Dear Editor, The manuscript “Source brightness fluctuation
Contents
1. Page 449 line 21 replaced SBF corr by c Page 449 line 23 replaced right by d Page 449 line 28 inserted The residual plot in Fig 3 e shows the difference between the AC spectrum and the SBF corrected spectrum Deviations of up to 4 are revealed 14 Page 449 line 24 It is not obvious to me why the SNR should improve by the SBF correction Some explanation for this improvement would be helpful How has the SNR be determined Is the increase from 156 to 164 significant or is it within the accuracy of the SNR determination The impact of source brightness fluctuations on the SNR is described by Beer1992 page 68f lower source brightness fluctuation frequencies result in a variable offset higher fluctuation fre quencies result in an increased noise level The improvement in the SNR mentioned in the paper is small and thus we now omit this aspect to not confuse the reader gt Page 449 line 23 deleted The SBF correction slightly gt Page 449 line 24 changed In addition the procedure to The procedure 15 Page 449 line 27 What is the explanation for the oversaturation in the spectral range between 990 and 1070 cm 1 The oversaturation is due to an incorrect expression of the line depth caused by the source brightness fluctuations as described in the introduction on page 445 line 11 This has now been clarified in the manuscript gt Page 449 line 27 changed2. For the actual comparison of the different methods SBF correction for detectors with an un known offset vs SBF correction for detectors without offset it would be instructive to com pare vertical profiles or total column concentrations of species that are measured with differ ent detectors preferably in the same spectral range Is there a spectral overlap between InSb and MCT detector that would allow such a comparison If not is it possible to find one or more species where a comparison of total column concentrations derived from different spec tral ranges with different detectors could be sensible Such a comparison would enhance the value of this paper A sensitivity study concerning different filters may be interesting especially for the high require ments of the retrieval of CO2 However the effect of different low pass filters applied in the spec tral domain has been discussed by Keppel Aleks Sect 6 The effect of different filters in the in terferogram domain is small and is discussed in comment 6 referee 1 However a detailed comparison between both filters may still be interesting but we think that such a study is outside the scope of this manuscript Technical corrections Page 447 line 9 an increased appearance of clouds was expected gt This aspect has been corrected in the manuscript Page 448 line 17 instantanously gt instantaneously gt This aspect has been corrected in the manuscript Page 450 line
3. calculation of the offset The impact of small errors in the offset is now in cluded in the manuscript in terms of a short sensitivity study in Sect 3 1 This study shows that small errors in the offset do not significantly affect the method gt see comment 7 referee 1 5 How do you determine B Is it determined in the vicinity of the interferogram peaks How do you make sure that B is measured at the same source brightness as A in presence of source brightness fluctuations A and B should be measured right at the position of the interferogram s centerburst to avoid er rors in the offset due to a change in the source brightness This is now mentioned in the manu script gt Page 448 line 1 inserted The variables Aycr and Bycr both need to be determined in the vicinity of the interferogram s centerburst to avoid errors in the offset due to a change in the source brightness 6 I assume that the determination of O using equation 6 could be easily done with two differ ent but stable sources e g by looking at the sun under clear sky conditions and then looking at a room temperature target What would be the result of such a characterization measure ment How does it compare to the methods presented in the paper We agree with the referee that the proposed method can in principle be used to determine the offset The method would also lead to two equations and two variables modulation efficiency and offset which can the
4. gt Page 448 line 16 replaced measured spectral region by used detector 3 How do you handle the offset in this automatic application Is the offset also determined automatically If yes which method is used Please clarify We mentioned on page 448 line 16 that The procedure can be applied automatically In this study the offset has been removed manually Ila Section 3 1 Fig 3 left G nterferograms At page 447 line 10 you state that the instrument has measured in DC mode In figure 3 however you present AC and DC data Please explain how the AC data has been generated The AC interferograms are numerically deduced from the DC interferograms This is now men tioned in the manuscript see also comment 8 referee 1 gt Page 448 line 24 inserted In Sect 3 DC interferograms are compared to SBF corrected interfero grams and AC interferograms The AC interferograms are numerically deduced by subtracting the smoothed interferogram from the raw interferogram within OPUS 11b The interferogram intensity is given in What is the reference i e what corresponds to 100 same question for Fig 4 The interferograms intensity is the percentage efficiency of the instrument s analog to digital con verter This is now mentioned in the manuscript gt Page 449 line 6 inserted The interferograms intensity is described as the percentage efficiency of the instrument s ADC analog to digital conve
5. is calculated A convolution of the interferogram with a rectangle corresponds to a multiplication of the spectrum with a sinc function This sinc function then serves as a lowpass filter So in fact the comparison presented in section 3 2 is not really a comparison between two methods but a comparison between two different ways of applying a lowpass filter while us ing in principle the same method Since none of the filters is described in detail the evalua tion of this comparison is difficult We assume a misunderstanding here As discussed in comment 2 we want to test if our SBF correction method which is the reweighting of the SBF influenced interferogram is correct Since our reweighting differs from the approved application presented by Keppel Aleks we in cluded Sect 3 2 to show that our SBF correction is suitable for a reweighting of DC interfero grams It seems however worthwhile to evaluate the effect of the filter parameters on the accuracy of the method Such a study should not only comprise two examples of lowpass filters as it is done in the present paper but it should be performed in a more systematic way investigating the effect of different filter widths and filter forms and maybe also discuss the differences when the filter is applied in the interferogram or spectral domain respectively Therefore I suggest to discuss the effect of the different filters in more detail and in a more systematic way in this section
6. 16 identically gt identical gt This aspect has been corrected in the manuscript
7. Dear Editor The manuscript Source brightness fluctuation correction of solar absorption Fourier Trans form mid infrared spectra has been revised according to the referees comments Comments to Referee 1 1 Introduction The description of the effects due to variable DC is vague intensity fluctua tions can distort the fractional line depth in FTIR It would be more appropriate to describe the primary effect as ILS distortion due to an additional apodisation which results from the variable atm transmission during interferogram recording We agree with the referee and included this aspect in the manuscript gt Page 445 line 12 inserted The distortion of the fractional line depth is due to an additional apodisa tion which results from the variable source intensity and distorts the instrumental line shape ILS 2 Please give more details on the MCT detector Do you use photoconductive or photovoltaic detectors In this study we use a standard Bruker photoconductive MCT detector This has been mentioned in the introduction on page 446 line 10 measured with a photoconductive MCT detector For clari fication we now mention the photoconductive MCT detector also in Sect 2 gt Page 447 line 11 changed Measurements with the MCT detector to Measurements in the mid infrared spec 2b tral region were performed using a photoconductive MCT detector at low signal intensity in order to
8. However the AC spectrum shows a strong oversaturation in this range whereby this effect to However the AC spectrum shows a strong oversaturation in this range due to an incorrect expression of the line depth caused by the source brightness fluctuations This effect 16 Section 3 1 figure 4 How has the offset been determined for these data Has the same offset been used to correct all three interferograms The offset has been calculated using the first method with a fixed modulation efficiency of Minsb 87 This is now mentioned in the manuscript gt Page 450 line 6 inserted which is removed for the SBF correction using the first method with a fixed modulation efficiency of Mins 87 17 18 Section 3 2 page 451ff The new SBF correction method presented up to know determines and subtracts an unknown offset from the detector DC signal prior to the normalization of the interferogram This is in my opinion the essential innovation with respect to the method presented by Keppel Aleks et al For offset free detectors the differences in the two methods are reduced to the application of different spectral lowpass filters that are once applied in the spectral domain Keppel Aleks et al and once in the interferogram domain this paper Smoothing the interferogram is equivalent to a convolution with a rectangle where the width of the rectangle is determined by the number of datapoints n over which the running mean
9. avoid non linearity effects What is the origin of the offset voltage A basic preamp circuit diagram would help The offset is due to a constant voltage which leads to a current flow through the detector element as described on page 446 line 12 The used detector is a standard Bruker photoconductive MCT detector A circuit diagram of the preamplifier of this detector can be found in each Bruker User s Manual Thus we did not include a preamp circuit diagram into the manuscript instead we now refer to the Bruker User s Manual with a footnote in the manuscript gt Page 446 line 12 inserted footnote A preamplifier circuit diagram can be found in the 120M Bruker user s manual 2c 2d 2e 3 Is the offset primarily of electrical origin or is instrumental selfemission relevant as well See comment 3 To which extent would the suggested methods be affected by nonlinearity effects Nonlinearities have an effect on the interferogram recording However for the solar absorption measurements in this manuscript we use low signal intensities to avoid non linear effects as men tioned on page 447 line 11f How stable is the deduced offset as function of time See comment 7 referee 2 Would instrumental self emission inflict a loss on the methods proposed Selfemission might contribute to both the DC and AC parts of the interferogram but will not scale with the atm transmission changes Concerning instru
10. derstand this sentence Why should an SBF correction be wavelength dependent apart from the fact that different wavelength ranges are covered by different detectors If this is the point you want to make then I suggest to replace independent of the measured wave length by independent of the used detector We want to express that the SBF correction is independent of the used detector This has been changed in the manuscript gt Page 446 line 25 replaced measured wavelength by used detector 2 Page 446 line 29 In addition we compare our SBF correction method to the method pre sented by Keppel Aleks et al 2007 I do not agree with this statement The SBF correction method presented in this paper deals with the unknown offset of MCT detectors but the comparison is done for offsetfree InGaAs detectors in both cases See also comments on section 3 2 below We assume a misunderstanding here In the introduction we distinguish between the SBF correc tion which is the reweighting of the interferogram according to Eq 1 and the removal of the off set Page 446 line18 Our SBF correction method the reweighting of the interferogram differs from the approved method used in previous studies Keppel Aleks as described on Page 445 line24ff and Page 448 line 11 Thus Sect 3 2 is considered as a test of our SBF correction method as mentioned in the abstract For a comparison with previous studies we apply our source brightn
11. dy to understand how accurately the offset determination needs to be I would expect that the correction stays largely intact even if we allow some uncertainty in the offset determination You could check this by applying different offsets in the retrieval study underlying Fig 4 We agree with the referee that a short sensitivity study to understand how accurately the offset determination needs to be is interesting for the manuscript We performed such a sensitivity study based on the examples of ozone total column concentrations underlying Fig 4 The sensitivity study shows that the impact of errors in the offset increases when the influence of SBF on the in terferogram increases However it also shows that small errors in the offset do not have a sig nificant impact on the total column retrieval gt Page 450 line 29 inserted Based on the three examples in Fig 4 we now study the impact of errors in the determination of the offset on the retrieval of O total column concentrations In Tab 2 the to tal column concentrations of O and their percentage deviations to the AC reference value for the three examples and for different applied offsets are presented In the undisturbed case 1 the offset does not affect the O retrieval However the impact of errors in the offset on the retrieval increases in the second 2 and third 3 case Large errors in the offset 8 5 lead to an error of up to 3 1 in the total column concentration Small
12. e modulation efficiency is dependent on the wavelength with typical values of 50 in the UV to 90 in the mid infrared The impact of different detec tor responsivity within the filter region is in our point of view small and only has a small impact on the determination of the modulation efficiency and a even smaller impact on the calculation of the offset The same argument applies in our opinion to the change in the modulation efficiency towards smaller wavenumbers To our experience the modulation efficiency does not drastically change between 1900 2700cm and 900 1300cm These assumption are confirmed by comparing the offset calculated with the modulation efficiency to the offset calculated with the second method Sect 3 1 The small error in the offset does not have a significant impact on the method as now showed in a short sensitivity study in Sect 3 1 see comment 7 For our measurements of the modulation efficiency we used a filter with a filter bandwidth from 1900 2700cm This is now mentioned in the manuscript gt Page 447 line 18 inserted bandwidth 1900 2700cm 5 Application of the OPUS running mean function Do you apply the OPUS smoothing function with a broad data point window just once We applied the smoothing function twice with a data point window of 1000 This is now men tioned in the manuscript gt Page 448 line 14 deleted mean over n datapoints gt Page 448 line 15 inserted The smoo
13. errors in the offset determination on the other hand do not significantly change the total column concentrations an error of 1 3 in the offset in case 3 leads to an error of less than 1 in the total column concentration gt Inserted Table 2 gt Inserted caption Table 2 Impact of errors in the offset determination on the retrieval of O total column concentrations percentage deviations are related to the initial value 6 6877x10 mol cm calculated from the undisturbed AC spectrum Numbers are given in 10 mol cm 8 Technical phrasing typos Abstract line 12 ff The analysis of trace gas concentrations is fundamental It is not clear to me what the authors want to express with this statement This phrase is repeated on page 446 line 7 We wanted to express that a SBF correction of spectra in the mid infrared spectral region is im portant to improve the data quality We omitted the sentence in the abstract and changed the cor responding sentence on page 446 gt Abstract line 12 deleted The analysis of trace gas concentrations gt Abstract line 13 inserted in the mid infrared spectral region gt Page 446 line 7 changed However in the mid infrared spectra region the analysis of many im portant trace gases is fundamental to However the application of a SBF correction to spectra in the mid infrared spectral region 700 3800cm for the analysis of many important t
14. ess fluctuation correction For clarification we changed the word choice on page 446 line 29 gt Page 446 line 29 changed In addition we compare our SBF correction method to the method to In addition we test our SBF correction method by comparing it to the method 3 Page 447 line 10ff Please indicate which spectral range exactly is covered by which detector for the two spectrometers A table would be helpful We agree with the referee and added a table with the spectral range of each detector in the two spectrometers gt Page 447 line 14 inserted The spectral coverage of each individual detector used in the IFS125HR and the IFS 120 5M during solar absorption measurements is shown in Tab 1 gt Inserted table with spectral range of each detector and instrument gt Inserted table caption Spectral range of different detectors used during solar absorption measure ments with the 120 5M and 125HR instrument 4 Section 2 Method How accurate is the determination of the offset What are the requirements for the accuracy of the offset determination in order to get a good SBF correction How accurate is the determina tion of A Does the interferogram sampling have an impact of the determination of A e g if the interferogram maximum and or minimum is not hit by a sampling point The interferogram sampling only has a small impact on the determination of A and an even smaller impact on the
15. mental self emission two cases have to be distinguished instrumental self emis sions which result from emissions in front of the interferometer block and self emissions which result from emissions behind the interferometer block Self emissions in front of the interferome ter block are mainly blocked by two apertures the first one in front of and the second one behind the interferometer block Instrumental self emissions behind the interferometer block contribute to the offset but will be corrected by the presented method We believe that instrumental self emissions do not significantly inflict a loss on the proposed method Furthermore due to the high temperature of the sun compared to the instrument the contribution from self emissions will be very small 4 Why does the first approach of determining the offset by using a measurement taken with an InSb detector works so properly We should expect that the interferometer s modulation effi ciency depends on wavenumber Since the InSb spectral responsivity differs from the MCT even identical filter and optical settings will not guarantee that the same modulation efficiency is found Which filter bandwidth do you recommend did you use in your setup for this exer cise Finally the InSb longwave response cut is located at higher wave numbers how can you cover the MCT spectral region of interest 700 1300 cm 1 with an InSb reference meas urement We agree with the referee in this point Th
16. n be used to determine the offset following equation 6 It is also possible to use a blackbody with different intensities to determine the offset following equation 6 However our intension with the second method is to use available solar absorption measurements to deter mine the offset without further measurements 7 How stable is the offset in time Is it sufficient to determine the offset once using dedicated measurements or must it be determined regularly during measurement campaigns Using method one to determine the offset implies a regular check of the modulation efficiency of the instrument since the offset is dependent on the modulation efficiency This is now mentioned in the manuscript gt Page 448 line 1 inserted Since the offset is dependent on the modulation efficiency of the instrument the modulation efficiency has to be determined regularly to avoid errors in the offset due to a change in the instrument alignment Using method two the offset is independent of the instrument alignment We now mention this aspect in the manuscript gt Page 448 line 9 changed This method is independent of using an additional detector to This method is independent of the alignment of the instrument 8 Page 448 line 1 1ff Here you describe the SBF correction using Eq 1 How do you handle the offset during the processing Is it subtracted beforehand or is it subtracted within the OPUS software This is not clear The
17. offset is subtracted with the OPUS software before applying the SBF correction We now mention this aspect in the manuscript gt Page 448 line 13 inserted the offset is subtracted from the raw interferogram using the OPUS calcula tor gt Page 448 line 14 inserted again 9 Page 448 line 14 mean over n datapoints How large is n We applied the smoothing function twice with a data point window of 1000 This is now men tioned in the manuscript gt Page 448 line 14 deleted mean over n datapoints gt Page 448 line 17 inserted The smoothing is accomplished by applying twice a running mean with a data point window of 1000 Have you tried different values for n See comment 6 referee 1 10 Page 448 line 15 The procedure has the advantage that the correction is independent of the measured spectral region and it can be applied automatically and instantaneously together with the measurement process 1 Advantage An advantage when compared to what We want to express that the procedure is independent of the used detector For clarification we changed the word choice in this sentence gt Page 448 line 15 deleted has the advantage that the correction 2 Independent of the measured spectral region Again I guess you mean it is independent of the used detector the dependency on the spectral region is then implicit because different de tectors are used in different spectral regions
18. race gases such as QO is reasonable Page 450 line 11 sicnificantly gt significantly gt This aspect has been corrected in the manuscript Section 3 1 I assume that the AC IFG in Fig 3 is deduced from the DC IFG numerically no parallel digitization of AC and DC ifgs as also possible with the Bruker acquisition in princi ple Please state somewhere where this AC ifg comes from The AC interferograms are numerically deduced from the DC interferograms This is now men tioned in the manuscript gt Page 448 line 24 inserted In Sect 3 DC interferograms are compared to SBF corrected interfero grams and AC interferograms The AC interferograms are numerically deduced by subtracting the smoothed interferogram from the raw interferogram within OPUS Fig 3 Please mark the zero level for both the black and red spectra e g by dashed lines gt Fig 3 zero levels have been inserted Please state that the spectra have been shifted along the ordinate for clarity gt Fig 3 caption SBF correction of a solar absorption MCT interferogram a AC interferogram b DC interferogram c SBF corrected interferogram d AC spectrum and SBF corrected spectrum shifted along ordinate e residual plot of AC spectrum and SBF corrected spectrum Comments to Referee 2 1 Introduction page 446 line 26ff We present a method of source brightness fluctuation cor rection which is independent of the measured wavelength I do not un
19. rter 12 Section 3 1 Fig 3 right spectra Please indicate where the zero level is A difference plot showing the residual between the two spectra may be instructive 13 Figure 3 is densely packed with information I suggest to enlarge the plots and to use the full width of the page for the interferograms as well as for the spectra and to supplement the spec tra by adding a difference plot We agree with the referee and changed Fig 3 according to the referee s suggestions gt Fig 3 Zero levels have been inserted The whole figure has been enlarged A difference plot has been included gt Fig 3 caption caption has been adjusted according to the new figure SBF correction of a solar ab sorption MCT interferogram a AC interferogram b DC interferogram c SBF corrected interfero gram d AC spectrum and SBF corrected spectrum shifted along ordinate e residual plot of AC spec trum and SBF corrected spectrum gt Sect 3 1 has been adjusted according to the new figure e Page 449 line 4 changed sentence Three interferograms to In Fig 3 a c three interfero grams are shown AC DC SBF corrected in Fig 3 d two spectra created from the AC and the SBF corrected interferograms are presented and in Fig 3 e a residual plot determined from the AC and the SBF corrected spectra is displayed Page 449 line 8 replaced AC by a Page 449 line 9 replaced DC by b
20. thing is accomplished by applying twice a running mean with a data point window of 1000 6 Did you alternatively try repeated smoothing runs with narrower kernel widths instead Is the result identical If not which approach seems preferable How do you determine the required amount of smoothing Is this choice critical We applied the same smoothing in the MCT examples 3 1 as well as in the InGaAs study 3 2 and prove that the choice of the smoothing is suitable for the SBF correction For clarification we now mention the setup of the smoothing also in Sect 3 2 gt Page 451 line 13 inserted For generating I mootn the OPUS smoothing function is applied twice with a data point window of 1000 In addition we tested the impact of different kernel widths 2x5000 2x1000 3x500 4x100 5x10 The choice of the different kernel is not critical as long as the smoothing does not affect the modu lation frequencies A good lower boundary is 500 data points This is now mentioned in the manu script gt Page 448 line 17 inserted The choice of the data point window is not critical as long it does not affect the modulation frequencies Here a reasonable lower boundary for the window size is 500 data points 7 Section 3 1 It is shown that both methods indicate very similar values for the offset How sensitive the SBF correction is wrt errors in the offset determination It would be interesting to include a short sensitivity stu
Download Pdf Manuals
Related Search