The current dev-branch also has this difference in it. I will try to investigate tomorrow.
(There are even -250 K in one place of your results. So clearly something weird is happening here.) With hope, //Richard Den tors 6 maj 2021 kl 15:00 skrev Fox, Stuart <stuart....@metoffice.gov.uk >: > Dear ARTS developers, > > > > I’m trying to understand some changes to clear-sky simulations when moving > to ARTS v2.4 and have noticed that there seems to be a difference in how a > cutoff frequency for absorption lines is handled. I’ve put a basic example > illustrating this at https://github.com/stuartfox/arts-cutoff-problem, > which has the controlfiles I’ve used at ARTS v2.4 and ARTS v2.3.1277 and > the input data I used. > > > > The example performs a clear-sky simulation using only water vapour > absorption lines read from an ARTSCAT-3 format file (which was originally > derived from the HITRAN-style file provided by AER). Note that Ive used a > slightly silly cutoff frequency of 20GHz around each line to highlight the > problem; it shows small (but non-negligible) differences in brightness > temperature around the main water vapour lines, and at v2.4 the output also > has a couple of very large negative brightness temperatures which leads me > to think there may be a bug in that version. If I remove the line cutoff > (by changing line 27 of the v2.4 controlfile to cutoff_option="None" and > line 25 of the v2.3 controlfile to cutoff=-1) then I get very good > agreement between the two versions. > > > > Is this a problem with ARTS or am I doing something wrong with my setup? > > > > Thanks, > > > > Stuart >
