Never seen something like this with SWAT results. The water balance is totally out of range. How comes surface runoff (16,487 mm) and groundwater percolation (4,743 mm) are many times larger than precipitation (just 532 mm). Seems to be a strange bug, which I have not encountered before. Is there a large surface water body in the catchment (e.g. big lakes, reservoirs) or huge large snow/ice storage? Only those could create something like this in reality. You may like to check the output files (.sub and/or .rch) to see where are these large surface flows - are they on a particular days, do they follow a certain pattern, etc? Same with the percolation values. Such checks may give some clues.