Re: calculating total emissions

[I meant to send this earlier, but got side-tracked. Some of what I
say below echoes Dennis' comments.]

Hi Lei,

Do loops in any scripted language can be slow, and especially nested
do loops.

Since you can do array arithmetic in NCL without having to loop, I
highly recommend using this feature. Also,, anything inside a do loop
that doesn't need to be there should be moved outside the loop.

Here's another UNTESTED version of your code, that's a bit shorter.
This is probably not quite right, but hopefully you get the idea:

alat2d = conform((/natime,nalat/),alat,1)
rlat2 = (alat2d+1.429)*3.14159265/180.0 ; convert degree to radians
rlat1 = (alat2d-1.429)*3.14159265/180.0
temp = sin(rlat2)-sin(rlat1)
diff_lon = 2.8152d*3.14159265d/180.0d
R = 6371.0d*1000.0d ; unit is meterncd
area = R*R*diff_lon*temp
area = where(area.lt.0,0.0,area)
totalfch4 =0.0d

do i =0, natime-1
   do j=0,nalat-1
     kk = ind(.not.ismissing(aFCH4(i,j,:)))
     if(.not.all(ismissing(kk))) then
       totalfch4=sum(aFCH4(i,j,kk)*area)
       afch4(i) = afch4(i)+totalfch4
     end if
     delete(kk)
   end do
end do

On Dec 15, 2009, at 1:34 PM, Lei Meng wrote:

> Hi,
> I was wondering if there are some faster ways to calculate
> globally total emissions. I used the following code and it looped
> over each grid cell to calculate total values. It takes so long to
> finish the calculation.
> --------------------------
> totalfch4=0.0d
> do i =0, natime-1
> do j=0,nalat-1
> rlat2= (alat(j)+1.429)*3.14159265/180.0 ; convert degree
> to radians
> rlat1= (alat(j)-1.429)*3.14159265/180.0
> temp1 = sin(rlat2)
> temp2 = sin(rlat1)
> temp = temp1-temp2
>
> do k=0,nalon-1
> if(.not.ismissing(aFCH4(i,j,k))) then
> diff_lon = 2.8152d*3.14159265d/180.0d
> R=6371.0d*1000.0d ; unit is meterncd
> area=R*R*diff_lon*temp
> if(area.lt.0) then
> area=0.0
> end if
> totalfch4=aFCH4(i,j,k)*area ;
> convert from kgC/m2/s to kgC/s by multiplying area
> afch4(i) = afch4(i)+totalfch4
> end if
> end do
> end do
> end do
> ------------------------------------
> Any suggestions would be appreciated. Thanks,
> Lei
> _______________________________________________
> 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
Received on Tue Dec 15 21:01:35 2009