up, enabling this setting may impact query performance slightly.