NPP ATMS in Mc-V

How do I...?
User avatar
bobc
Posts: 988
Joined: Mon Nov 15, 2010 5:57 pm

Re: NPP ATMS in Mc-V

Post by bobc »

Thanks for the screenshots, Kris. I'm still having difficulty replicating the problem, but I've sent an email around to a couple programmers to get their thoughts. I'll get back to you as soon as I hear from them.

Thanks again -
Bob
User avatar
bobc
Posts: 988
Joined: Mon Nov 15, 2010 5:57 pm

Re: NPP ATMS in Mc-V

Post by bobc »

Hi Kris -

We are looking to get a little more information about your machine. Outside of McV, click the Apple button on the top toolbar and choose "About This Mac". This should pop up a window with 6 tabs. Please take a screenshot of the Overview tab and send it to us.

Thanks -
Bob
User avatar
kbedka1
Posts: 428
Joined: Wed Jan 28, 2009 7:27 pm

Re: NPP ATMS in Mc-V

Post by kbedka1 »

See attached. FYI, we were able to get a stripe free image from a windows machine using a nightly build from a couple weeks ago. For some reason the 89 GHz AMSRE data was coming up as all 0's though which was strange considering that channel was operational in 2006.
Attachments
Screen Shot 2017-07-07 at 10.40.28 AM.png
User avatar
kbedka1
Posts: 428
Joined: Wed Jan 28, 2009 7:27 pm

Re: NPP ATMS in Mc-V

Post by kbedka1 »

We are observing that the AMSR-E 89 GHz channel is providing all 0's for brightness temperature but other colleagues who have plotted 89 GHz data in IDL are getting valid values. Does Mc-V have issues with reading the 89 GHz channel data?
User avatar
bobc
Posts: 988
Joined: Mon Nov 15, 2010 5:57 pm

Re: NPP ATMS in Mc-V

Post by bobc »

Hi Kris -

With regards to the 89 GHz channel, I'm seeing similar results to you with "89.0 GHz-V-A Brightness Temperature" and "89.0 GHz-H-A Brightness Temperature" using your data file. Note that I see the same thing in McIDAS-X when displaying these bands with a local ADDE server. In McIDAS-X, the RAW value is -32768, TEMP is 0.00, and BRIT is 255 at all pixels.

On the NSIDC AMSR-E L2a page, if you click on User's Guide:

High_Res_A_Swath:
89 GHz observations from the A feedhorn AMSR-E scans; 486 observations per approximately 2000 scans. Note: Beginning 4 November 2004, the 89 GHz A-horn developed a permanent problem resulting in a loss of those observations. Consequently, after 3 November 2004, the High_Res_A_Swath data fields contain values of 0.

Looking around through the McIDAS-X inquiry system, I came across Inquiry 12503 where it is mentioned in the Testing section that bands 13 and 14 display as white, as the 89A channels of level 2A have RAW values of -32768. After a scale and offset is applied to this value, the brightness temperature of 0 is produced. I verified this in McX, where displaying band 13 or 14 from your data (the 89.0 GHz A channels) have these IMGPROBE values. We do have some older sample data from 2003, before the problem with the satellite, and these 89.0 GHz A channels can be displayed there without error in McIDAS-V and McIDAS-X.

Looking at your data in ToolsUI, there are a variety of low resolution 89.0 fields that you can ncdump and see values and an image. Perhaps your colleagues are working with these lower resolution 89.0 fields in IDL? These low resolution fields have a size of 1997x243 (10km x 10km) which matches the other bands/channels (as opposed to the high resolution A channels that are 1997x486 or 10km x 5km). An ncdump of the 89.0 A-horn fields in ToolsUI produces a solid black image where all of the values are the same at every pixel, matching the McIDAS-V/X output.

I wrote up adding the ability to access the low resolution 89.0 GHz fields as McIDAS-X Inquiry 16392. I also noticed that the B-horn 89.0 GHz channels have a different issue where McIDAS-V thinks the raw image size is 0,0. There's some IMGPROBE problems of these bands in McIDAS-X that I think may be the root of this problem. I wrote this up as Inquiry 16393.

As an FYI, I'm still trying to look into the striping problem you are seeing. One thing I noticed is that the graphics on the OS X machines I have tested with are NVIDIA and yours are Intel. I'll keep you updated on any findings.

Thanks -
Bob
User avatar
kbedka1
Posts: 428
Joined: Wed Jan 28, 2009 7:27 pm

Re: NPP ATMS in Mc-V

Post by kbedka1 »

Thank you Bob, the 89 GHz channel is quite useful for depicting regions of hail in a storm. Even the low res data will be useful for this application. The inquiry and background work you did is appreciated!
User avatar
bobc
Posts: 988
Joined: Mon Nov 15, 2010 5:57 pm

Re: NPP ATMS in Mc-V

Post by bobc »

No problem, Kris.

I did find a way that you can get access to these low-resolution fields in McV without an ADDE server, but there are a few steps:

  1. Load your AMSR-E HDF file through the General>Files/Directories chooser with the "Grid files (netCDF/GRIB/OPeNDAP/GEMPAK)" data type.
  2. In the Field Selector, you can expand the "Low_Res_SwathData_Fields" tree and find that there are 8 of them for the 89.0 GHz channel.
  3. Select the one you are interested in, and the Image Display display type.
  4. In the Times tab, note that there are well over 1,000 times listed. I'm thinking that each line in the swath has its own timestep and all of them are being listed. However, I found that selecting just one of the times displays the entire swath. Select one of the times and hit Create Display.
  5. Note that the display centers itself off the data, so you may want to change to the World projection and turn Auto-Set Projection off in the display panel so you keep the World projection if another layer is displayed.

At this point, if you probe the data (or look at the enhancement range in the Legend) you will notice that you are looking at the Raw values, not the calibrated temperature values you get through an ADDE server. However, from the AMSR-E document, it looks to be pretty easy to go from raw to temperature by just multiplying by a scale factor and adding an offset:

Tb ;(kelvin) = (stored data value * 0.01) + 327.68

To apply this formula to the data, I set up a function in the Jython Library:

Code: Select all

def amsreRawToT(data):
    amsreT = (data * 0.01) + 327.68
    return amsreT

To access this function, you can set up a formula. I set the formula up as follows:
Description: AMSR E Raw to Temperature
Name: amsreRawToTemp
Formula: amsreRawToT(data)

Now, to display the temperature data, do the following:
  1. Load your data through the General>Files/Directories chooser as described above.
  2. In the Field Selector, choose Formulas on the left side of the tab and select the "AMSR E Raw to Temperature" formula.
  3. Choose the Image Display display type and hit Create Display.
  4. In the Field Selector window, select the Low Res 89.0GHz channel.
  5. At the bottom of the Field Selector window, select just one time in the Times tab.
  6. Click OK

Now, your data will display as temperature values.

Note a few things:
  1. The display defaults to "Gray Scale" and the range matches the lowest/highest data values. To make the display match what you would expect to see with temperature data, change the enhancement to "Inverse Gray Scale" and the range to "180 to 320". Note that you can set up a parameter default for the formula (using the "amsreRawToTemp" Name field when setting up the formula) to default to this enhancement/range so you don't have to change it manually every time.
  2. If you realize you want to work with the same 89.0GHz field all the time, you can set this formula up as a derived formula so you don't manually have to select the field through the Field Selector window every time. If you are unsure of how to do this, let me know and I can give you instructions. You can find documentation on creating derived fields on the Derived Data page in the User's Guide.
  3. The navigation is slightly different when using the formula to display the data loaded through the local file chooser than it is when displaying through the local ADDE server. However, the navigation looks to actually be better when displaying data from the local file with the formula.

If you have any questions about this, let me know.

Thanks -
Bob
User avatar
kbedka1
Posts: 428
Joined: Wed Jan 28, 2009 7:27 pm

Re: NPP ATMS in Mc-V

Post by kbedka1 »

Bob, I am seeing 4 pairs (i.e.1 pair=a V and H polarization) of 89 GHz temperatures. Any idea which one we should be looking at and/or what the difference between them are?
User avatar
kbedka1
Posts: 428
Joined: Wed Jan 28, 2009 7:27 pm

Re: NPP ATMS in Mc-V

Post by kbedka1 »

I have verified that I am able to load the 89 GHz data through the process you detail. Thank you for the help. I am running into a problem though. I have to use both the vertical (V) and horizontal (H) polarization channels in a formula to get what's called a polarization corrected temperature (PCT). I initially select my polarization corrected temperature formula, which then prompts me to select V and H data fields. For both V and H I select AMSR E Raw to Temperature, because I need to convert to temperature to make the formula work. But then I am only prompted once to select Data, i.e. I only have the opportunity to select data for the V channel, but never get asked again for the H channel. So I am unable to get what I need done. Mc-V should be prompting me to select Data twice, for the each of the calls to AMSR E Raw to Temperature. You can test this yourself, make up a formula called polarization corrected temperature: PCT = 1.82 * V - 0.82 * H
User avatar
bobc
Posts: 988
Joined: Mon Nov 15, 2010 5:57 pm

Re: NPP ATMS in Mc-V

Post by bobc »

Hi Kris -

It sounds like you are doing things correctly, and the process is working for me on Windows and OS X using recent nightlies. To be sure we are following the same procedure, here's some steps outlining what I did:

  1. I added the new PCT formula you mentioned. Now, there are two AMSR-E formulas that I have under the Help Desk group in the Field Selector. It doesn't matter what group you have them in, I just have them here to keep things organized. I selected the new PCT formula, the Image Display display type, and clicked Create Display:
    Field Selector formula listing
    Field Selector formula listing
  2. Next, the Field Selector window appears where I define what V and H are. In both cases, they are the AMSRE Raw to Temperature formula. I selected this for V and H and hit OK:
    First Field Selector window
    First Field Selector window
  3. In the next Field Selector, I selected the V field to pass through the amsreRawToT formula, just the first timestep, and hit OK:
    Second Field Selector window
    Second Field Selector window
  4. In the next Field Selector, I selected the H field to pass through the amsreRawToT formula, just the first timestep, and hit OK:
    Third Field Selector window
    Third Field Selector window

At this point, I was able to get the data to display without error. Does this sound like the process you are following?

As an FYI, you can make things a little easier on yourself by combing your polarization corrected temperature formula with the amsreRawToT function. In my Jython Library, I have the following:

Code: Select all

def amsreRawToT(data):
    amsreT = (data * 0.01) + 327.68
    return amsreT

def amsrePCT(V, H):
    V_temp = amsreRawToT(V)
    H_temp = amsreRawToT(H)
    PCT = 1.82 * V_temp - 0.82 * H_temp
    return PCT

I set up a formula to use this amsrePCT function as follows:

Description: Polarization corrected temperature expanded
Name: PCTE
Formula: amsrePCT(V,H)

When I evaluate this formula, there is just one Field Selector window prompting for the V and H fields. This is returning the same results as the process listed above, but with a couple fewer steps. Give this a try and let me know if you run into any problems.

Thanks -
Bob
Post Reply