McIDAS V1 beta5 vs V1.5

How do I...?
Post Reply
User avatar
Cloudy
Posts: 25
Joined: Tue Mar 02, 2010 12:15 am

McIDAS V1 beta5 vs V1.5

Post by Cloudy »

When I load the same file, for example from: http://nomads.ncep.noaa.gov/cgi-bin/filter_gfs_0p50.pl? dir = % 2Fgfs.2015122012, in McIDAS-V V1.0 beta5 and also in McIDAS-V V1.5 the list of Fields in Field Selector tab is different. In McIDAS-V V1.5 the precipitation fields are averages of 3 or 6 hours and those of total of clouds low, middle or high do not exist. Is possible to recover the fields presented in beta5 to have them in V1.5? Many thanks. Cloudy.
User avatar
bobc
Posts: 939
Joined: Mon Nov 15, 2010 5:57 pm

Re: McIDAS V1 beta5 vs V1.5

Post by bobc »

Hello -

I'm replicating what you see with this data. If I load my file (gfs.t12z.pgrb2full.0p50.f030) into 1.0beta5 I see a field called "Total_cloud_cover @ high_cloud", and this same field in 1.5 is called "Total cloud cover (6_Hour Average) @ High cloud layer". If I display this same field in 1.0beta5 and 1.5, they look to be the same.

Looking online, it seems like this was a change that was made to the NetCDF/Java library in version 4.3, which changed the way that GRIB variables are named.

For a couple of sources of information, see Announcement of NetCDF/java 4.3.8
- At the bottom of this post, it includes the new and old variable names, for example:
OLD Total_cloud_cover_high_cloud
NEW Total_cloud_cover_High_cloud_12_Hour_Average
NEW Total_cloud_cover_High_cloud_Mixed_intervals_Average

Another source of information is GRIB variable name changes in 4.3. The bottom part of this post (2/27/12 at 11:17AM from John Caron) goes over some explanation of why the changes were made.

Are you actually noticing any fields that are in 1.0beta5 that are missing in 1.5, or is this just confusion over the naming of the fields?

Thanks -
Bob Carp
User avatar
Cloudy
Posts: 25
Joined: Tue Mar 02, 2010 12:15 am

Re: McIDAS V1 beta5 vs V1.5

Post by Cloudy »

Hi Bob, many thanks for you response. Is not a confusion, if I load several files taking a period from, for example 48 hours, in 1.0 beta 5 shows the field every 3 hours, but in 1.5 it takes only the initial period and it does not cover the 48 hours. Same happens with any fields 3 or 6 hours average(Total Clouds or Precipitation and so on). I not understand what happen.
Cloudy
User avatar
bobc
Posts: 939
Joined: Mon Nov 15, 2010 5:57 pm

Re: McIDAS V1 beta5 vs V1.5

Post by bobc »

Hello -

Thanks for the explanation. I was working with one file at a time and thus didn't see the problem. I'm able to replicate this now.

It looks like this came along with the field name changes. If you aggregate the grid files together into one data source using the 'Aggregate Grids By Time' data type, you can notice some differences in the metadata between 1.0beta5 and 1.5. To see the metadata, load the aggregated data source, right-click on the data source in the Field Selector and choose Properties. In this Properties window, go to the Metadata tab.

In 1.0beta5, the only time dimension in the metadata is 'time'. All of the fields have this time dimension. For example:
float Total_cloud_cover_high_cloud(time=17, lat=361, lon=720)

In 1.5, there are two dimensions in the metadata, 'time' and 'time1'. All of the fields that are averages over a time period use 'time1' instead of 'time'. For example:
float Total_cloud_cover_high_cloud_3_Hour_Average(time1=1, lat=361, lon=720);

Notice how time= in 1.0beta5 has a length of 17, and time1 in 1.5 has a length of only 1. This is why you can't get loops of the average fields to display in McV using 'Aggregate Grids By Time', since this is looking for a dimension called 'time' in the fields.

I haven't worked through this myself, but you may be able to create a wrapper file using NcML aggregation, defining 'time1' as the dimension to aggregate with. See NcML Aggregation for more info.

I will send an email to NetCDF support (at Unidata) to try finding out why this change was made. Investigating this data in NetCDF ToolsUI shows that this 'time1' dimension was introduced in NetCDF 4.2.

Thanks -
Bob
Post Reply