Re: divergence problem

Thanks, Jimmy.

I was running regcm at 30km. 

I was concerned that the wind components didn't look smooth enough, so
I tried using smth9 on them and then computing the divergence.  That
helped produced a smoother plot of divergence but I feel uncomfortable
about this approach.  And I did the same for my control and experiment
and when I differenced their divergences, the difference was also noisy
and looked like a numerical issue.

On 03/26/11, James Correia wrote:
> Hi Michael-
> Perhaps the technique is fine, but the resulting plot is not showing the divergence at the resolved scales? I would recommend using a Gaussian smoother on the resulting divergence. I don't know what RegCM's resolution is, but MM5 was around 10 delta x and WRF is around 7 delta x, where x is the grid spacing. So you could apply the gaussian filter over a radius of between 7 and 10 grid points to capture the resolved scales.
>
>
> jimmyc
>
>
> On Sat, Mar 26, 2011 at 1:43 PM, Michael Notaro <mnotaro@wisc.edu <mnotaro@wisc.edu>> wrote:
>
> >
>
> > Does anyone have any suggestions on either a function
> >
>
> > for divergence to apply directly to lat2d x lon2d data
> >
>
> > or how to refine my approach below (1st fit to a regular
> >
>
> > grid and then compute divergence)?
> >
>
> >
> >
>
> >
> >
>
> > Dear Michael
> >
>
> >
> >
>
> > I would compute the divergence on the 2d model grid (using the model&#39;s own
> >
>
> > discretization) from the vertically interpolated winds, then average in time
> >
>
> > and at the very end interpolate to the new 2d grid.
> >
>
> >
> >
>
> > Regards,
> >
>
> > Andrea
> >
>
> > --
> >
>
> >
> >
>
> > Andrea N. Hahmann
> >
>
> > Senior Scientist
> >
>
> > Wind Energy Division
> >
>
> > Risø DTU
> >
>
> >
> >
>
> > Technical University of Denmark
> >
>
> > Risø National Laboratory for Sustainable Energy
> >
>
> > Frederikborgvej 399, P.O. Box 49
> >
>
> > 4000 Roskilde, Denmark
> >
>
> >
> >
>
> > Direct +45 4677 5471(tel:%2B45%204677%205471)
> >
>
> > Mobil: +45 2133 0550(tel:%2B45%202133%200550)
> >
>
> > ahah@risoe.dtu.dk <ahah@risoe.dtu.dk>
> >
>
> > www.risoe.dtu.dk(http://www.risoe.dtu.dk)
> >
>
> >
> >
>
> >
> >
>
> >
> >
>
> > > From: Michael Notaro <mnotaro@wisc.edu <mnotaro@wisc.edu>>
> >
>
> > > Date: Fri, 25 Mar 2011 09:40:36 -0500
> >
>
> > > To: ncl-talk NCL <ncl-talk@ucar.edu <ncl-talk@ucar.edu>>
> >
>
> > > Subject: divergence calculations
> >
>
> > >
> >
>
> > > My calculated monthly mean 250-hPa divergence map
> >
>
> > > looks extremely noisy (see attached plots of divergence
> >
>
> > > and interpolated u wind).
> >
>
> > >
> >
>
> > > I started with 6-hourly U and V on a regional model lat2d x lon2d
> >
>
> > > grid.  I used int2p to interpolate U and V from sigma to 250 hPa.
> >
>
> > > Then I used rcm2rgrid to interpolate to a fixed grid.  Then I used
> >
>
> > > uv2dv_cfd to compute 250 hPa divergence of these interpolated
> >
>
> > > wind components.  Finally I averaged the divergences across
> >
>
> > > all time steps to get the monthly mean.
> >
>
> > >
> >
>
> > > Any suggestions?  I assume the interpolated wind components
> >
>
> > > are not the best, and so the derivates for divergence look messy.
> >
>
> > >
> >
>
> > > Michael
> >
>
> > >
> >
>
> > >
> >
>
> > >
> >
>
> > >
> >
>
> > > load "$NCARG_ROOT/lib/ncarg/nclscripts/csm/gsn_code.ncl"
> >
>
> > > load "$NCARG_ROOT/lib/ncarg/nclscripts/csm/gsn_csm.ncl"
> >
>
> > > load "$NCARG_ROOT/lib/ncarg/nclscripts/csm/contributed.ncl"
> >
>
> > > load "$NCARG_ROOT/lib/ncarg/nclscripts/csm/shea_util.ncl"
> >
>
> > > begin
> >
>
> > >
> >
>
> > > nlat=148
> >
>
> > > nlon=158
> >
>
> > >
> >
>
> > > ctl=new((/8,148,158,10/),float)
> >
>
> > >
> >
>
> > > months=(/"MAR","APR","MAY","JUN","JUL","AUG","SEP","OCT"/)
> >
>
> > >
> >
>
> > > yrs=(/"1991","1992","1993","1994","1995","1996","1997","1998","1999","2000"/)
> >
>
> > > yr=(/"91","92","93","94","95","96","97","98","99","00"/)
> >
>
> > > mns=(/"03","04","05","06","07","08","09","10"/)
> >
>
> > > hr=(/"06","00","00","00","00","00","00","00"/)
> >
>
> > > days=(/31,30,31,30,31,31,30,31/)
> >
>
> > >
> >
>
> > > do iy=0,9
> >
>
> > >   do im=0,7
> >
>
> > >     print(iy+" "+im)
> >
>
> > >     if ( (yrs(iy)+""+mns(im).eq."199103").or.(yrs(iy)+""+mns(im).eq."199406")
> >
>
> > > ) then
> >
>
> > >       a=addfile("../con9096/Southwest_ATM."+yrs(iy)+""+mns(im)+"0106.nc(http://0106.nc)","r")
> >
>
> > >     else
> >
>
> > >       a=addfile("../con9096/Southwest_ATM."+yrs(iy)+""+mns(im)+"0100.nc(http://0100.nc)","r")
> >
>
> > >     end if
> >
>
> > >     if ( (yrs(iy)+""+mns(im).eq."199103").or.(yrs(iy)+""+mns(im).eq."199406")
> >
>
> > > ) then
> >
>
> > >       u=a->u(3::,:,:,:)
> >
>
> > >       v=a->v(3::,:,:,:)
> >
>
> > >       ps=a->ps(3::,:,:)
> >
>
> > >       time=a->time(3::)
> >
>
> > >     else
> >
>
> > >       if (mns(im).eq."03") then
> >
>
> > >         u=a->u(4::,:,:,:)
> >
>
> > >         v=a->v(4::,:,:,:)
> >
>
> > >         ps=a->ps(4::,:,:)
> >
>
> > >         time=a->time(4::)
> >
>
> > >       else
> >
>
> > >         u=a->u(:,:,:,:)
> >
>
> > >         v=a->v(:,:,:,:)
> >
>
> > >         ps=a->ps(:,:,:)
> >
>
> > >         time=a->time
> >
>
> > >       end if
> >
>
> > >     end if
> >
>
> > >     lat2d=a->xlat
> >
>
> > >     lon2d=a->xlon
> >
>
> > >     sigma=a->sigma
> >
>
> > >     ptop=a->ptop ; hPa
> >
>
> > >     nt=dimsizes(time)
> >
>
> > >
> >
>
> > >     lat=fspan(min(lat2d),max(lat2d),148)
> >
>
> > >     lon=fspan(min(lon2d),max(lon2d),158)
> >
>
> > >     lat!0="lat"
> >
>
> > >     lat&lat=lat
> >
>
> > >     lat@units="degrees_north"
> >
>
> > >     lon!0="lon"
> >
>
> > >     lon&lon=lon
> >
>
> > >     lon@units="degrees_east"
> >
>
> > >
> >
>
> > >     p=new((/nt,18,148,158/),float)
> >
>
> > >     do ilev=0,17
> >
>
> > >       p(:,ilev,:,:)=ps*sigma(ilev)
> >
>
> > >     end do
> >
>
> > >     copy_VarCoords(u,p)
> >
>
> > >
> >
>
> > >     ulev=int2p(p(time|:,iy|:,jx|:,kz|:),u(time|:,iy|:,jx|:,kz|:),250.,1)
> >
>
> > >     vlev=int2p(p(time|:,iy|:,jx|:,kz|:),v(time|:,iy|:,jx|:,kz|:),250.,1)
> >
>
> > >
> >
>
> > >     ulevel=ulev(:,:,:,0)
> >
>
> > >     vlevel=vlev(:,:,:,0)
> >
>
> > >     copy_VarCoords(ps,ulevel)
> >
>
> > >     copy_VarCoords(ps,vlevel)
> >
>
> > >
> >
>
> > >     ugrid=rcm2rgrid(lat2d,lon2d,ulevel,lat,lon,0)
> >
>
> > >     vgrid=rcm2rgrid(lat2d,lon2d,vlevel,lat,lon,0)
> >
>
> > >     ugrid!0="time"
> >
>
> > >     ugrid!1="lat"
> >
>
> > >     ugrid!2="lon"
> >
>
> > >     ugrid&time=time
> >
>
> > >     ugrid&lat=lat
> >
>
> > >     ugrid&lon=lon
> >
>
> > >     copy_VarCoords(ugrid,vgrid)
> >
>
> > >
> >
>
> > >     div=uv2dv_cfd (ugrid,vgrid,lat,lon,2)
> >
>
> > >     copy_VarCoords(ps,div)
> >
>
> > >
> >
>
> > >     ctl(im,:,:,iy)=dim_avg_Wrap(div(iy|:,jx|:,time|:))
> >
>
> > >
> >
>
> > >     out=addfile("out.nc(http://out.nc)","c")
> >
>
> > >     out->ctl=ctl(im,:,:,iy)
> >
>
> > >     out->ugrid=ugrid
> >
>
> > >     out->vgrid=vgrid
> >
>
> > >
> >
>
> > >
> >
>
> > >     delete(a)
> >
>
> > >     delete(time)
> >
>
> > >     delete(u)
> >
>
> > >     delete(v)
> >
>
> > >     delete(ps)
> >
>
> > >     delete(p)
> >
>
> > >     delete(ulev)
> >
>
> > >     delete(vlev)
> >
>
> > >     delete(ugrid)
> >
>
> > >     delete(vgrid)
> >
>
> > >     delete(div)
> >
>
> > >
> >
>
> > >   end do
> >
>
> > > end do
> >
>
> > >
> >
>
> > > end
> >
>
> > >
> >
>
> > >
> >
>
> > > _______________________________________________
> >
>
> > > ncl-talk mailing list
> >
>
> > > List instructions, subscriber options, unsubscribe:
> >
>
> > > http://mailman.ucar.edu/mailman/listinfo/ncl-talk
> >
>
> >
> >
>
> > _______________________________________________
> >
>
> > ncl-talk mailing list
> >
>
> > List instructions, subscriber options, unsubscribe:
> >
>
> > http://mailman.ucar.edu/mailman/listinfo/ncl-talk
> >
>
> >
>
>
>
>
> --
> ------------------------------------------------------------------------------------------
> The views expressed in this email do not necessarily reflect those of NOAA, the National Weather Service, or the University of Oklahoma.
>
>
> ------------------------------------------------------------------------------------------
> James Correia Jr.
> OU CIMMS Research Associate
> SPC HWT Liaison
>
>
>

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Sat Mar 26 13:22:32 2011