Re: mass-conservative regridding

Wenshan

As a general rule (whether using NCL or some other language or tool), if you a looking at a variable where mass conservation is important you should NOT regrid; no regridding technique (at least none that I know of) will conserve mass. This is well known in oceanography, and is one of the reasons why e.g. CMIP5 data are released on each model's native grid rather than a common, regular lat/lon grid which would be much easier to work with.

If mass conservation is important for your task (and it sounds like it is) then you need to plot and calculate on the data's native grid.

Will

From: Wenshan Wang <wenshanw@uci.edu<mailto:wenshanw@uci.edu>>
Date: Tuesday, 18 February 2014 7:49 AM
To: Dennis Shea <shea@ucar.edu<mailto:shea@ucar.edu>>
Cc: ncl-talk <ncl-talk@ucar.edu<mailto:ncl-talk@ucar.edu>>
Subject: Re: mass-conservative regridding

Hi Dennis,

Thank you very much for the reply.
But I am sorry I didn't quite get it.
I mean I have a global emission data map. No matter the resolution is, the total emission should be the same, right?
I am trying to regrid 1x1 degree fixed grid to 1.9x2.5 degree fixed grid.
The total emission of 1x1 is about 70,000, but either using ESMF or area conservative remap, the total emission of 1.9x2.5 is no larger than 20,000. I guess it is because 99% grid points of my emission map are 0, the difference of average is not large.

I found in ncl-talk that we can re-scale the regridded to map to match the original total emission, 70000. But it was an old message. So, right now, ESMF is the best we have, right?

Thank you for your time and patience!

Best,
Wenshan
------------------------------------------------------------------------------
Graduate Student Researcher
Earth System Science
University of California, Irvine

On Sat, Feb 15, 2014 at 9:36 AM, Dennis Shea <shea@ucar.edu<mailto:shea@ucar.edu>> wrote:
ESMF is as 'state of the art' as it gets.

re: "The avg is good enough but the sum of all grids is quite off."

This is rather obscure. The "sum of all grids" means nothing.
If you had 100 grid points an the other grid had 1000, the
sums would be different.

On 2/13/14, 2:03 PM, Wenshan Wang wrote:
Hi all,

I am now trying to regrid emission data, which asks for mass-conservative.
I found a message talking about emission regridding about 6 years ago.

I was wondering whether NCL has some new functions that can perform a
mass-conservative regrid since then.

I tried 'area_conserve_remap', 'area_hi2lores' and ESMF conservative method.
The avg is good enough but the sum of all grids is quite off.

Thank you!

Best,
Wenshan
------------------------------------------------------------------------------
Graduate Student Researcher
Earth System Science
University of California, Irvine

_______________________________________________
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 Mon Feb 17 14:40:12 2014