It seems that the NRFA has changed the output of the API output for amax and pot records: need to see if this is permanent and will also trickle down to winfap files. If so one can automatically derive the water year from the datetime information (the water year starts a 9am), and there is no longer a need to keep track of the events for which the water_year function "failed" since they happened before 9am. Note this also has consequences for the rnrfa packkage.