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

Mike Burton – AOPP– 26 July 2013

Introduction to open-path FTIR measurements of volcanic gases (and aerosols…)

Dr. Mike BurtonIstituto Nazionale di Geofisica e Vulcanologia

Pisa, Italy

Talk Outline

• Infrared radiative transfer• FTIR technology and signal processing• Open-path FTIR applied to volcanology• Future directions for OP-FTIR in volcanology


Wien’s Displacement Law

The peak wavenumber of the Planck curve increases with increasing temperature, following ~1.96 x T(K)

0

1

2

3

4

5

6

7

8

9

10

0 1000 2000 3000 4000 5000 6000 7000 8000 9000

Wavenumber (cm-1)

Inte

ns

ity

400°K

600°K

800°K

1000°K

1200°K

Peak 1200 K at ~2400 cm-1

Peak 1000 K at ~2000 cm-1

Peak 800 K at ~1600 cm-1

Peak 600 K at ~1200 cm-1

Peak 400 K at ~800 cm-1


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).

Ambient Temperature is T, assuming surrounding is blackbody the gas is exposed to radiation with intensity = B(T)

Gas with temperature T absorbs 20% of radiation

Radiation with intensity B(T)*0.8 is transmitted and B(T)*0.2 is emitted, so B(T)

is observed

Gas is ‘invisible’: thermal contrast with background is needed to ‘see’ gas


Gas with temperature T emits B(T)*0.20

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…


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)) . t2 + (B T2) . (1-t2)

Michelson Interferometer


Requires collimated lightHalf the source radiation returns to the sourceThroughput is highAll wavelengths measured simultaneously

Input and output from the interferometer

Broadband source


Finite mirror movement distance


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

Phase errors and correction


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.

Field of view


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

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)

• 2001-2003 several measurements on Stromboli (Burton et al., 2007, Science)

• 2005-2007 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)


Hardware

• Midac

• Bruker

• MCT

• InSb

• Cerberus


FLIR Photon 320

Scanner

Acid-resistant sealing

Midac M4401-S

Measurement Geometries


FTIR

IR Source




Examples of OP-FTIR spectra

Lava Source

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0 1000 2000 3000 4000 5000 6000

Wavenumber cm-1

Inte

nsity

Lava Source


0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0 1000 2000 3000 4000 5000 6000

Wavenumber cm-1

Inte

nsi

ty300°K

700°K

1200°K

Lava Source


Passive FTIR spectrum of volcanic gases

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

2400 2500 2600 2700 2800 2900 3000

Wavenumber (cm-1)

Inte

nsi

ty (

un

calib

rate

d)

SO2 1+3 combination band

P-branch of HCl



0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0 1000 2000 3000 4000 5000 6000 7000 8000

Wavenumber cm-1

Inte

nsi

ty

Solar

Lava Source

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0 1000 2000 3000 4000 5000 6000

Wavenumber cm-1

Inte

nsi

ty

Lava Source


Emission spectra: if the source of radiation is cooler than the gas itself we observe emission spectra

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

760 860 960 1060 1160 1260

Wavenumber (cm-1)

Inte

nsi

ty

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.

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.

0

0.2

0.4

0.6

0.8

1

1.2

2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756

Wavenumber

Tra

nsm

ittan

ce

Measured HCl Spectrum Theoretical HCl spectrum

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

0

0.2

0.4

0.6

0.8

1

1.2

2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756

Wavenumber

Tra

nsm

ittan

ce

Measured HCl Spectrum Theoretical HCl spectrum

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.

Potential problems

0

0.2

0.4

0.6

0.8

1

1.2

2720 2730 2740 2750 2760 2770 2780

Wavenumber

Tra

nsm

itta

nce

HCl 3e17 correct

HCl 3 * 1e17 xsec

Potential problems

0

0.2

0.4

0.6

0.8

1

1.2

2720 2730 2740 2750 2760 2770 2780

Wavenumber

Tra

nsm

itta

nce

HCl 3e18 correct

HCl 3 x 1e18 xsec

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.

Simple retrieval: Masaya Volcano active source.

FTIR

IR Source

520 m

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

Complex retrievals: Etna lava fountain

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.

Complex retrieval: Etna lava fountain

Tran

smit

tan

ce

a

b

Trade-offs: Spectral resolution vs weight and SNR

Higher spectral resolution:

Pro – Better quality fits, less sensitive to non-linearity problemsCon – Larger spectrometer, with more weight and more delicateCon – 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

Trade-offs: SNR vs measurement frequency

Higher signal to noise ratio: achieved through longer integration times

Pro – Better quality spectraCon – Lower spectrum acquisition frequencyCon – potential for large variations in absorbing gas amount, leading to non-linearity problems

Higher measurement frequency:

Pro: resolve short-term variationsPro: can always average spectra to increase SNR after data collectionCon: inferior snr on individual spectra

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

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

Documents