Changes between Initial Version and Version 2 of Ticket #203
- Timestamp:
- Aug 24, 2018, 6:00:20 PM (6 years ago)
Legend:
- Unmodified
- Added
- Removed
- Modified
-
Ticket #203
- Property Milestone changed from to FLEXPART 10
- Property Component changed from FP physics/numerics to FP coding/compilation
- Property Summary changed from inconsitency/bug in verttransform_gfs.f90 to inconsistency / bug in verttransform_gfs.f90
-
Ticket #203 – Description
initial v2 3 3 in FLEXPART 9.2 I stumbled upon what is most probably a bug, and it is still present in the FLEXPART git dev branch. 4 4 5 Comparing verttransform.f90 (now seems to be verttransform_ecmwf.f90) to verttransform_gfs.f90, around line 500, one notices pieces of code which probably should be identical, but they aren't:5 Comparing `verttransform.f90` (now seems to be `verttransform_ecmwf.f90`) to `verttransform_gfs.f90`, around line 500, one notices pieces of code which probably should be identical, but they aren't: 6 6 7 7 for ECMWF it looks like: 8 8 9 if (vv(nx/2-1,0,iz,n).lt.0.) then 9 10 {{{#!fortran 11 if (vv(nx/2-1,0,iz,n).lt.0.) then 10 12 ddpol=atan(uu(nx/2-1,0,iz,n)/ & 11 13 vv(nx/2-1,0,iz,n))+xlonr … … 13 15 ddpol=pi+atan(uu(nx/2-1,0,iz,n)/ & 14 16 vv(nx/2-1,0,iz,n))+xlonr 17 }}} 15 18 16 19 (+ + xlonr) … … 18 21 whereas for GFS: 19 22 20 if(vv(nx/2-1,0,iz,n).lt.0.) then 23 24 {{{#!fortran 25 if(vv(nx/2-1,0,iz,n).lt.0.) then 21 26 ddpol=atan(uu(nx/2-1,0,iz,n)/vv(nx/2-1,0,iz,n))+xlonr 22 27 elseif (vv(nx/2-1,0,iz,n).gt.0.) then 23 28 ddpol=pi+atan(uu(nx/2-1,0,iz,n)/vv(nx/2-1,0,iz,n))-xlonr 29 }}} 24 30 25 31 (+ - xlonr)