NULL pixels after projecting Galileo NIMS data

ISIS ISIS Support
#1

I’m attempting to map-project some Galileo NIMS cubes that were obtained in the Long-Spectrometer Mode (specifically cubes 15e019 and 17e009). I’m using the data from Joshua Cahill’s PDART pre-release as suggested by Trent Hare. My process so far has been: (1) Create longitude and latitude cubes using cubeatt, then (2) Map-project using nocam2map with information from the PDS “browse” images.

It seems that regardless of which projection type I use (sinusoidal, cylindrical, etc.), about half of the data is projected as NULL; for example, the cube ‘15e015’ projects NULL for longitudes 107.1 W through 112.8 W. It’s clear from the “browse” image that these data do indeed exist within this longitude range. Also, in qview, the latitude/longitude information appears when the mouse hovers over the NULL pixels.

I’m curious if anyone has experienced something like this, or if anyone might happen to know what I’m doing wrong? I know the Galileo NIMS data has always been finicky with ISIS3, but since half of the data does show up when projected, I expect it’s an issue with how I’ve processed the data?

Thanks in advance!!

Map Galileo NIMS cube and mosaic it with Galileo SSI cube
#4

So, I pulled down the data and reproduced what you’re talking about. I tinkered around a bit and it seems that the resolution of the output projected image has a big impact on what parts of the scan are visible and which aren’t. By increasing the resolution by a factor of 10 I was able to get more of the data to show up in the output.

I think the narrow width of the scan is causing nocam2map to sometimes skip over the scan when it is attempting to interpolate. I would try again with 2x, 4x, 8x, and 16x resolution and see if you get better results. Something like pixel2map would handle this better, but that requires a full camera model.

1 Like
#5

This seems to work much better than anything else I’ve tried, thank you so much! It seems that the resolution increase factor maxes at ~50x, where the output cube that is generated reaches the maximum file size that nocam2map is able to create. Even at ~50x, there are some empty patches in the strip, though it’s difficult to tell if those patches are inherent to the data and don’t exist, or if they’re just unrecoverable. Either way, this is significantly better than where I started, so thank you again!

#2

That process sounds correct to me. Can you provide the exact command lines (or copy and paste them from the bottom of the gui) that you used for one of the images?

#3

Of course! I’ll use the 15e015 cube as the example, specifically the 15e015ti_spt.cub and 15e015ti_lot.cub files from the Cahill PDART pre-release linked above.

For the creation of the latitude/longitude files:
$ cubeatt from=15e015ti_lot.cub+1 to=latitude.cub
$ cubeatt from=15e015ti_lot.cub+2 to=longitude.cub

And for the map projection:
$ nocam2map from=15e015ti_spt.cub latcube=latitude.cub loncube=longitude.cub londir=positivewest pixres=ppd resolution=18.106 equradius=1562600.0 polradius=1559500.0

(where the resolution was computed from the mrad/pixel value provided in the browse image linked above, and the radii are for Europa from USGS).

#6

@jodiberdis Have you managed to use nocam2map on the full NIMS Europa dataset (320 images)? If so, from where did you take the resolution from? Did you extract them and read them in for the corresponding cube?

I am stuck at a very similar problem: half of the cubes throw an error without succesful mapping.

E.g.
15e008cr_core.cub

Group = Error
  Degree                  = 3
  NumberOfPoints          = 55
  SampleMinimumError      = -0.8386347112049
  SampleAverageError      = -7.42840132840105e-15
  SampleMaximumError      = 0.96056501148041
  SampleStdDeviationError = 0.45237015474128
  LineMinimumError        = -0.62690792566256
  LineAverageError        = -3.70774118478452e-14
  LineMaximumError        = 0.46682586101453
  LineStdDeviationError   = 0.28370503072671
End_Group

(what does this tell me?)

and to the command line, it throws something like: **PROGRAMMER ERROR** Invalid cube size specifications [ns=-2147483648,nl=-2147483648,nb=102]. And sometimes I get **ERROR** Unable to solve-least squares using QR Decomposition. The upper triangular R matrix is not full rank. What do these errors want to tell me?

The other half only throws also the Group = Error message, but then proceeds with mapping.

Any help is appreciated. Also a link to map-projected NIMS files.