«ABSTRACT Studies were performed to carry out semi-empirical validation of a new measurement approach we propose for molecular mixing ratios ...»
Semi-empirical validation of the cross-band relative absorption
technique for the measurement of molecular mixing ratios
Denis Pliutau, and Narasimha S. Prasad
NASA Langley Research Center, 5 N. Dryden St., MS 468, Hampton VA, 23681
Studies were performed to carry out semi-empirical validation of a new measurement approach we propose for
molecular mixing ratios determination. The approach is based on relative measurements in bands of O2 and other molecules and as such may be best described as cross band relative absorption (CoBRA).. The current validation studies rely upon well verified and established theoretical and experimental databases, satellite data assimilations and modeling codes such as HITRAN, line-by-line radiative transfer model (LBLRTM), and the modern-era retrospective analysis for research and applications (MERRA). The approach holds promise for atmospheric mixing ratio measurements of CO2 and a variety of other molecules currently under investigation for several future satellite lidar missions. One of the advantages of the method is a significant reduction of the temperature sensitivity uncertainties which is illustrated with application to the ASCENDS mission for the measurement of CO2 mixing ratios (XCO2).
Additional advantages of the method include the possibility to closely match cross-band weighting function combinations which is harder to achieve using conventional differential absorption techniques and the potential for additional corrections for water vapor and other interferences without using the data from numerical weather prediction (NWP) models.
Keyword list: lidar, molecular mixing ratio, HITRAN, spectroscopy, absorption, radiative transfer
1. INTRODUCTION Several active sensing lidar missions relying on Integrated Path Differential Absorption (IPDA) approach for the mixing ratio measurements of the atmospheric molecules such as CO2, CH4 and N2O are being investigated.1-7 Examples of these future lidar missions include Active Sensing of CO2 Emissions Over Nights Days and Seasons (ASCENDS), Methane Remote Lidar Mission (MERLIN), and Advanced Space Carbon and Climate Observation of Planet Earth (A-SCOPE).5-7 Previous studies have identified a number of spectral interferences such as those due to temperature, pressure variations and water vapor absorption as making significant contribution to the total estimated mixing ratio uncertainties.1-7 One of the ways these uncertainties due to variations in atmospheric parameters of temperature, pressure and humidity are carried over to the derived mixing ratios is through the extraction of the altitude distribution of the column averaged mixing ratio values by means of fitting techniques which utilize atmospheric profiles data only known with a limited precision.8 The uncertainties introduced this way could be significantly reduced if the need in the altitude mixing ratio fitting could be eliminated or the accuracy requirements on the known atmospheric data relaxed.
To address the accuracy limitations imposed by the temperature and pressure induced uncertainties in the IPDA technique and provide a way for additional optimizations to reduce the influence of water vapor interferences, we have suggested a Cross-Band Relative Absorption technique which relies on spectral lines matching in bands of various molecules (i.e. CO2, CH4, N2O etc.) with those located in a reference band of oxygen such as that at 1.26-1.27 µm.9 The key component of the proposed CoBRA methodology is the reduction of the temperature and total pressure induced uncertainties though the selection of measurement and reference wavelengths by matching individual spectral lines in the measurement and reference channels having similar spectral characteristics. The CoBRA method also has the advantage of providing a possibility to closely match weighting functions in the measurement (i.e. molecule of interest) and reference (i.e. oxygen.) bands holding potential for self-sufficient measurements of atmospheric molecular mixing ratios without external corrections for temperature and other atmospheric parameters.9 The key component of the CoBRA methodology is the temperature compensation throughout the entire range of total pressure variations and the weighting functions matching by properly selecting combinations of spectral lines and excitation wavelengths. In this paper we describe selected components of the validation methodology we employed to verify the temperature and total pressure sensitivity compensation potential of the Cross-Band Relative Absorption Technique (CoBRA) for the purpose of molecular mixing ratio measurements in the atmosphere. Results of the CoBRA temperature / total pressure sensitivity analysis in selected spectral bands for the molecules of CO2 and CH4 are presented including suitability comparisons of the A-Band and 1.26-1.27 micron band of oxygen for the CoBRA technique. Additional considerations related to the water vapor and aerosols interferences as well as the background correction techniques are discussed.
2. CoBRA TECHNIQUE OVERVIEW AND VALIDATION ALGORITHM IMPLEMENTATION
The implementation of the CoBRA technique is based on two components: (1) the selection of spectral lines with close-matching spectral parameters and (2) the proximity of half-width relative spectral distances from the center of the line to the location of the measurement wavelengths ( f m and f r ) as shown in Fig. 1
The exact match of line intensities (Im and Ir) is not required, however the proximity of line intensities is desirable due to limitations on the usable optical depth values suitable for the space lidar applications.10 These limits on the optimum optical depth impose restrictions on the available spectral line regions for cross-band line matching. Such limitation dictated by the line intensity induced constraints is illustrated further in this paper by performing the CoBRA analysis for the Methane (CH4) molecule using the A-Band of oxygen.
The validation methodology for the CoBRA technique we have selected is based on the HITRAN database, Lineby-line radiative transfer model (LBLRTM), and the modern-era retrospective analysis for research and applications (MERRA).11-13 Our method employs pre-analysis of the MERRA dataset on a temperature / pressure grid to reduce the number of calculation required as described in our precious presentation.14 The selected temperature / pressure binning methodology for a span of year 2009 and all locations is further explained in Fig. 2. As can be seen, the original analysis of the “inst6_3d_ana_Nv” MERRA data was performed on a grid of 1 Kelvin x 1 mbar as shown in Fig 2a.
The finer 1mbar bins where then merged together to correspond to the 72 tabulated layers of the MERRA dataset thus reducing the number of pressure level grids to 72. These 72 bins were selected so that the tabulated MERRA pressure level values shown in Fig 2band 2c correspond to the centers of the pressure level bins formed. The reduction of the altitude (pressure) level bins was partially dictated by the need to reduce the number of calculations necessary to perform the complete temperature / pressure sensitivity analysis.
Fig 2 Results of the MERRA dataset temperature analysis on a 1K x 1mbar grid with color coded frequency of occurrence of temperature /.total pressure combinations (a), binning methodology and indexing with MERRA tabulated pressure level grid of 72 layers applied to the analyzed MERRA data in (a), and the illustration of the individual bin selected dimensions (c).
where i – is the index of the temperature bin within a pressure layer j, Ti and Pj and the temperature and pressure corresponding to the temperature bin i and a pressure level j respectively, τmeas and τref are the optical depth values in the measurement and reference bands respectively at the measurement and reference wavelengths of ωmeas and ωref These variables are further used to calculate the variances and mean value for further determination of the relative coefficient of variation representative of the temperature induced error within each layer. The combining of the individual layer errors into the total path uncertainty is accomplished by scaling individual layer errors by the total molecular density at the corresponding pressure levels. We have selected a validation approach based on the calculation of variances as opposed to multivariate techniques such as PCA partly due to the ease of comparison with the results of experimental measurements.
The capability to calculate 3 different cases of temperature dependences (Uncorrected, IPDA, and CoBRA) enables cross-comparison of pure temperature sensitivity with that resulting from the application of IPDA and CoBRA methods.
Pure temperature analysis for the alternative bands of CO2 and O2 considered for the ASCENDS mission applications have been reported in Ref. 15. The results of such comparison will be presented in the next section
3. UNCORRECTED, IPDA AND CoBRA ANALYSIS FOR SELECTED LINES OF CO2
We have used the calculation methodology described in section 2 to determine the temperature induced optical depth uncertainties as a function of wavelength for the CO2 in the 1.57µm band for the uncorrected, IPDA and CoBRA approaches. Other molecules and spectral bands were also investigated. Such analysis allowing comparison with the uncorrected optical depth uncertainties introduced due to temperature variations for different spectral lines and regions is useful as it provides quantitative reference information for evaluating the level of improvements in temperature induced uncertainties due to the introduction of an off-line wavelength (i.e. IPDA) or of the reference oxygen channel (CoBRA).
Fig. 4 shows the comparison of the maximum and minimum achievable CoBRA technique accuracies in the measurement CO2 band of 1.57 micron referenced to the band of.oxygen in the 126-1.27 micron region. The point distribution plots displayed in Fig 4c and 4e are the maximum and minimum possible accuracies at as a function of measurement wavelength for all possible off-line combinations of reference wavelengths in the band of oxygen. For comparison the uncorrected coefficient of variation is plotted in Fig 4b and 4d. As can be seen, the CoBRA method results in significant accuracy improvements compared to uncorrected uncertainties at the optimum wavelength combinations but may also result in increased errors exceeding the uncorrected values for the worst case selection of wavelength pairs. The “uncorrected” temperature sensitivity analysis shown in Fig 4b and 4d was carried out in the same way as that for the IPDA and CoBRA calculations with the exception that no off-line or reference wavelengths were used by performing single wavelength uncorrected temperature sensitivity calculations.
It should be noted that the distribution of the low temperature sensitive regions in the 1.57 micron band of CO2 shown in Fig 4b and 4d is in agreement with the temperature sensitivity analysis results for the 1.57 micron band previously reported by Menzies et. al. further supporting the correctness of the results obtained using our validation methodology.1 Fig 4 Comparison of uncorrected temperature induced uncertainties in the 1.57 micron band of CO2 (b and d) with the maximum (c) and minimum (e) achievable temperature induced uncertainties using the CoBRA method. Spectrum in (a) is a US Standard atmosphere calculated spectrum for an a vertical path of 80km. The results are for due to annual variations in atmospheric temperature and pressure for year 2009.
Fig 5 Comparison of maximum and minimum achievable temperature induced errors using IPDA (c, e) and CoBRA (d, f) techniques with the uncorrected temperature induced error levels (b) for a sample CO2 line in the 1.57 µm band (a).
The comparison of the results displayed in Fig 4d and 4e also shows that low temperature sensitivity levels may be achieved using the CoBRA approach even for spectral regions which exhibit the highest uncorrected temperature sensitivities. Error reductions for such regions drastically expand the span of possible excitation wavelengths. This result is particularly important from the stand point of additional optimizations such as those involving selection of wavelengths to reduce water vapor interferences.