More data - Gridded Nexrad data, Loss method SCS Curve Number, Transform method Modclark to work with the gridded data, no baseflow in this watershed, Routing method Muskingum-Cunge, with no Loss/Gain method
HEC-HMS is not a heavy duty flood routing software like HEC-RAS. It only can solve kinematic wave and lump methods (like Muskingum-Cunge) and it can only solve for prismatic, uniform cross-sections. You may use it for very simple routing but this is not a channel design tool and should not be used as a channel design/analysis tool. HEC-HMS is hydrology toll for converting rainfall to runoff.
Back to your question:
It cannot keep track of water if it exceeds the channel's freebaord, and if that happens you will loss mass in your system. Just virtually increase the depth of your channel.
One limitations in many cross sectional surveys, is that they just survey the defined channel, and do not extend to the full extent of floodplain to confining slope or abandoned terrace. As suggested by Dr. Zamani, HECRAS should work. Some years back in one if the Rosgen training courses, I used Rivermorph software to help assess if the channel section we were working on had sufficient velocity to pass its fine sediment and bedload in flood stage, resulting in assessment of likelihood of aggravation, due to insufficient velocity to move entrained load. And not all channels are stable in floodstage.