Hi Greg, This note is not the definitive answer you are looking for. Please read on. re: > Ok - I understand the details of how to do the channel differencing > with IMGOPER. OK. > It's when I get to the color bar that I am confused. My original inclination was that this should be straightforward. After looking at the problem in detail last night, however, I find that the situation is quite a bit more complicated. The main reason for the complication, at least for GOES data, is the two-slope calibration that the images follow. If we are talking about GOES images, the the calibration of brightness to temperature for infrared images is represented by the values in the stretch table IRTEMP.ST (I created the stretch IRTEMP.ST using the standard GOES calibration): SU TABLE IRTEMP BREAKPOINTS STORED IN TABLE : IRTEMP.ST INPUT OUTPUT ----- ------ 163 255 241.9999 177 242 176 330 0 CALIBRATION TYPE : GVAR CALIBRATION UNITS : TEMP BAND NUMBER : -1 INTERPOLATION TYPE: LIN SU: DONE The two-slope calibration indicates that the the same difference in brightness will represent different temperature differences depending on the original brightness values for the IR (10.7 um) and IRS (IR short, 3.9 um) images. This difference is well illustrated by the scatter plot contained in the attachment I have included. The axes in the scatter plot are: X -> 63 + (irBrit - irsBrit) Y -> 63 + (irTemp - irsTemp) Notice how the scatter plot illustrates the two slope calibration effect. The two-slope relationship is likely the reason that IMGOPER did not provide a correct PRD calibration in the output image created by IMGOPER when specifying UNIT=TEMP. I think that it should have, but I will contact SSEC for comment. Nonetheless, I _think_ that one should be able to create a stretch table that captures the essence of the two-slope calibration. This stretch table would then be used in a BAR labeling of the product generated with IMGOPER as: BAR SU=IRDELT where IRDELT.ST is the as yet non-existent stretch table. > So > I've done a channel differencing using UNIT=TEMP. If I do an IMGPROBE on > the results, I see numbers like: RAW=160 BRIT=160 PROD=160 etc. The > values vary but RAW=BRIT=PROD is always true. I am not surprised that the RAW=BRIT since the image will be calibrated. I am surprised by PROD also having the same value AND there is not a TEMP value. This is something I will need to discuss with SSEC. > If I type BAR to put a color bar on the image, I get a 0-255 > grayscale that has the word PROD above the scale but BRIT above the > numbers. This agrees with PROD=BRIT. The question is why the PRD value doesn't reflect the fact that the values are a difference?! > Ideally I'd like the units to be TEMP and the range to scale > from -5 to +10. My conjecture is that one should be able to create a stretch table that represents the temperature difference values in spite of the two-slope calibrations. I am playing with this now, and will send you my results if/when I am satisfied with them. > When I ran the IMGOPER command I used the same ACON and > MCON as suggested in the fog product demo (150 and 10). OK. > Is it possible to get the colorbar to read the way I want it to? I believe so. > If yes, how do I do that? I'm working on it... Cheers, Tom **************************************************************************** Unidata User Support UCAR Unidata Program (303) 497-8642 P.O. Box 3000 address@hidden Boulder, CO 80307 ---------------------------------------------------------------------------- Unidata HomePage http://www.unidata.ucar.edu **************************************************************************** Ticket Details =================== Ticket ID: OKG-462913 Department: Support McIDAS Priority: Normal Status: Open
Attachment:
irTirsT_irBifsB.gif
Description: GIF image
