JD-EE, seem to me your numbers about transmission losses are about right. However, since the source frequency is 37.5 ± 1 kHz, I think we can safely assume that the bandwidth of the TPL reciever is at least 2 kHz. Counting only sea state noise (7 to 10 kt, no rain) we have about 34 dB re 1 uPa / 1Hz wich gives a 67 dB noise level. Using your numbers for transmission losses this leaves 160 - 75 - 67 = 18 dB available for detection @ 1700 m, a value considered minimal for simple detection (learned from auv-ee).
By the way, I am wondering if the towed array the french navy submarine dragged was really a usefull asset. Theese things may not be efficient at all at 37.5 kHz even using DSP.