[NARCCAP-discuss] why 50km x 50km
Gutowski, William J [GE AT]
gutowski at iastate.edu
Thu May 23 23:02:45 MDT 2013
Jianhua:
It is important to realize that NARCCAP started about 8 years ago. At
that time, 50-km was largely the state-of-the-science for RCM simulation,
especially for such a large domain for multiple decades of simulation. As
you may know, the computing demand for numerical simulation increases
inversely with the cube of resolution. Thus, a 10 km grid for the same
domain would impose 5x5x5 = 125 times more computing demand. That might
not matter if computers were so fast that that increased demand was not an
issue, but unfortunately, that's not the case. Doing all the NARCCAP
simulations at 10-km grid spacing would require about 160 Million
core-hours. For comparison, NCAR's annual allotment on the new
Yellowstone machine is about 170 M core-hours. Thus, even today, it would
be difficult to justify such a demand on computing time, even if spread
out over 3-4 years.
In addition, the finer resolution means a much larger volume of data to
post-process and serve to users. The NARCCAP data set is about 40-50 TB
at 50-km grid spacing. Going to 10-km resolution would push that up to
about a petabyte or more.
Beyond that, there always are going to be yet finer processes that one
would like to include, but there are important questions that must be
addressed before launching into finer resolution, such as - will the
models do better at finer resolution? The long history of numerical
modeling is that eventually finer resolution helps, but initially,
simulations often get worse, and it can take many months or even years to
figure out why.
Ultimately, the key issue (at least for impacts studies) is getting
climate information at appropriate scales, and that is not necessarily
synonymous with producing numbers on a finer grid. The most efficient
way to do that may not be to run an RCM at yet finer resolution but to
learn how to use a combination of dynamical simulation and statistical
techniques to produce it.
Good question - but lots more we need to learn and be able to do before
we can truly do a "NARCCAP" at that grid spacing.
Bill
--
William J. Gutowski, Jr.
3021 Agronomy Hall
Dept. of Geological & Atmospheric Sciences
Dept. of Agronomy
Iowa State University
Ames, Iowa 50011-1010
Tel: 1-515-294-5632
Fax: 1-515-294-2619
http://www.ge-at.iastate.edu <http://www.ge-at.iastate.edu/>
http://rcmlab.agron.iastate.edu <http://rcmlab.agron.iastate.edu/>
On 5/23/13 4:39 PM, "Jianhua Huang" <jianhua.huangsdu at gmail.com> wrote:
>Dear all:
>
>
>
>Anyone knows why the NARCCAP models choose the 50km x 50km resolution?
>This
>looks a naïve question, but I was asked in a recent presentation. I
>checked
>the NARCCAP website thoroughly, but found no answer. Why the resolution is
>50km x 50km, why not 10km x 10 km? Is 50 km resolution good enough to
>catch
>the spatial heterogeneity of climate change in the United States?
>
>
>
>Thanks for any clue or suggestion.
>
>
>
>Jianhua
>
>_______________________________________________
>narccap-discuss mailing list
>narccap-discuss at mailman.ucar.edu
>http://mailman.ucar.edu/mailman/listinfo/narccap-discuss
More information about the narccap-discuss
mailing list