Opened 11 years ago

Closed 11 years ago

#420 closed help (fixed)

Inconsistency check failure for new ancillary file

Reported by: a.elvidge Owned by: willie
Component: UM Model Keywords: ancillary file format, LNER test
Cc: Platform:
UM Version: 7.1

Description

Hi,

I am getting the following error when trying to use a new sst ancillary file. Looking at previous tickets, I see that this is a problem with a consistency check that the um performs. Unlike the previous tickets, however, 2 of my new ancil values (files 5 and 6) are way off those of the model dump values. What do files 5 and 6 refer to exactly? It also looks like files 1 and 4 are also narrowly outside the 1E-6 threshold.

1 Anc File 3.60107421875E-2 Model Dump 3.59999999999999973E-2
2 Anc File 3.60000133514404297E-2 Model Dump 3.59999999999999973E-2
3 Anc File -3.4000000953674316 Model Dump -3.3999999999999999
4 Anc File 175.89999389648438 Model Dump 175.90000000000001
5 Anc File 157.25 Model Dump 22.75
6 Anc File 475.75999999999999 Model Dump 295.75999999999999

Error in INANCILA
CMESSAGE INANCILA: REAL header Error.
ErrorStatus? 8

*
ERROR!!! in reconfiguration in routine Rcf_Ancil_Atmos
Error Code:- 8
Error Message:- INANCILA: REAL header Error.
Error generated from processor 0
*

Not sure if it is useful but here is an ncdump on my new ancil:

netcdf ostia_analysed_sst {
dimensions:

x = 216 ;
y = 190 ;
surface = 1 ;
t = UNLIMITED ; (1 currently)

variables:

float temp(t, surface, y, x) ;

temp:name = "temp" ;
temp:title = "SURFACE TEMPERATURE AFTER TIMESTEP" ;
temp:date = "01/01/00" ;
temp:time = "00:00" ;
temp:long_name = "SURFACE TEMPERATURE AFTER TIMESTEP" ;
temp:units = "K" ;
temp:north_pole = 475.76, 157.25 ;
temp:missing_value = 2.e+20 ;
temp:_FillValue = 2.e+20 ;
temp:valid_min = 271.0636f ;
temp:valid_max = 273.71f ;

double x(x) ;

x:units = "degrees_east" ;
x:point_spacing = "even" ;

double y(y) ;

y:units = "degrees_north" ;
y:point_spacing = "even" ;

double t(t) ;

t:units = "days since 0000-01-01 00:00:00" ;
t:time_origin = "01-JAN-0000:00:00:00" ;

float surface(surface) ;

surface:units = "level" ;
surface:positive = "up" ;

global attributes:

:title = "analysed_sst ancillary file" ;
:history = "02-May-2010 15:58:32: ostia_analysed_sst\n",

"02-May-2010 15:58:32: Edited by Andrew Elvidge" ;

The ancillary file is on hector under work/n02/n02/aelvidge/UMAncil/AP4km5/sst
You should have access to it.

Help is much appreciated.

Cheers, Andy

Change History (8)

comment:1 Changed 11 years ago by willie

  • Keywords ancillary file format added
  • Owner changed from um_support to willie
  • Status changed from new to assigned

Hi Andy,

They're not files, but entries within the file. Paper F3 on ancillary file formats gives,

1. EW (x) grid spacing in degrees
2. NS (y) grid spacing in degrees
3. Latitude of first PTR row in degrees (latitudes in range 90 to .90)
4. Longitude of first PTR point on row in degrees( longitudes in range 0.360)
5. Real latitude of `pseudo' N pole in degrees
6. Real longitude of `pseudo' N pole in degrees

So your ancillary file has unusual values for the rotated pole compared to the model dump.

I hope that helps,

Regards,

Willie

comment:2 Changed 11 years ago by a.elvidge

Thanks, I have adjusted the poles so that they are now correct. However, the conistency test will still fail because of the following:

  1. EW (x) grid spacing in degrees

Anc File 3.60107421875E-2 Model Dump 3.59999999999999973E-2

difference = 0.000011

AND

  1. Longitude of first PTR point on row in degrees( longitudes in range 0.360)

Anc File 175.89999389648438 Model Dump 175.90000000000001

difference = 0.000006

These values are practically identical - is there any way of bypassing this check or decreasing its sensitivity?

Cheers, Andy

comment:3 Changed 11 years ago by willie

  • Keywords format, LNER test added; format removed

Hi Andy,

You can decrease the sensitivity of the test by modifying the code. The definition of the LNER test is at the top of the subroutine where the error message is written. The procedure for modifying the code can be found in the FCM tutorial at http://puma.nerc.ac.uk/trac/UM_TUTORIAL/wiki/UmTutorial/vn7.1/CodeChanges

You may have to login to the tutorial page to see this.

Regards,

Willie

comment:4 Changed 11 years ago by a.elvidge

Hi Willie,

What do you mean by the 'definition of the LNER test'? I have looked through the .leave output and above the error message I see no reference to the LNER test..? (you have permission to ses my .leave file (job xenol), if you wish to see it.)

Thanks, Andy

comment:5 Changed 11 years ago by willie

Hi Andy,

If you go to the source code browser http://ncas-cms.nerc.ac.uk/index.php/um-code and select Um7.1, you can type in the error message in the search panel on the left. There are two responses and one of them is the reconfiguration. If you select the reconfiguration and scroll up to near the top you will find the lines

      REAL P1,P2
      LOGICAL LNER
      LNER(P1,P2) = ((ABS(P1-P2))  >   (1.E-6*ABS(P1+P2)))

and this is the line you need to change.

regards,

Willie

comment:6 Changed 11 years ago by a.elvidge

Willie,

I dont appear to be able to access the source code - I am asked for a user name and password, but the one I use here does not work..?

Cheers, Andy

comment:7 Changed 11 years ago by a.elvidge

Hi,

This is solved now. The code change worked.

Thanks,
Andy

comment:8 Changed 11 years ago by willie

  • Resolution set to fixed
  • Status changed from assigned to closed
Note: See TracTickets for help on using tickets.