Public Lab Research note


Desktop Spectrometry Kit (V3) for VNIR Field Spectroscopy

by lukebrown | July 11, 2015 11:29 | 78 views | 8 comments | #12047 | 78 views | 8 comments | #12047 11 Jul 11:29

lukebrown was awarded the Basic Barnstar by warren for their work in this research note.


Introduction

I was interested in whether the Desktop Spectrometry Kit (V3) could be successfully used for VNIR field spectroscopy, in the context of vegetation analysis. I conducted a quick experiment outside using a bamboo leaf and a sheet of white paper to act as a reference target. It should be noted the IR filter was removed from the webcam before this experiment was carried out.

Suitability of the card enclosure

Having taken the spectrometer outside, I suspected there may be substantial transmittance through the enclosure, as the spectra displayed on screen varied considerably depending on how I held the unit. I fashioned an external sleeve out of thicker cardboard and recorded two spectra, holding the spectrometer to my body to prevent light entering through the slit. The degree of recorded background noise was greatly reduced using the sleeve (Figure 1).

noise.png

Figure 1: Background noise recorded with and without the use of an external cardboard sleeve.

Spectral 'reflectance' of a bamboo leaf

At each wavelength, the ratio between the DN associated with the bamboo leaf and the sheet of white paper was calculated. It is important to note the resulting values are not true reflectance factors because no radiometric calibration was carried out prior to this calculation. Additionally, the sheet of paper is by no means a well characterised reference target!

Nevertheless, some key features can be easily identified, including the chlorophyll absorption features and the red edge (Figure 2). The magnitude of values is greater than would be expected, particularly at visible wavelengths. This perhaps highlights the need for radiometric calibration. The spectral reflectance of a bamboo leaf obtained using an ASD FieldSpec 3 is provided for some context.

bamboo.png

Figure 2: Reflectance spectra for a bamboo leaf obtained using the Desktop Spectrometry Kit (V3) and an ASD FieldSpec 3.

Next steps

To reduce background noise, I believe the design of the enclosure needs to be revised. The results presented here suggest that thicker cardboard may be advantageous.

To be of real use for VNIR field spectroscopy, a means of radiometric calibration is required, enabling measurements of radiance to be obtained. True reflectance factors could then be calculated, enabling recorded spectra to be directly compared to measurements from other instruments.


8 Comments

HI, Luke! Great post. What paper did you use for the sleeve? We chose the black paper as it's absorbent in the near infrared range, but do you think the extra light is entering through cracks and joints in the enclosure, or through the paper itself? How bright was your light such that it got into the case? How bright must light be before it begins entering the chamber?

We'd love to see a picture of the sleeve -- its design may answer some of these questions, along with an exact description of the type of material.

It's great to see the ASD FieldSpec data too -- is it so smooth because of a longer exposure, with time-averaging to reduce noise? Or is it otherwise smoothed? Tests on an Ocean Optics device seemed similarly noisy to our device: http://publiclab.org/notes/warren/09-23-2014/oil-fluorescence-scanning-on-an-ocean-optics-spectrometer so I'm curious why it's so smooth in yours.

If the ASD is calibrated, it'd be really terrific to get a baseline sensitivity graph for our webcams with a full-spectrum light. They may not be totally exposure linear (though @stoft has more data on this) but just knowing how sensitive our type of webcam is to each wavelength would be a good thing to know, and we could in theory build in a correction to Spectral Workbench. I wonder how much it varies camera to camera. @stoft - do you have thoughts on this -- maybe I'm overlooking something?

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


@warren Thanks! The sleeve was just a piece of corrugated cardboard, about 3 mm thick, folded around the enclosure and secured with some tape. I’m afraid I can’t provide a photo right now as I’m away from home.

I believe light is entering through the paper itself. It’s quite thin, particularly on the lid where there’s only one layer, so there is likely to be some transmittance. When looking at the webcam image itself I was sometimes able to discern the shadow of my fingers!

The experiments were carried out under bright sunlight, so this may not be so much of an issue if other sources of light are used, but it definitely is for field spectroscopy. Having said this, even under indoor lighting I did notice the black areas of the webcam image appeared darker when the sleeve was used. This probably needs some further experimentation.

In terms of the ASD data, they are the average of three samples - each time you take a measurement, the average of a specified number of samples is produced to reduce noise as you mention. You may be correct about a longer integration time too - this is automatically determined when the device is optimised for the current illumination conditions. Dark current measurements are also obtained during optimisation using an electro-mechanical shutter. They are automatically subtracted from subsequent measurements. I'm not familiar with the Ocean Optics device, but if dark current subtraction wasn’t undertaken this might provide some further noise reduction?

The ASD is calibrated, although getting access to it would be subject to approval - I’d need to have chat with some people in the department. The ASD data presented here were actually collected as part of some work for my master's degree.

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


We would love some input on selecting a better material for the spec v3 -- it's complex since not every black cardstock is actually black in the near-infrared range. Would you have any interest in testing out some different materials that we could still die-cut and print on?

One additional suggestion has been to use a water-resistant paper or thin plastic. I'm wondering if we could make a water-resistant "sleeve" to slide the spectrometer into. Would it be enough to have a sleeve with open ends? Or would it have to have an end flap as well? Could we improve both water resistance and opacity at once?

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


Hmm, I just did a test in (admittedly cloudy) sunlight vs. closing the blinds, and covering the device with a thick blanket, and didn't see a big difference. It doesn't seem likely to me that light is leaking in through this big dark folded-over blanket. How does this compare to your noise?

This has had 10 pixel rows of data smoothed to try to reduce noise:

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


Hmm, the embed isn't working in comments. I'm going to post a quick note:

http://publiclab.org/notes/warren/07-13-2015/testing-if-light-leaks-through-the-walls-of-the-pl-spectrometer-v3

Reply to this comment...


@warren Have posted a response in your note with some webcam images to try and illustrate what is going on at my end. In terms of sleeve design, mine is in fact open ended (please excuse how rudimentary it looks):

IMG_20150714_171830294.jpg

IMG_20150714_171839542.jpg

I think a water-resistant sleeve is a great idea, the critical thing in my opinion is that the material used is thick enough to prevent any substantial transmittance.

Reply to this comment...


Great, thanks. I think we could relatively easily find a heavier material to make this work. Would you be able to post the picture of your sleeve plus your suggestion here, where we collect input for the next revision? https://github.com/publiclab/spectrometer3/issues

Thanks for your excellent empirical work and suggestion.

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


Hi, we're compiling input here: https://github.com/publiclab/spectrometer3/issues/25 for an October redesign. Thanks!

Reply to this comment...


Login to comment.

Public Lab is open for anyone and will always be free. By signing up you'll join a diverse group of community researchers and tap into a lot of grassroots expertise.

Sign up