| Author | Message | ||
     jimboy bactong (Jimboy11) New member Username: Jimboy11 Post Number: 7 Registered: 3-2006 |
whew! thanks a lot, i feel a ton better now. your right, i've been reading a lot of journals and books for a couple of months and for all that time i'm reading heiroglyphics but discussing it here make a lot clearer i just want to clarify something... you think that spectra from 400-700 is good? the known absorption maxima of chlorophyll are 430, 450, 640 and 660, the absorbance from 500-600 as we expect should be low since its green or chlorophyll supposedly transmits that wavelengths, but as you can see the rise of my spectra is just different we already discarded that region in our discussion but since you mention it to be good, makes me think that maybe i'm not doing the right thing. and by the way the samples are already gone thanks again!! | ||
| David W. Hopkins (Dhopkins) New member Username: Dhopkins Post Number: 76 Registered: 10-2002 |
Hi Jimboy, Okay, the good news is that your spectra appear to be good from 400 to 700 and from about 1160 to 1400. Unfortunately, it appears that only about 8 of the scans appear good from 700 to 1160, the count is difficult because it is not clear what curves may be superimposed, I can only count about 16 scans, depending on what region I examine. It appears that the reference curves were not adjusted to fall within the range of the ADC, and some of the samples read too low in that region. You have twice as many points in the visible region as in the NIR, so the break at 700 looks weird, but we can see the 670 nm band of Chl in many of the scans. It looks like you have some noise at the higher wavelengths, and I think you might be best not to trust the data beyond 1360 nm. You may be able to do some useful regressions using PCR and PLS over the 1160-1360 nm range. You should read the help files on the methods in your software, or read the useful little book by Richard Kramer, Chemometric Techniques for Quantitative Analysis. Marcel Dekker, Inc, New York, 1998. ISBN 0-8247-0198-4. Maybe as "reference values" you could get chl amounts using a 2-wavelength baseline correction for the readings at 670 nm, say 2*A670 - A650 - A690. It would be better to test this spectral method vs. an extraction method, if you think you can still do that. You have not answered whether the samples are still available? I think you are learning a lot of spectroscopy from your work, even if you find that the spectra were not perfect. An important lesson is to look at the spectra and ask if they are reasonable, as you have done. Best wishes, Dave | ||
| (Unregistered Guest) Unregistered guest |
the instrument we are using here uses a 500 watts halogen lamp as light source, a lense to focus the light, an oriel 77250 as monochromator and the gratings are 77299 for the nir or from 700-1400 nm then 77298 for the visible. the sensor is a phototransistor which is connected to an ADC then interfaced to a computer. the blank or the reference we use is a two piece glass slide, which is measured separately. when a sample is measured it is inserted between the two glass slide... the spectra is measured in transmittance which we do by placing the sensor behind the sample thats how we do things here, hope there is something in my spectra | ||
| David W. Hopkins (Dhopkins) New member Username: Dhopkins Post Number: 75 Registered: 10-2002 |
Hi Jimboy, I am sorry to say, your spectra look very unusual to me. We need to back up several steps to determine what you have, and whether anything can be salvaged. What instrument did you use? Were you measuring the leaves in transmission or reflection? What sample did you use to record the T0 reference scan? Did you record the T0 scans separately, so you can view them? Did you adjust the gain so that the T0 scans fell within the operating range of the amplifier? I think there were problems at this stage, because the shapes of the sample scans between approximately 690 and 1170 nm are not well behaved. You are far away from worrying about the details of PLS regression or PCR Principal Component Regression. Both methods are available in Unscrambler and many other software packages. We need more details to see whether we can help you. Best wishes, Dave | ||
| jimboy bactong (Jimboy11) New member Username: Jimboy11 Post Number: 6 Registered: 3-2006 |
hi everyone hi sir dave, you have mention about this PCR how is it done? or what is its requirement? is it in the unscrambler 9.5? for i have obtain a trial copy of it? about my spectra, i have aploaded it as an attachment, the legend in left stands for the rice variety, and one more thing that bothers me, the peaks in my spectra does not match to those known chlorophyll peaks? is it wrong?  | ||
| David W. Hopkins (Dhopkins) New member Username: Dhopkins Post Number: 74 Registered: 10-2002 |
Hi Jimboy, It is not clear how much you are expected to do for this project. To do PLS, you need to have independent lab values for the constituent(s) you wish to measure. You probably should do a search to determine the best way to measure chlorophyl and decide whether you have the capability to do it. This assumes that you still have the samples and that they have not changed during the time between your NIR-vis scans and the extraction. Of course, you can calculate the chlorophyl content on a dry-weight basis to get around a problem of drying out in the period. I suppose that you could use the A670 values as chlorophyl determinations, and go through the motions of PLS and PCR calibrations. Hard to say about your spectra. Why don't you upload some scans of a few representative samples, so we can advise you. Do a search on the threads for instructions, which I think are also on the left side bar of the discussion area. Best regards, Dave | ||
| jimboy bactong (Jimboy11) New member Username: Jimboy11 Post Number: 5 Registered: 3-2006 |
so, i really have to do this chlorophyll extraction? or not? sori guys if my questions may sound a bit stupid or shallow but i'm here in philippines, nir is still an alien here even pls, no buks about it yet and no professors to ask about, i'm on my own... and one more thing, absorbance in nir is still solved using log10(1/t) right? as t is transmittance and transmitance is ratio of voltage of sample over voltage of blank... right? also i did try to compare my absorbance spectra to what a read, its just different, it has no peaks but it has plateaus a very broad one, is something wrong with it? | ||
| David W. Hopkins (Dhopkins) New member Username: Dhopkins Post Number: 73 Registered: 10-2002 |
Hi Jimboy, Since you have recorded the vis-nir spectra on the samples, you may already have good chlorophyl information. The extraction of the leaves using a solvent such as acetone would give you a good measure of the chlorophyl. It would be interesting to compare the extraction results to the peak height at 670 nm, which is approximately where the chlorophyl a should absorb maximally, if I remember correctly. Best wishes, Dave | ||
| hlmark (Unregistered Guest) Unregistered guest |
I should have continued: You collected 20 leaves. That's generally considered "skimpy" these days, for NIR calibration. However, they might be satisfactory, if they contain suitable variations in their composition, as I was describing. You would also do well to reread Dave Hopkins' advice, and collect another, say, 10 leaves to use for validating the model you generate. \o/ /_\ | ||
| hlmark (Unregistered Guest) Unregistered guest |
Jimboy - well that, in a nutshell, is the second step, anyway. The FIRST step is to collect a suitable set of samples on which to perform the calibration. This sample set should consist of leaves with varying amounts of chlorophyll, hopefully covering the range of chlorophyll content that might exist in those leaves, or as near to the full range as you can manage. Ideally you should have more than a minimum amount of leaves, so that hopefully they will also vary in whatever other components might vary in the leaves. You should also try for a variety of leaves that might exhibit different physical characteristics: thickness, number of veins, whatever there is. Howard \o/ /_\ | ||
| jimboy bactong (Jimboy11) New member Username: Jimboy11 Post Number: 4 Registered: 3-2006 |
hi, its me again... guess having a qualitative discussion is not enough, anyways i just want to ask question about this pls thing, to make sure i'm doing right since my sample is an intact leaf and let say i want to have a n0n-destructive way of chlorophyll estimation... to do this, i have to subject my samples to nir spectroscopy and to have this so called calibration sets i have to still extract the chlorophyll in it then measure it the traditional way... right? | ||
| David W. Hopkins (Dhopkins) New member Username: Dhopkins Post Number: 72 Registered: 10-2002 |
Jimboy, Another great reference that specifically addresses the spectroscopy of agricultural products and shows shere protein, moisture, oil and fiber have absorption bands is: Phil Williams and Karl Norris, eds., Near-infrared Technology in the Agricultural and Food Industries. Second Edition, American Association of Cereal Chemists, Inc, St. Paul, Minnesota, 2001. ISBN 1-891127-24-1. Hopefully that is available through your library, at least via inter-library loan. Best regards, Dave | ||
| Howard Mark (Hlmark) New member Username: Hlmark Post Number: 10 Registered: 9-2001 |
Jimboy - FOSS (one of the major NIR instrument manufacturers) used to provide a very nice one. I don't know if they still do. I know someone there I can forward your question to, and see if they can provide you with one. \o/ /_\ | ||
| jimboy bactong (Jimboy11) New member Username: Jimboy11 Post Number: 3 Registered: 3-2006 |
i would love to have that pdf file, i'm pretty 90% sure that book is still not in our library, my e-mail is [email protected] again thanks, you guys are great | ||
| jimboy bactong (Jimboy11) New member Username: Jimboy11 Post Number: 2 Registered: 3-2006 |
that was fast, thanks, anyways that colthup chart seems really interesting but where can i get those? is it available online? do you have a copy of it? | ||
| David W. Hopkins (Dhopkins) New member Username: Dhopkins Post Number: 71 Registered: 10-2002 |
Hi Jimboy, Yes, there is an excellent chapter in a set of books that may be in your library that I can recommend highly. See the chapter by Lois G. Weyer & S.-C. Lo, Spectra-Structure Correlations in the Near-infrared, Vol 3 pp 1817-1837. It is in John M. Chalmers and Peter R. Griffiths, eds., Handbook of Vibrational Spectroscopy. 5-Volume Set, John Wiley & Sons, Ltd, Chichester, U.K., 2002. ISBN 0471988472. It has band assignments and good discussion for the NIR region. If you send me your email address, I will send you Lois Weyer's address, and I am sure she would be glad to send you a pdf copy of the chapter for you to study. However, for the purposes of your project, I think that only 20 leaves is at most half of a project. I recommend that you obtain the scans of 20 new leaves to test your calibrations. That is the critical part, as you may find that your best calibration will not perform as well as another! I would recommend that you try a Principal Components Regression (PCR) as well as PLS. With so few calibrations, you may find PCR better than PLS on your test set. Your first lesson should be, calibration is only the start of the process, and you need to test your calibrations on a separate test set of data. Best regards, Dave Hopkins | ||
| Howard Mark (Hlmark) New member Username: Hlmark Post Number: 9 Registered: 9-2001 |
Hi Jimboy! Welcome to the discussion group. I can't help you in detail, except to point you to where you need to look. There are compilations of spectral band assignments available. There are also graphic illustrations of which underlying molecular vibrations cause absorptions at various frequencies/wavelengths, called Colthup charts. In the NIR, the absorptions are moderately specific to the nature of the functional group, but not so specific as to be able to tell much more about the overall molecule. Absorptions in the visible part of the spectrum tend to be less specific, although you can pretty well guess that absorption in a leaf will be due to chlorophyll - - and you don't even need to look at the spectrum to know that! But absorptions in the visible region are, in general, due to the overall structure of the molecular electronic transitions, and therefore tell you less about the details of the molecule than the vibrational bands in the NIR do. Howard \o/ /_\ | ||
| jimboy bactong (Jimboy11) New member Username: Jimboy11 Post Number: 1 Registered: 3-2006 |
hi!, have done a vis-nir absorbance spectroscopy of 20 rice leaves for my undergraduate thesis. Besides PLS, is there any other way to discuss the absorbance spectra? is there a way to know what are those peaks and "valleying" stands? i hope you can really help me on this? thanks |