error when attempting a scatter analysis

[ghansham]

Hi All.


One more feature.
If I dont have a wheel mouse how do I use zoom in/out Image Displays in Scatter Plot...

I was just trying to understand what actually is happening in this dataset.
Actually as someone already tried a plan view display, the bounding box is getting drawn somewhere else.
And as a result the probe is getting centered on that bounding box and not display.

Another interesting thing.. is that this data in this product is packed along the longitude lines direction.
the bounding box is not getting calculated properly. If I get sometime, I will try to dig why it is happening like that.

regards
Ghansham

[tomw]

I thought a coordinates attribute was required for all 2-D arrays?

This is only required for those "data" parameters that are to be geolocated in order to connect to the arrays that contain the coordinates. Think of it as the "coordinates" attribute naming the variables that contain the geolocation information.

[bobc]

Hi Ghansham -

If I dont have a wheel mouse how do I use zoom in/out Image Displays in Scatter Plot...

Does Shift+right click and drag work for you to zoom in and out?

- Bob

[ghansham]

Hi there..
Sorry but it didnt work for me. I was thinking the same.
Ghansham

[ghansham]

hi Sir..
Sorry for false alarm.. It works...
But I think we can add in some controls like the one available in Map Display
to keep the things simple from user point of view.

Ghansham

[bobc]

Hi Ghansham -

I'm glad that Shift+right click and drag works for you to zoom in on the images. I agree with you that it would be nice to have navigation controls as a part of the scatter analysis display to make it easier for the user to navigate through the displays.

I have written up McIDAS-V Inquiry #1490 for this enhancement:
http://mcidas.ssec.wisc.edu/inquiry-v/?inquiry=1490

- Thanks for the suggestion
Bob

[ghansham]

hi All..

There is another issue that I found related to this topic.
Before loading it as a plan view, just select "User Selected" in Region Selection
and then set Bounding Box to occupy full area by drawing a box and then doing "Ctrl+R".
And then try to load it. You will find an exception coming from NetCDF java. It seems
like that netcdf java is not able to handle the lat/lon arrays backing this datasets properly.

I could not dig much due to shortage of time. But this seems to be a typical case that needs to
be handled with care. I think this issue needs to be addressed first.

regards
Ghansham

[tomw]

Back to original issue....

There appears to be a problem in the netCDF 4.3 library that subtracts 360 from the longitude values for this file. I have reported this to Unidata.

In the meantime,

In the meantime, if you want to use this file, you can employ this NcML to "fix" the issue (note you'll need to change the "location" attribute to point to your file on disk):

Code: Select all

<?xml version="1.0" encoding="UTF-8"?>
<netcdf xmlns="http://www.unidata.ucar.edu/namespaces/netcdf/ncml-2.2" location="D:/DATA_DRIVE/kris/NSS.GHRR.NF.D86275.S0857.E1038_NASA_LaRC_cloud_properties_3chan5.nc">
  <variable name="longitude" shape="lat lon" type="float">
    <remove type="attribute" name="coordinates" />
    <remove type="attribute" name="valid_range" />
    <attribute name="add_offset" type="float" value="360.0" />
  </variable>
</netcdf>

I also suggest that if you are creating these files you rename the "dimensions" to be "x" and "y", and remove the "coordinates" attribute from the "latitude" and "longitude" variables...just for clarity...

[tomw]

We have heard back from the netCDF developers -- this issue will be fixed in their next release.

[tomw]

Unidata released an update to the netCDF library which fixes this issue, and it is available in the McIDAS-V nightly builds.