The FTIR has relatively large measurement cell with a large surface area. Sampling lines and buffer volumes also lengthen the response time of the calibration system. The combination of the high frequency calibration requirement and the slow response times makes it impractical to perform the 2 standard calibrations that have been outlined for in-situ analysers by \citet{Bastrikov_2014}. The best approach for the FTIR is to chose a calibration strategy that reduces memory effects and enables the calibration of the humidity-isotope effects and the VSMOW scale \cite{Steen_Larsen_2013}. For the FTIR system calibrations performed every 3 hours presents a suitable trade-off between minimising the effect of instrument drift and loss of measurement time. This presents an inadequate amount of time to measure 2 standards or characterise the humidity-isotope effect over the expected H2O range for a long term-deployment. We therefore explore using the direct isotopologue calibration method introduced by \citet{Griffith_2012} and employed by \citet{Vardag_2015} for CO_2 isotope ratio measurements on a similar FTIR instrument.
The direct isotopologue calibration determines the instrument response function for each isotopologue used to calculate isotope ratios. This technique has the advantage that there is no need to charcterise both the VSMOW calibration and humidity-isotope response functions, as these are explicitly accounted for. Breifly, the instrument response function is characterised by supplying the instrument with known mixing ratios of each isotopologue (X and X_i) that covers the range expected to be observed. By then applying the determined calibration functions the corrected isotope ratios are calculated:
\[\delta=\bigg[\frac{\chi_i}{\chi}-1\bigg]\]
The minor water vapour isotopologue calibrations for the FTIR are non-linear (see figure 1), so the measured isotope ratios including their instrument response functions are shown by:
\[\delta=\bigg[\frac{a_0+a_1\chi_i+a_2\chi_i^2+\cdots+a_n\chi_i^n}{b_0+b_1\chi}-1\bigg]\]
Clearly the instrument response function for \(\chi\) are required for the direct calibration and for calculating the true \(\chi\)_i mixing ratio from the true \(\delta\) and \(\chi\). The \(\chi\) calibration is also subject to drift and is not possible to characterise using the FTIR calibration system. For a general assessment of the instrument response function of the minor isotopologues:
use co-located meteorological measurements to determine the instrument response function for the water vapour mixing ratio. These were collected over several months and are shown in figure 1a. Show some curvature but are ok for a general characterisation of the instrument.
Using a single isotope standard vary the mixing ratio of the water vapour supplied to the FTIR.
Use the known \(\delta\) of the standard and the calibrated mixing ratio to calculate the true \(\chi\)_i.
Fit linear regression to the instrument response function and calculate residuls of the linear fit to determine the curvature of the instrument function.
residuals calculated as \(\Big[\frac{\chi_i-\rm residuals}{\chi_i}\Big].1000\)
increase order of polynomial fit to instrument response function until residuals are minimised.
, we use the co-located meteorological measurements over a period of several months to first est
The real time calibration uses equation 2 to calibrate the FTIR isotope ratio measurements. To account for instrumental drift, regular measurements of a standard are performed:
\(\chi\) from met - how?
\(\chi\)_i from met calibrated \(\chi\) and known \(\delta\)
only small \(\chi\)_i and \(\chi\) to dynamically span the observe mixing ratios - similar to \citet{Wen_2008}
apply linear correction.