A tools that has been used to identify unanotated splicing events for plants and that, in my opinion, gives robust results is SplAdder (See: http://biorxiv.org/content/early/2016/02/05/017095).
In my experience, it is also important that one chooses a good read aligner, e.g. STAR and filters out reads that map to multiple locations, as these can otherwise lead to false positives.
A good description of a pipeline to detect and identify differential splicing events in plants, can be found in: http://www.plantcell.org/content/25/10/3726.short