Opened 11 years ago

Closed 10 years ago

#416 closed help (fixed)

Ice edge field

Reported by: a.elvidge Owned by: jeff
Component: UM Model Keywords:
Cc: Platform:
UM Version: 7.1

Description

Hi,

I am attempting to create a new sea ice ancillary file from an outside source (OSTIA). I have successfully interpolated the new sea ice fraction data so that it is identical (i.e. set to the same grid and land sea mask) to the "iceconc" field within the original "qrclim.ice" ancillary file.

I am unsure however as to what to replace the "ICE EDGE - IN THE ANCILLARY FILE ONLY" (iceconc_1) field with. What is this exactly? Having plotted it out it doesnt seem entirely obvious to me..

Thanks, Andy

Change History (14)

comment:1 Changed 11 years ago by willie

Hi Andy,

UMDP C7 mentions sea ice edge indirectly. I have some documents on the sea ice model (UMDP 055 and UMDP 45) for UM6.1 that I'll email to you.

Are you aware of the central ancillary program? - see http://ncas-cms.nerc.ac.uk/CAP_INTERFACE/cap_general.php - this will generate sea ice ancillary files.

I hope that helps,

Regards,

Willie

comment:2 Changed 11 years ago by a.elvidge

Hi Willie,

Thanks for the documents. Yes I am aware of the ancillary file service, but I need higher resolution, non-climatolgical sea ice fields.

Thanks, Andy

comment:3 Changed 11 years ago by jeff

Hi Andy

If you use xancil (http://ncas-cms.nerc.ac.uk//index.php?option=com_wrapper&Itemid=274) that will create the ice edge fields for you. You need to have you input files in NetCDF format.

Jeff.

comment:4 Changed 11 years ago by a.elvidge

Hi Willie, Jeff
Thanks, turns out the ice edge field is ignored anyway.
Andy

comment:5 Changed 11 years ago by jeff

  • Owner changed from um_support to jeff
  • Status changed from new to accepted

Hi Andy

What makes you think the ice edge field is ignored? As far as I know it is used.

Jeff.

comment:6 follow-up: Changed 11 years ago by a.elvidge

Hi Jeff,

I probably should have said that it is not necessarily required rather than it is ignored. In xancil there is an option as to whether you want to calculate sea ice edge or not. And when I try calculating it, it gives me a blank data set. Also, under the sea ice ancillary options there are 2 tick boxes - for configuring sea ice proportion and for configuring sea ice thickness, but nothing concerning sea ice edge.

Cheers, Andy

comment:7 follow-up: Changed 11 years ago by a.elvidge

Hi Jeff/Willie?,

Regarding my sea ice ancillary - does the UM recognise 'nan' (not a number) as an invalid number (i.e. land), or does it require a -1.0737e+09?

I seem to remember attempting to specify a number which was not within the specified valid range, but that when I plotted it out in xconv, it didnt seem to recognise it as an invalid number, and instead plotted it out as the large negative number that it is. Hence I used nan's instead. I had forgot that this was the case and ran the model with these ancils included. However, to my surprise the model appears to run fine. So, does the UM indeed recognise nan's?

Additionally, I notice that in the default sea ice ancillary files, ice depth is specified in the southern hemisphere as 2m. However, when calculating ice depth for my new ancils it automatically gave me thicknesses of 1m. Is this to be expected? Also, again in contrast to the ice depth field in the default sea ice ancil, thicknesses over the land are given as 0m rather than an invalid value.

Thanks for any help,

Andy

comment:8 in reply to: ↑ 6 Changed 11 years ago by jeff

Hi Andy

I probably should have said that it is not necessarily required rather than it is ignored. In xancil there is an option as to whether you want to calculate sea ice edge or not. And when I try calculating it, it gives me a blank data set. Also, under the sea ice ancillary options there are 2 tick boxes - for configuring sea ice proportion and for configuring sea ice thickness, but nothing concerning sea ice edge.

Xancil shouldn't give you a blank data set unless the ice fraction field is always the same for every time, if this isn't the case then there is something wrong. If you only have 1 time in the ancillary file then the ice edge will be blank (actually all missing data values) and this isn't a problem.

The sea ice edge field controls the time interpolation of sea-ice fraction/thickness and sst (over sea ice) from the fields in the ancillary files. It isn't used elsewhere in the UM so isn't a configured or updated field.

Jeff.

comment:9 in reply to: ↑ 7 Changed 11 years ago by jeff

Hi Andy

Regarding my sea ice ancillary - does the UM recognise 'nan' (not a number) as an invalid number (i.e. land), or does it require a -1.0737e+09?

I seem to remember attempting to specify a number which was not within the specified valid range, but that when I plotted it out in xconv, it didnt seem to recognise it as an invalid number, and instead plotted it out as the large negative number that it is. Hence I used nan's instead. I had forgot that this was the case and ran the model with these ancils included. However, to my surprise the model appears to run fine. So, does the UM indeed recognise nan's?

You shouldn't use NaN in the UM, it will probably be okay if the UM doesn't ever try and read these data values. If it does read them on some machines it will crash on others it may not do what you expect. Sometimes the UM checks the data values against RMDI (-230) and this won't work if they are NaNs?.

I don't believe you can specify a valid range in a UM file only a missing data value so was this a netcdf file? If it was and xconv plotted a value outside the valid_min/valid_max or valid_range attribute values, then this is a bug and I will look into it.

Additionally, I notice that in the default sea ice ancillary files, ice depth is specified in the southern hemisphere as 2m. However, when calculating ice depth for my new ancils it automatically gave me thicknesses of 1m. Is this to be expected? Also, again in contrast to the ice depth field in the default sea ice ancil, thicknesses over the land are given as 0m rather than an invalid value.

This doesn't sound right, this was with xancil correct? Let me know where the files are you used to create this ancillary file and I will take a look.

Jeff.

comment:10 Changed 11 years ago by a.elvidge

Hi Jeff,

The files are in hector under work/n02/n02/aelvidge/UMAncil/Ostia/

Within this directory there are 2 further directories: 'withnan' and 'with2e20'.
The withnan refers to where I have specified invalid values with a NaN, the with2e20 where I have specified these values as being equal to the "var_missing_value" value i.e. 2*1020. In both folders are the following files:

ostia_sea_ice_fraction.nc
ostia_analysed_sst.nc
ice
sst

'ostia_sea_ice_fraction.nc' and 'ostia_analysed_sst.nc' are the netcdf files I fed into xancil to give the UM ancil files 'ice' and 'sst' respectively. You should have full access to all these files.

1) Sea Ice field - you say that "If you only have 1 time in the ancillary file then the ice edge will be blank (actually all missing data values) and this isn't a problem." Indeed this is the case, so this is not a problem.

2) I used the 'withnan' ancil files successfully in a UM job. I assume this is because, as you say, "it will probably be okay if the UM doesn't ever try and read these data values". The reason I used NaNs? initially rather than 2e20 is because, upon plotting out the ancils in xconv, the 2e20s are read as numbers rather than invalid regions. - you can check this out for yourself.

3) Finally note that the sea ice thickness field gives a uniform value of 1m (rather than 2m) for areas of sea ice, and that thicknesses over the land are given as 1m rather than an invalid value.

Although the files appear to work ok in the model, I would like to ensure they are set up correctly. So, any help you can offer is much appreciated.

Thanks, Andy

comment:11 Changed 11 years ago by jeff

Hi Andy

Do you have the xancil .job files (or namelist files if not), you used to create the ancillary files?

Jeff.

comment:12 Changed 11 years ago by a.elvidge

Hi Jeff,

Yep, both .job files are in the same Ostia directory.

Cheers, Andy

comment:13 Changed 11 years ago by jeff

Hi Andy

I've looked at your netcdf files and think I have found the problem, your variables have type float (real) but the attribute missing_value has type double (double precision). Although its possible to do this it doesn't make any sense because any software which wants to compare data values to the missing_value (i.e. xancil) will not be able to do so correctly. I have written a work around for the next version of xancil which should now work for this case. To make correct ancillary files either remake your netcdf files with float type for missing_value or convert your land/sea mask ancillary file into netcdf and use that in xancil to create the land mask.

As for the sea-ice thickness, as far as I'm aware the thickness value should be 1m in the south and 2m in the north, do you have any reason to think otherwise?

Also I noticed your netcdf files haven't defined the time value, this seems to upset xconv but not xancil.

Jeff.

comment:14 Changed 10 years ago by jeff

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