The manual says that theta in an input sounding is moist (see page 44 "where qv and θm are given in the input sounding"). This seems unlikely because 1) dry temperature is usually measured and 2) the get_sounding subroutine appears to be using the value as a dry theta, where later the moist theta is calculated with qvf = 1. + rvovrd*qv_input(k) and th_input(k)*qvf