Re: Question on the time coordinate of IPCC AR4 model output data?

Lin,

There are several different calendar systems used in IPCC output
files. Your script forces the Gregorian calendar interpretation on
the time coordinates in all cases. This will result in severe time
shifts and invalid analytical results, including the exact kind of
inconsistencies that you showed.

The latest NCL version 5.1.1 has an upgraded calendar function that
supports most of the IPCC calendar types. (Also 5.1.0 I think, but
5.1.1 is better.) Please ensure that you are using a new version of
NCL, and that you are not overriding the calendar type of your time
coordinates.

It takes some care and double checking in the application to get
this right, since the definition of "what is a month" changes from
calendar to calendar.

There is an alternate calendar function that works with older
versions of NCL. It supports IPCC calendar types including one or
two that are not handled by the latest NCL version. Its
functionality is limited in other ways. Also I have not checked
lately to make sure it still works with NCL 5.1.1. If interested,
download a copy here:

   http://www.cdc.noaa.gov/people/dave.allured/data/ncl/lib/
   (wait a few minutes, I just fixed this link...)

Also see a related discussion in the NCL archives, starting 2008-oct-20.

Dave Allured
CU/CIRES Climate Diagnostics Center (CDC)
http://cires.colorado.edu/science/centers/cdc/
NOAA/ESRL/PSD, Climate Analysis Branch (CAB)
http://www.cdc.noaa.gov/psd1/

leo.aries.g wrote:
> Hi, all
>
> I met some problems when dealing with monthly data of some IPCC AR4
> model output.
> The problem is from the time coordinates, in general there are more than
> 1 value for some months
> in contrast to no values for some other months.
>
> For example , one output of psl (pressure at sea level) for GFDL in
> sresa1b scenario,
> with a filename "psl_A1.200101-220012.nc" , '
> downloaded from
> "ftp-esg.ucllnl.org/data5/sresb1/atm/mo/psl/gfdl_cm2_0/run1/psl_A1.200101-220012.nc"
>
> Here below is my scripts:
>
> path_name = "..."
> ymStrt = 200101
> ymLast = 209912
>
> fi = addfile(path_name + "psl_A1.200101-220012.nc" , "r")
> time = fi->time
> time_at_calendar = "gregorian"
> yyyymm = ut_calendar(time,-1)
> ntStrt = ind(yyyymm.eq.ymStrt)
> ntLast = ind(yyyymm.eq.ymLast)
> data = fi->"psl"
> time2 = yyyymm(ntStrt:ntLast)
>
> ntime = dimsizes(time2)
> time4 = new((/ntime/),integer)
> time4(0:ntime-2) = time2(1:ntime-1)
> time4(ntime-1) = 0
> nt = ind(time2.eq.time4)
> print(nt) ;
> here I got 4 output ( nt= 720, 732,744,756 ), corresponding to the cases
>
> with more than 1 values for some months
>
> yyyymmdd = ut_calendar(time,-2)
> print(time2(718:758) + " " + yyyymmdd(718:758)) ; to get more
> details, print the time coordinates in monthly and daily resolution
>
> ; Here below is the results
> (0) 206011 20601101
> (1) 206012 20601201
> *(2) 206101 20610101*
> *(3) 206101 20610131*
> (4) 206103 20610301
> (5) 206104 20610401
> (6) 206105 20610501
> (7) 206106 20610601
> (8) 206107 20610701
> (9) 206108 20610801
> (10) 206109 20610901
> (11) 206110 20611001
> (12) 206111 20611101
> (13) 206112 20611201
> *(14) 206201 20620101*
> *(15) 206201 20620131*
> (16) 206203 20620301
> (17) 206204 20620401
> (18) 206205 20620501
> (19) 206206 20620601
> (20) 206207 20620701
> (21) 206208 20620801
> (22) 206209 20620901
> (23) 206210 20621001
> (24) 206211 20621101
> (25) 206212 20621201
> *(26) 206301 20630101*
> *(27) 206301 20630131*
> (28) 206303 20630301
> (29) 206304 20630401
> (30) 206305 20630501
> (31) 206306 20630601
> (32) 206307 20630701
> (33) 206308 20630801
> (34) 206309 20630901
> (35) 206310 20631001
> (36) 206311 20631101
> (37) 206312 20631201
> *(38) 206401 20640101*
> *(39) 206401 20640131*
> /*(40) 206402 20640229*/
>
> ; remove the seasonal cycle
> x = rmMonAnnCycLLT(data({lat|-60.:-50.},lon|:,time|ntStrt:ntLast)) ;
> rmMonAnnCycLLT can not run because time dimension was not
>
> multiple of 12 (1 more value added between 20640101 to 20640229)
>
>
> Here is only one example, similar cases were often met when I deal with
> IPCC AR4 model output.
> What cause it and how should I deal with them? Does anyone know about
> these and can give me some advice?
>
> Thank!
>
>
> Lin
>
> Chinese Academy of Meteorological Sciences (CAMS),CMA
>
>
> 2009-07-07
> ------------------------------------------------------------------------
> leo.aries.g
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> 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 Jul 06 2009 - 13:12:04 MDT