Negative optical depth exists

[WinKevin66]

Hi,
I hope you are OK.
When I try to get gas absorption optical depth, I found that there have some negative values in some wavenumber and layers.
The TAPE5 file is attached, using O2-A band(12738.000-13422.000), select no any continuum (ICNTNM = 0), the LNFL just use O2 with LNOUT EXBRD, some result is just like figure 1: first is wavenumber and then are each layers' value.
And if I change ICNTNM = 6, and select only 'XO2CN O2 continuum absorption multiplicative factor', there also have negative values.
The zip includes TAPE5 file and my result.
So, what should I do?
Thank you for your reply!
WinKevin

zip.tar.gz