1. Istituto Nazionale di Geofisica e Vulcanologia Pisa, Italy
Introduction to open-path FTIR measurements of volcanic gases (and aerosols…)
Dr. Mike Burton
Istituto Nazionale di Geofisica e Vulcanologia
Pisa, Italy
Mike Burton – AOPP– 26 July 2013

2. Mike Burton – AOPP– 26 July 2013
Talk Outline
Infrared radiative transfer
FTIR technology and signal processing
Open-path FTIR applied to volcanology
Future directions for OP-FTIR in volcanology
Mike Burton – AOPP– 26 July 2013

3. Wien’s Displacement Law
The peak wavenumber of the Planck curve increases with increasing temperature, following ~1.96 x T(K)
Mike Burton – AOPP– 26 July 2013

4. Gas with temperature T absorbs 20% of radiation
If a gas is in thermal equilibrium then the amount of radiation emitted must be equal to the amount of radiation absorbed (ignoring convection and conduction).
Gas with temperature T absorbs 20% of radiation
Ambient Temperature is T, assuming surrounding is blackbody the gas is exposed to radiation with intensity = B(T)
Radiation with intensity B(T)*0.8 is transmitted and B(T)*0.2 is emitted, so B(T) is observed
Gas with temperature T emits B(T)*0.20
Gas is ‘invisible’: thermal contrast with background is needed to ‘see’ gas
Mike Burton – AOPP– 26 July 2013

5. Transmittance is I/I0 = exp (-ecl) = t
Beer-Lambert Law
Transmittance is I/I0 = exp (-ecl) = t
Therefore observed intensity is I0.exp(-ecl) = I0.t
Adding more gases produces a multiplicative effect, e.g.
I = I0.tgas1.tgas2.tgas3.tgas4…
Mike Burton – AOPP– 26 July 2013

6. Gas layers both emit and absorb radiation
I(n)
Atm. Layer with temperature T1 and transmittance t1(n)
I(n) . t1 (n) B(T1) . (1-t1)
Atm. Layer with temperature T2 and transmittance t2(n)
(I(n) . t1 (n) + B(T1) . (1-t1)) . t B(T2) . (1-t2)
Mike Burton – AOPP– 26 July 2013

7. Michelson Interferometer
Requires collimated light
Half the source radiation returns to the source
Throughput is high
All wavelengths measured simultaneously
Mike Burton – AOPP– 26 July 2013

8. Input and output from the interferometer

9. Mike Burton – AOPP– 26 July 2013
Broadband source
Mike Burton – AOPP– 26 July 2013

10. Finite mirror movement distance
Mike Burton – AOPP– 26 July 2013

11. Multiplication in mirror displacement / Fourier space is a convolution in frequency space

12. Phase errors and correction
Mike Burton – AOPP– 26 July 2013

13. Mike Burton – AOPP– 26 July 2013
Field of view
As well as OPD, the field of view of the instrument can degrade spectral resolution and produce wavelength shifts.
This is because off-axis rays travelling through the spectrometer travel further than the on-axis rays.
Mike Burton – AOPP– 26 July 2013

14. Field of view, resolution and instrument lineshape
Spectral resolution is controlled by the distance of mirror movement within the FTIR
Normal estimates for the spectral resolution are given by
Resolution = 0.9 / OPD
Where OPD is the optical path difference produced by moving one of the mirrors in the FTIR. In the OPAG-22 the maximum OPD is 1.78 cm, producing a resolution of 0.9/1.78 ~0.5 cm-1
FTIR spectrometers may have OPD’s of several meters, for the highest resolution spectrometers. Typical absorption line-widths at atmospheric pressure are 0.1cm-1, so an FTIR with OPD of 9cm would be optimal.
Higher resolution = larger spectrometer = greater weight and less portability

15. Applying OP-FTIR to volcanic gas
First demonstrated by Japanese team on Unzen (Mori et al., 1993)
1996 Francis and Oppenheimer measured SiF4 on Vulcano (Francis et al., 1996)
1998 Francis, Oppenheimer, Burton used sunlight on Etna and active on Masaya (Francis et al., 1998 Nature, Burton et al., 2000, Geology)
1998 Love & Goff measure SO2 and CO2 in emission at Popo (Nature, 1998)
From 2000 regular measurements on Etna (Allard et al., 2005, Nature)
several measurements on Stromboli (Burton et al., 2007, Science)
regular measurements at Kilauea (Edmonds and Gerlach, 2008)
2010 Review of more than 10 years results from Unzen, Usu, Aso, and Satsuma-Iwojima (Notsu and Mori, 2010)
2012 Cerberus on Stromboli (La Spina et al., JVGR, 2012)
2013 FTIR on LUSI mud volcano (submitted)
Mike Burton – AOPP– 26 July 2013

16. Hardware Midac Bruker MCT InSb Cerberus FLIR Photon 320 Scanner
Midac M4401-S
Acid-resistant sealing
Mike Burton – AOPP– 26 July 2013

17. Measurement Geometries
FTIR
IR Source
Mike Burton – AOPP– 26 July 2013

18. Mike Burton – AOPP– 26 July 2013

19. Mike Burton – AOPP– 26 July 2013

20. Mike Burton – AOPP– 26 July 2013

21. Examples of OP-FTIR spectra

22. Examples of OP-FTIR spectra

23. Examples of OP-FTIR spectra

24. Examples of OP-FTIR spectra

25. Examples of OP-FTIR spectra

26. Examples of OP-FTIR spectra
Emission spectra: if the source of radiation is cooler than the gas itself we observe emission spectra

27. This is achieved by producing a best-fit simulated spectrum, such that
Data Analysis
The basic objective of any retrieval scheme is to retrieve quantitative information from the measured spectra.
This is achieved by producing a best-fit simulated spectrum, such that
y = F(x)
Where y is the measured spectrum, F is a model which simulates the measured spectrum and x is a state vector containing the model parameters, such as gas amounts.

28. Field of view, resolution and instrument lineshape
The measured spectrum is affected by the instrument, primarily by smearing of the spectrum due to its finite spectral resolution.

29. Potential problems
Many measurements are made with a lower resolution spectrometer than the linewidth of the target gases. In this case care must be taken when fitting the spectrum to take account of the non-associative nature of the ILS convolution.
F = exp(-ec1l)  ILS

30. ln(exp(-ec1l)  ILS) x 3 ≠ ln(exp(-ec2l)  ILS)
Potential problems
When performing the fit it is important to re-convolve the theoretical spectrum with the ILS at each iteration, because…
c2 =3 x c1
ln(exp(-ec1l)  ILS) x 3 ≠ ln(exp(-ec2l)  ILS)
i.e. you cannot use a reference spectrum measured at one gas concentration to fit measured spectra with markedly different gas concentration.
Using weak absorption lines is generally a good idea, as they are less affected by these problems.

31. Potential problems

32. Potential problems

33. Single beam or ratio retrievals
Single beam retrieval: simulating the original measured spectrum
Ratio retrieval: prior to analysis divide the measured spectrum by a reference spectrum, ideally identical to the measured spectrum but without the target gases.
In OP-FTIR reference (or background) spectra can be hard to come by; conditions can change rapidly.
Single-beam retrievals are preferable.

34. Simple retrieval: Masaya Volcano active source.
FTIR
520 m
IR Source

35. Simple retrieval: Masaya Volcano active source.
Volcanic gas temperature ~ atmospheric temperature
High volcanic gas amount
Stable source, hotter than gas
Simple atmospheric model, single layer, no temperature retrieval

36. Complex retrievals: Etna lava fountain

37. Complex retrieval: Etna lava fountain
Volcanic gas temperature is subject to very strong gradients
Highly variable volcanic gas amounts (difficult to always use weak absorption lines)
Highly unstable source (need to eliminate spectra in emission, and take account of the changing field of view)
Complex retrieval:
First determine volcanic gas temperature from SO2 rotational absorption structure
Use two layer model, one for the atmosphere one for volcanic gases, with simultaneous fit of field of view parameter.

38. Complex retrieval: Etna lava fountain
Transmittance a b

39. Trade-offs: Spectral resolution vs weight and SNR
Higher spectral resolution:
Pro – Better quality fits, less sensitive to non-linearity problems
Con – Larger spectrometer, with more weight and more delicate
Con – longer data acquisition time for each spectrum and lower snr
Up until now typical portable open-path FTIR’s have been 0.5cm-1 resolution and weighing ~15 kg: can be improved

40. Trade-offs: SNR vs measurement frequency
Higher signal to noise ratio: achieved through longer integration times
Pro – Better quality spectra
Con – Lower spectrum acquisition frequency
Con – potential for large variations in absorbing gas amount, leading to non-linearity problems
Higher measurement frequency:
Pro: resolve short-term variations
Pro: can always average spectra to increase SNR after data collection
Con: inferior snr on individual spectra

41. Future Directions for OP-FTIR
Emission spectroscopy
Examine limits for lower resolution gas measurements, in order to increase measurement frequency or SNR
Mud volcanism
Aerosol and ash quantification
Satellite validation (IASI)
Gas solubility model validation