A method to correct sampling ghosts in historic near-infrared Fourier transform spectrometer (FTS) measurements

[摘要] The Total Carbon Column Observing Network (TCCON) has been established toprovide ground-based remote sensing measurements of the column-averaged dryair mole fractions (DMF) of key greenhouse gases. To ensure network-wideconsistency, biases between Fourier transform spectrometers at differentsites have to be well controlled. Errors in interferogram sampling canintroduce significant biases in retrievals. In this study we investigate atwo-step scheme to correct these errors. In the first step the laser samplingerror (LSE) is estimated by determining the sampling shift which minimisesthe magnitude of the signal intensity in selected, fully absorbed regions ofthe solar spectrum. The LSE is estimated for every day with measurementswhich meet certain selection criteria to derive the site-specific time seriesof the LSEs. In the second step, this sequence of LSEs is used to resampleall the interferograms acquired at the site, and hence correct the samplingerrors. Measurements acquired at the Izaña and Lauder TCCON sites areused to demonstrate the method. At both sites the sampling error historiesshow changes in LSE due to instrument interventions (e.g. realignment).Estimated LSEs are in good agreement with sampling errors inferred from theratio of primary and ghost spectral signatures in optically bandpass-limitedtungsten lamp spectra acquired at Lauder. The original time series ofX_air and X_CO2 (X_Y: column-averagedDMF of the target gas Y) at both sites show discrepancies of 0.2–0.5%due to changes in the LSE associated with instrument interventions or changesin the measurement sample rate. After resampling, discrepancies are reduced to0.1% or less at Lauder and 0.2% at Izaña. In the latter case,coincident changes in interferometer alignment may also have contributed tothe residual difference. In the future the proposed method will be used tocorrect historical spectra at all TCCON sites.

[学科分类] 几何与拓扑