I’m guessing from the name of your directory path (“EMAQ1p3km”) that your grid resolution is 1.3333 km and that there may be very slight differences in the XCELL and/or YCELL attributes between the CCTM_ACONC file opened as INFILE1 and the METCRO3D files opened as INFILE2. You could verify this by doing an ncdump -h on both files and comparing the grid attributes.
There is a known shortcoming of the portion of combine that ensures consistency of grids across files before combining variables from the different files. This shortcoming affects domains with “odd” grid spacing and/or grid origin values. There are multiple examples of this issue in this thread, along with suggested code changes to solve this issue by relaxing the tolerance criteria or making these criteria relative rather than absolute (the more robust solution).