Improve MQUASILAG option

[pesei]

Quasi-Lagrangian output, i.e. particle IDs and positions, is not well implemented yet.

In readcommand.f90:

  ! For quasilag output for each release is forbidden
  ! For quasilag backward is forbidden

In partoutput_short.f90:

  ! Convert positions to integer*2 variables (from -32768 to 32767)
  ! Do this only for region of main interest, i.e. extended North Atlantic region,
  ! and for the tracer of interest, i.e. the North American one
  !*****************************************************************************

      if (xlon.gt.180.) xlon=xlon-360.
      if (xlon.lt.-180.) xlon=xlon+360.

      numshortall=numshortall+1
      if ((xlon.gt.-140).and.(xlon.lt.60).and.(ylat.gt.10).and. &
           (xmass1(i,1).gt.0.)) then
        numshortout=numshortout+1
        idump(1,numshortout)=nint(xlon*180.)
        idump(2,numshortout)=nint(ylat*360.)
        zlim=min(ztra1(i)+topo,32766.)
        idump(3,numshortout)=nint(zlim)
        i4dump(numshortout)=npoint(i)
      endif

    endif
  end do

I don't see the reasons for these limitations, and in any case, they are not desirable (cf. recent FLEXPART discussion list posting, Subject: Particle tracking).

[pesei]

      if ((mquasilag.eq.1).and.(itime.eq.(loutstart+loutend)/2)) &
         call partoutput_short(itime)    
         ! dump particle positions in extremely compressed format

Why is quasilag output in the middleof the output interval?

[cluekenwinkels]

Hello,
I also have problems with the output of the partposit files in backward runs on a limited domain.
As far as I can tell there is no possibility to attach an id to each of the particles in this case (which would usually be done with mquasilag in forward mode). However I need the positions of each particle to recover its trajectory which, at the current state, is not possible out of the box.
I am now working on a workaround in postprocessing, but I am sure many could profit from the addition of a particle id to the position-output.

Context, where to use the trajectories:
I am interested in the source-receptor relation in a limited domain and want to seperate the influence from within the domain from the outside. For a background, measurements at the endpoints of the trajectories are used. Without consistent trajectories this is not possible.

Thanks in advance!
Greetings,
Christopher

[pesei]

I have talked to Andreas Stohl who originally implemented MQUASILAG. The limitiations are there because originally it was for a specific application only.

I think we should remove the limitations, and I should do it, or you, or in some form of cooperation (for contacting me, see DevelopersListing)