Hi Adnan, Can you describe your problem? Where does the model stop? Has rfcst.x run and produced r_fcstout ? or the model stopped before rfcst.x ? Hideki

scchan.2009

Segementation Fault with rfcst.x with a * spdmx

scchan.2009 09 October 2008 20:02:32

I was trying the rsim script with a new domain size (288x191 -- a size that should be okay according to the guidelines), and I had received the following error in r_fcstout.ft24 :

0nmc regional spectral model-current version created march 2003
ncpus = 1
con(i),i=1,50
-1. 0. 0. 6. 6. 0. 24. 0. 6. 0.
6. 21600. 0. 1. 1. 0. 312. 0. 1. 24.
6. 6. 0. 0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
rsm: run regional only with global files
fftfax is called in rftini. igrd2 = 576
in latbnd beta= 1.666666666666667E-002 factor= 15.0000000000000
------ initiate latbnd finished ----
------- into rsminp -----
SIGANLB2

----- begin rsetsig ----
delx dely 60000.0000000000 60000.0000000000
fixio field read in from unit= 32
fh, idate= 0.0 0 11 1 2002
fixrdrec completed for ts
fixrdrec completed for smc
fixrdrec completed for sno
fixrdrec completed for stc
fixrdrec completed for tg3
fixrdrec completed for z0
fixrdrec completed for cv
fixrdrec completed for cvb
fixrdrec completed for cvt
fixrdrec completed for alb
fixrdrec completed for sli
fixrdrec completed for vegcov
fixrdrec completed for canop
fixrdrec completed for f10m
fixrdrec completed for vegtyp
fixrdrec completed for soiltyp
fixrdrec completed for albf
fixrdrec completed for ustar
fixrdrec completed for fm
fixrdrec completed for fh
fixrdrec completed for prcp
fixrdrec completed for srflag
fixrdrec completed for snodph
fixrdrec completed for slc
fixrdrec completed for shdmin
fixrdrec completed for shdmax
fixrdrec completed for slope
fixrdrec completed for snoalb
fixio completed.
hprimes has max=0.1213E+04 min=0.5144E-01 k= 1 nzero= 43012 nmis= 0
------ into rsmsmf -----
shour= 0.000000000000000E+000
run rsm from 0.000000000000000E+000 to 21600.0000000000
------ into rsmltb -----
in rsmltb nestsecrdt 21600 4.629629629629629E-005
archv data from da,mo,yr= 0 3 66
...last date/time and current itim
0 1 1 90 8760.0
1
co2 concentration is 3.480000000000000E-004
1 ozone climatology for month,day= 11 1
------------initial step------------------
kdt before smooth start= 1 dt= 3.0
rsm spdmx(01:10)=*** 26.77 29.51 27.60 28.1
0 30.16
rsm spdmx(11:20)= 29.75 31.08 30.40 32.16 35.71 36.68 44.10 52.97
opening file r_pgb.ft00
set both input and output indices
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source

rfcst.x 000000000049CA32 Unknown Unknown Unknown
rfcst.x 000000000046D962 Unknown Unknown Unknown
rfcst.x 0000000000469022 Unknown Unknown Unknown
rfcst.x 0000000000408F04 Unknown Unknown Unknown
rfcst.x 0000000000404AAB Unknown Unknown Unknown
rfcst.x 0000000000403702 Unknown Unknown Unknown
libc.so.6 0000003F2361D8B4 Unknown Unknown Unknown
rfcst.x 0000000000403629 Unknown Unknown Unknown

It has been listed that reducing the time step size will fix a blown up spdmx. All DELTAT_REG_* are set to 6 when this error occurred. The issue with RAND() is also changed in the code.

What else I can try to fix this problem? Thank you very much.

mkanamitsu

mkanamitsu 10 October 2008 01:54:25

Before jumping in to the rsim script, please try rsm script first. Your error may be coming from the specification of initial condition and boundary data. Running the rsm script will isolate this problem.

scchan.2009

mpirun -machinefile

scchan.2009 14 October 2008 18:28:31

Hello,

I have been trying to utilize the 4 out of the 8 local processors on the computer I am using to do some test rsim script runs (so I am not calling another machine to run the model). And I am receiving the following errors:

...
+ mpirun -machinefile -np 4 ./rfcst.x
+ '[' 255 -ne 0 ']'
+ echo 'r_fcst failed'

with r_fcstout.ft24 reporting:

unable to open machinefile

Hello,

I have been trying to get some of my colleagues to be able to work with ERA-40 data. I have gotten into some obstacles -- some I managed to solve, but some I do not.

First thing ins the p2sig(_era) script. Using the version of RSM that is released before the RSM-ROM announcement I have gotten on 10/01/2009, the file: orogrd.smth -- needed for p2sig is not created on my Intel compiler using systems. I have managed to create one by brute force by manually executing mtn to create orogrd.smth.

The second problem is that orogrd.smth generated by a low-res topo installation appears to be incompatible with a p2sig compiled for hi-res topo. Default mtn.sh in hi-res topo compile seeks for libs/con/top0.5m_???.* which are files do not exist.

I will post more findings when I am looking more into this.

Thanks
Steven Chan

kei

kei 02 October 2009 18:04:13

Steven,

I recommend to compile default GSM at somewhere else. In GSM, orogrd.smth and p2sig programs should be made without any problem. You can use the created global orography files (orogrd.smth) for the p2sig programs. Then the created global spectra files (aka "sig" files) will be used for initial and boundary conditions for your RSM runs.

Currently, there is inconsistency for (global) mtn and p2sig programs while GTOPO30 option is used for RSM. (It is possible to resolve the issue by yourself, but little tricky.)

Kei

scchan.2009

ieee2grb_sst, RSM's w3lib and NCEP's w3lib (v1.7)

scchan.2009 16 December 2009 21:50:41

Hello All:

I have been trying to check a certain issue with ieee2grb_sst converted grib files and run into a few interesting issues and obstacles:

Recently I obtained some CFS data from NCEP (and its SST grib files). Those original grib files well with G-RSM. Then I try to do bias correction to those CFS SST by converting to Grads format, do the bias correction, and use ieee2grb_sst to convert back to grib, but something related to the header(?) has changed and lead to sfc0.x to crash.

The original NCEP grib files has this as part of sfc0.x output, and reads the file size as 16586880:

==== start reading sstskint ===

status= 25 -1945108103 33188 1 1079 1079
0 16586880 1260819706 1260819706 1260819706 4096

the ieee2grb_sst generated grib file gives the following output from sfc0.out this:

==== start reading sstskint ===

status= 25 -1827306045 33204 1 1079 1079
0 0 1260916266 1260916266 1260916266 4096

the latter leads to an error in naint.F when it reads the file size as 0. I can change naint.F to read 1260916266 instead, but I am not sure if that is the right thing to do. So I wish to ask.

That leads to another issue is that I try to re-compile ieee2grb_sst.x with NCEP's new version of w3lib. While the program compiles, it generates no output file. I can compile ieee2grb_sst.x with RSM's own w3lib, but then that leads to apparent changes I shown above.

Sorry if this is long or confusing post.

Steven Chan

scchan.2009

scchan.2009 18 December 2009 19:25:20

Sorry to have asked.

I fixed the problem with the grib file itself.

vmisra

restarting

vmisra 12 January 2010 16:23:44

I am doing a 30 year run on a remote machine. As and when it completes one year I download the output and get rid of the output directory for that specific year. However, when I want to restart in between the model likes to start from the initial date and does not recognize the restart files residing in another year directory. How do I circumvent this issue?

kei

kei 13 January 2010 21:16:26

Vasu,

Are there all restart files (r_sigit, r_sigitdt, and r_sfci) with the same time stamp from which you like to restart? If not, you should copy the latest r_sig.ftXX and r_sfc.ftXX files by yourself.

Usually rsim script automatically detects the restarting time if you have appropriate files.

Kei

vmisra

version?

vmisra 26 January 2010 22:19:30

how to find the version of the model we are running? is there some cvs command
-Vasu

mkanamitsu

mkanamitsu 28 January 2010 00:07:57

cvs --help will list all the cvs command and there are several. One is cvs st.

scchan.2009

NCAR frost

scchan.2009 17 February 2010 20:22:28

Hello Everyone:

I am trying to get RSM to run on frost on NCAR. The xlf compiler in frost behaves quite differently than the xlf compiler on other NCAR machines, and I have noticed large incompatibilities dealing how the compiler reacts to "free form" and "fixed form" FORTRAN (especially it has difficulty if they are "mixed" in the same code). Removing/adding the compiler option -qfixed generates different type of errors.

I have been applying to fix to individual code (and I am still working in the lib directory) by inserting compiler specific directives into the pre-cpp code.

I am wondering if anyone has a better idea how can to do this be done more efficiently, and in a way that make this easier for a future RSM user that may be dealing with system like frost.

Thank you
Steven Chan

mkanamitsu

mkanamitsu 17 February 2010 21:01:47

Please post the error message, both for free form and fixed form.

We have never encountered such problem. If you are using DOS formatted source code file by mistake, it will cause all kinds of problems.

vmisra

What is vegetation fraction

vmisra 20 April 2010 15:35:00

What is vegetation fraction? It seems to change both in time and space. We already know that it is percentage of vegetation covered in a grid box. So when it varies with time, does it mean a fraction disappears and reappears? Is this fraction also related to leaf area index or greeneness?

mkanamitsu

mkanamitsu 20 April 2010 16:42:45

Vegetation fraction is a function of season.

ibooprok

GSM with new NCEP T574L64

ibooprok 02 August 2010 03:13:51

July 27, 2010, NCEP change resolutionn of GFS to T574L64.
I have been running GSM using GFS output. Do I need to patch the GSM codes?

ibooprok

ibooprok 13 August 2010 05:41:14

The initial condition from NCEP GDAS:

Initial date (2010080400) for:
1. gdas1.t00z.sanl 511376936 bytes
2. gdas1.t00z.sfcanl 272591284 bytes

Initial date (2010080300) for :
3. gdas1.t00z.engicegrb.grib2 55280 bytes
4. gdas1.t00z.snogrb.grib2 38196 bytes
5. gdas1.t00z.sstgrb.grib2 50477 bytes

grib2 had converted to grib1.

scchan.2009

Minor issue in incdte

scchan.2009 10 April 2011 14:34:28

Hello All:

Today, I notice for some previous version of RSM (at least the version up to middle of 2010) that incdte computes years that are divisible by 100 (unless divisible to 400 as well) as leap years. That is incorrect.

The change to address the problem is quick. I have made the change on my own copy of the source.

Cheers
Steven Chan

kei

rhg11c 05 December 2016 23:42:44

Hello,

I would like to find out how RSM averages the output r_pgb files for common variables like temp, u, v etc. For example if I have a run in which INTPGB=24 so that you get daily data, and then you run the exact same run but with INTPGB=1(hourly), can you average the hourly run such that you get back the daily data? I am trying to verify that 2 runs are from the exact same setup just one is daily and one is hourly.

Hourly run:
INCHOUR=24
INTSFCX=24
INTPOST=1
INTPGB=1
INTRESTART=1
Daily run:
INCHOUR=24
INTSFCX=24
INTPOST=24
INTPGB=24
INTRESTART=24

Thank you

kei

kei 06 December 2016 00:29:13

There are two types of saved variables; instantaneous and time-averaged. Fluxes (energy and water, moisture flux) are time-averaged, and temperature, humidity and wind, etc. are instantaneous. Averaging hourly data into daily would make different results in instantaneous variables, whereas the identical results come out in fluxes. Furthermore, extreme values (such as tmax and tmin) are within the specified range. Therefore, selecting the maximum value among 24 hourly-tmaxes would be identical as daily tmax, for example.

Kei

rhg11c

RSM low, middle and high cloud amounts using Slingo scheme

rhg11c 20 January 2017 23:28:12

Hello,

So in my understanding of the Slingo scheme code in RSM, the Slingo scheme groups clouds into 3 layers for low, middle and high clouds and it determines where the top and bottom pressure are for each of these 3 cloud layers based on some criteria? The top and bottom pressure for each layer can be found in the output for RSM (preslcb=low level cloud layer base). What I am getting at is, the cloud layer top and bottom can vary, that is, they don't stay constant so the pressure levels of low, mid and high level clouds can vary.

Additionally, the cloud amounts calculated in the low, middle and high levels are basically the total cloud amounts for each of those individual levels. Then the vertical column total cloud amount is calculated separately and takes into account overlapping within the whole column. This total cloud amount is in line with what the radiation scheme sees. In other words, Low cld + mid cld + high cld amounts = probably over 100%. And total vertical column cloud amount should be almost always <100%.

Just wanted to clarify that I am not missing something and that this is the correct way to interpret the cloud amounts.

Thank you
Russ

rhg11c

rhg11c 25 January 2017 17:57:12

Is anyone able to help answer this question?

Russ

Welcome to G-RSM Wiki

This is a Wiki forum for the G-RSM system at ECPC, SIO.

Please add your name, institution, and e-mail address to the User List.

Any user can edit any page of G-RSM Wiki.

You are encouraged to contribute to the forum by sharing your ideas about G-RSM.

Main Menu

Discussions

==== start reading sstskint ===

==== start reading sstskint ===