• Ultraviolet Photography
  •  

Snow in the SWIR

Infrared SWIR
12 replies to this topic

#1 Andy Perrin

    Member

  • Members+G
  • 3,995 posts
  • Location: United States

Posted 17 December 2020 - 22:46

Consider this a first attempt, which I plan to update or replace tomorrow when it's light out again. I started a SWIR pano of the snowy landscape out my window but unfortunately the sun set before I could finish. Oh well. In addition to missing data, there are some obvious places where the lighting changed between captures. I will try again!

Equipment
TriWave Ge-CMOS sensor camera
Wollensak Velostigmat 25mm lens (but the camera has a 1/2" sensor, so there is a huge crop factor of around 5.4, so the effective focal length is about 135mm)
1500nm long pass filter from Thorlabs
Tripod

Software
TriWave driver software (seems to be custom coded for this particular camera -- I definitely think my camera must have been a prototype)
Custom MATLAB script to batch process the images and prepare them for the panorama, including subtracting off fixed pattern noise from the sensor
Photoshop CS6
SmartDeblur for deconvolving the image to sharpen it

Observations
- ice and water have different absorption curves and it's definitely possible to distinguish wet snow (black areas near windows and on the ground) from well-frozen snow (the gray on the roofs).
ETA: it seems ice absorbs more light than water at 1500nm, not less, so my updated hypothesis is that the contrast is due to less scattering between ice/water interfaces than between ice/air interfaces. See discussion below.

- brick has gorgeous patterns in SWIR

Attached Image: Snow SWIR UVP first try.jpg

Attached Image: Snow vis UVP.jpg

Edited by Andy Perrin, 18 December 2020 - 00:52.


#2 Stefano

    Member

  • Members(+)
  • 1,719 posts
  • Location: Italy

Posted 17 December 2020 - 22:52

Oh yes! I love this things! Very nice that you can see the difference between wet and "dry" snow. I expected it to simply be all-black, but apparently that's not the case.

Also do you have any idea why those bricks have those patterns?

Edit:
Now I noticed you posted a visible reference too. It seems to me wet bricks are darker in SWIR. That makes sense, as water absorbs a lot there. I don’t know if you are catching the 1450 nm water absorption peak with your filter (I think not), if you do water will look much darker, as in your 980 nm experiments.

Edited by Stefano, 17 December 2020 - 23:14.


#3 Andy Perrin

    Member

  • Members+G
  • 3,995 posts
  • Location: United States

Posted 17 December 2020 - 23:18

It’s the downslope of the 1450nm peak, Stefano. Most of the light is right near 1500nm because the camera gain goes to 0 at 1600nm. So it is a bit like photographing UV with a 400nm short pass on a silicon sensor: most of the photo will be 400nm or very close because it will drown out the shorter wavelengths. In this case it is the long end that gets drowned out.

Edited by Andy Perrin, 17 December 2020 - 23:19.


#4 Stefano

    Member

  • Members(+)
  • 1,719 posts
  • Location: Italy

Posted 18 December 2020 - 00:02

So it’s like using a 1500-1600 nm bandpass. It’s the same logic behind Bernard’s tri-color IR where he used a 1000 nm longpass filter that was limited by his camera sensitivity at ~1100 nm.

I think ice actually peaks closer to 1500, according to the first graph here https://en.m.wikiped...rption_by_water (I should find actual numerical data).

#5 Andy Perrin

    Member

  • Members+G
  • 3,995 posts
  • Location: United States

Posted 18 December 2020 - 00:34

I have the ice data in excel actually if you want it.

Here's the range we're discussing:
Attached Image: Water and ice spectrum 1400-1600nm UVP.png

It actually seems like the ice should be the darker one? So I think that calls into doubt my original interpretation that water absorbs more so it's darker.

It could be that the frozen snow scatters light better because the refractive index difference between ice/air is bigger than between ice/water.

Edited by Andy Perrin, 18 December 2020 - 00:52.


#6 Stefano

    Member

  • Members(+)
  • 1,719 posts
  • Location: Italy

Posted 18 December 2020 - 06:03

Water's attenuation length is about 0.3 mm at 1450 nm, and ice peaks at about 0.2 and something (it seems). That's very short, so you can't test this with ice cubes as you did at 980 nm. One millimeter of either water or ice should be almost black.

#7 Andy Perrin

    Member

  • Members+G
  • 3,995 posts
  • Location: United States

Posted 18 December 2020 - 06:43

Yeah, and that’s actually more evidence for the scattering hypothesis. They should both be deep black otherwise.

#8 Stefano

    Member

  • Members(+)
  • 1,719 posts
  • Location: Italy

Posted 18 December 2020 - 06:56

Is this like the same thing behind wet things being darker (even in visible light)? Water filling the gaps?

#9 Andy Perrin

    Member

  • Members+G
  • 3,995 posts
  • Location: United States

Posted 18 December 2020 - 14:23

Yes exactly.

#10 Bernard Foot

    Bernard Foot

  • Members+G
  • 706 posts
  • Location: UK

Posted 18 December 2020 - 14:53

Fascinating!

But what's going on with the light-coloured building at the right of the image - there's grey and black appearing on the boarding: is that trapped snow & ice?

PS - you ought to tell the owners of the black building that they need to get their guttering seen to.
Bernard Foot

#11 Stefano

    Member

  • Members(+)
  • 1,719 posts
  • Location: Italy

Posted 18 December 2020 - 15:02

Bernard, noticed now about the building on the right. The grey and black are almost surely water-related. Water absorbs SWIR so strongly that SWIR cameras can even detect moisture.

I too noticed the roof of the back building. As long as it doesn't snow several inches, it shouldn't be a big problem, but it's just a matter of time before it collapses.

I also noticed someone is using a tungsten lamp, as it is bright in SWIR, and there is a "thing" on top of a roof (probably an air conditioning system or similar) in the exact center of the image that drastically changes color in SWIR. Most notably, the two halves appear the same in the visible reference but are different in SWIR.

Edited by Stefano, 18 December 2020 - 15:08.


#12 Andy Perrin

    Member

  • Members+G
  • 3,995 posts
  • Location: United States

Posted 18 December 2020 - 16:34

Bernard - the black stains on the side of the house are definitely water. Not sure if that’s what you are looking at though.

I’m going to redo it this afternoon.

Edited by Andy Perrin, 18 December 2020 - 16:36.


#13 Andy Perrin

    Member

  • Members+G
  • 3,995 posts
  • Location: United States

Posted 18 December 2020 - 20:30

So far I have been unable to take a pano because the sun came out, which made the dynamic range bigger than the 10 bits my capture card is capable of. I REALLY miss the 4 extra bits on my A7S at times like this!

Guess I have to wait for some clouds.