Interactive comment on “ NO 2 observations over the western Pacific and Indian Ocean by MAX-DOAS on Kaiyo , a Japanese research vessel ”

This is an interesting paper which reports unique MAX-DOAS observations of NO2 volume mixing ratios over the western Pacific and Indian Ocean. Such observations are scarce and particularly important for validation of chemical transport models and satellites. In addition, the study reports on a technical issue related to the choice of the optimal fitting window for NO2. This is relevant for future MAX-DOAS studies under comparable conditions. In general the paper is well-written and well-structured. I recommend it for publication in AMT after the following comments have been addressed:


Introduction
Multi-Axis Differential Optical Absorption Spectroscopy (MAX-DOAS) is a recently developed remote sensing technique designed for atmospheric aerosol and gas profile measurements using scattered solar radiation at several elevation angles (e.g.H önninger et al., 2004;Wagner et al., 2004;Sinreich et al., 2005;Frieß et al., 2006).It is useful for measuring a priori profiles for satellite retrievals and for validating chemical transport models.

Back Close
Full 2009; Sinreich et al., 2010), have been developed.Ship-borne measurements provide information on background concentrations over the ocean and can be used to clarify transport processes from polluted areas to the ocean, emissions from ocean to air, and emissions from ships.However, even in the case of NO 2 , spatial and temporal variations over the ocean are not fully understood, due in part to the difficulties encountered in measuring low concentrations.
In general, MAX-DOAS measures the trace gas content over a long light path (up to ∼10 km) with low elevation angles, thereby enabling the detection of low concentrations of the components of interest or weak absorbers near the ground.Thus, MAX-DOAS is useful for quantifying tropospheric trace gas over remote areas/ocean, where concentrations of the component of interest are generally low.As an example, NO 2 measurements by MAX-DOAS have been conducted at a remote Japanese island, Okinawa Island, yielding concentrations as low as ∼0.2 ppbv (Takashima et al., 2011).
Since 2007, the Japan Agency for Marine-Earth Science and Technology/Research Institute for Global Change (JAMSTEC/RIGC) has been conducting continuous MAX-DOAS measurements at several sites in Asia and Russia using compact, low-power spectrometers (e.g.Irie et al., 2009Irie et al., , 2011;;Takashima et al., 2009Takashima et al., , 2011)).The instrument has been validated with other instruments, yielding differences of less than ∼10 % for NO 2 and oxygen dimer (O 4 ) differential slant column densities (DSCDs) (Roscoe et al., 2010).Here, we report on the development of a MAX-DOAS instrument for use on ocean vessels, using an active-type gimbal to keep the telescope horizontal.
As a first step, we focus only on NO 2 because it is somewhat easier to retrieve than other components, although MAX-DOAS has the potential to perform simultaneous profile measurements on aerosol and several gas components, such as NO 2 , water vapor (H 2 O), SO 2 , IO, BrO, HCHO, and CHOCHO (e.g.Irie et al., 2011).We also performed a sensitivity analysis with three fitting windows for NO 2 retrieval and we discuss the detection limit of the NO 2 retrievals, because the sensitivity has yet to be fully investigated.Introduction

Conclusions References
Tables Figures
The second cruise (KY09-01) was conducted from 5 February to 10 May 2009 from Yokohama (Japan; 139.65  S), the Indian Ocean, Bari again, and finally returning to Yokohama (Fig. 1).We did not perform measurements in the territorial waters of Guam (United States) during cruise KY09-01 because we did not obtain permission from the relevant authorities.Note that this study is the first to report MAX-DOAS measurements over the western Pacific and Indian Ocean.

Compact, low-power instruments for MAX-DOAS
A compact, low-power and low-cost instrument for MAX-DOAS has been developed by JAMSTEC/RIGC and PREDE Co., Ltd.(Tokyo, Japan), and the instrument has been used for continuous measurements at several sites in Asia and Russia.In this study, we developed the instrument for use on ocean vessels by employing an active-type gimbal to keep the telescope horizontal on the vessel.The gimbal-mounted telescope unit was installed on the top deck of the vessel and the line of sight was toward the starboard side (Fig. 2).The movable mirror of the telescope unit rotates through six different elevation angles (ELs) of 3 • , 5 • , 10 • , 20 • , 30 • , and 70 • every 30 min, with a field of view of ∼0.9 • .For the first cruise (KY08-05), a miniaturized UV/visible spectrometer (USB4000, Ocean Optics) was installed inside the telescope unit, and the spectra data were recorded by a laptop located indoors on the vessel.The telescope and spectrometer Introduction

Conclusions References
Tables Figures

Back Close
Full were connected to each other by a 1-m (KY08-05) or 10-m (the second cruise, KY09-01) fiber optic bundle cable that consists of seven cores with radii of 100 µm.The typical exposure time was 0.08 s, and the spectra data were averaged and recorded every second by a laptop located indoors.Measurements were made over the spectral range of 230 to 560 nm with a spectral resolution of ∼0.6-0.7 nm.To suppress background noise and wavelength shifts of the spectrum, the temperature of the spectrometer was kept at 45 • C for KY08-05 and at 40 • C for KY09-01, for the entire observation period.
For comparison with in situ measurements, we use MAX-DOAS data obtained at Yokosuka, Japan (35.32 • N, 139.65 • E), which have been measured continuously since April 2007 using basically the same instrument as that used at Okinawa (Takashima et al., 2009), employing a USB4000 spectrometer and 5-m fiber optics.The azimuth angle of the observations was set to +37.0 from north (the plus sign indicates a clockwise direction).A movable mirror turns through six different ELs (3 • , 5 • , 10 • , 20 • , 30 • , and 90 • ) every 30 min, with a field of view of <1 • .The spectrometer was kept 20 • C.

Active gimbal system
To keep the telescope unit horizontal on the vessel, it was mounted on an activegimbal developed by JAMSTEC/RIGC and PREDE.In this system, the gimbal is controlled horizontally by reducing the difference between the standard horizontal level and the present level for both the roll and pitch angles, using two inclinometers (SEIKA Mikrosystemtechnik GmbH, N2).
To monitor how well the gimbal maintains a horizontal orientation, we installed another sensor inside the telescope unit (Honeywell, HMR3500) and recorded the roll and pitch angels at a repetition rate of 5 Hz.For the spectral analysis, spectra data were selected with a criterion for the elevation angle to be within ±0.2 • of the target (see below).In addition, we recorded the heading of the vessel, the roll and pitch angles of the vessel, longitude, latitude, and time.Introduction

Conclusions References
Tables Figures

Back Close
Full • of the target elevation angle for ∼60 % of the time.In general, in the case of a regular cycle of ship motion (e.g. a sine function), the gimbal performs well in controlling the horizontal level, but it is commonly unable to control the horizontal level in the case of an irregular cycle.Figure 3 shows an example of variation of the maximum intensity of the spectrum, in which we can see change in the intensity corresponding to change in the elevation angle (from 3 • to 5 • ).

Data analysis
The measured 1-s spectra were selected with a criterion for the elevation angle to be within ±0.2 • of the target elevation angle and averaged every 1 min.Here, we excluded spectra data for periods when we changed the target elevation angle of the telescopeunit mirror.The azimuth of the heading of the vessel, longitude, altitude were also averaged every 1 min.
To retrieve a vertical profile of NO 2 concentration, we used the Japanese MAX-DOAS profile retrieval algorithm, version 1 (JM1) (Irie et al., 2011).The averaged spectrum was analyzed using the DOAS method (Platt, 1994), employing nonlinear least squares spectral fitting (Rodgers, 2000) to derive the DSCD of the oxygen collision complex (O 2 -O 2 or O 4 ) and NO 2 .Here, DSCD is defined as the difference between the column concentration integrated along the sunlight path measured at a low EL (EL < 70 • to derive the DSCD of the oxygen collision complex) and that at EL = 70 • to derive the DSCD of the oxygen collision complex.
The box air mass factor (A box ), which is defined as the air mass factor for a given layer, was derived from the O 4 DSCD inversion with the Monte Carlo Atmospheric Radiative Transfer Simulator (MCARaTS) (Iwabuchi, 2006).Using the A box , we retrieved Introduction

Conclusions References
Tables Figures

Back Close
Full NO 2 profiles in the lower troposphere with a vertical step of 1 km from the NO 2 DSCD measurements.Details of the retrieval algorithm have been described elsewhere (e.g.Irie et al., 2008Irie et al., , 2011;;Takashima et al., 2011) 1).In the retrieval, we applied the 460-490 nm standard fitting window of JM1, but we also used the 425-450 nm window, which is one of the most widely used for NO 2 retrieval (e.g.Boersma et al., 2004).We also performed an additional sensitivity analysis using a fitting window of 338-370 nm, and using different NO 2 cross-section data at 220 K (Vandaele et al., 1998).

NO 2 retrieval for three fitting windows
First, the NO 2 profile was retrieved using the JM1 algorithm with a standard fitting window of 460-490 nm. Figure 4 shows NO 2 concentrations for the 0-1 km layer, close to mainland Japan (the area with highest concentrations during the cruises).A clear land-ocean contrast is observed: during KY08-05, NO 2 concentrations were low over the ocean (<1 ppbv), with no clear diurnal variations; in contrast, concentrations were high (>1 ppbv) and with a clear diurnal variation near the coast or when in port (19-Introduction

Conclusions References
Tables Figures

Back Close
Full DOAS analysis using a 425-450 nm fitting window also revealed a clear land-ocean contrast in NO 2 DSCDs, with quite good agreement over the Japan region (Fig. 5a, c), but poor fitting results (Fig. 6c) and negative DSCD values, especially at low elevation angles at low latitudes (in the cause of Fig. 6c, the NO 2 DSCD was positive).In general, lower NO 2 DSCD was obtained at lower ELs (see the following paragraph and Fig. 7).
This resulted in sparse NO 2 volume mixing ratio (VMR) data after applying our vertical profile retrieval methods (Fig. 5).
To consider the effect of H 2 O in the fitting for the 425-450 nm window, the relationship between H 2 O DSCD and NO 2 DSCD was investigated (Fig. 7); there is a clear negative correlation between the two.At the same time the fitting residual is high for high H 2 O DSCD (not shown).These results suggest that the poor fitting at 425-450 nm is due in part to the H 2 O interference in the fitting.At lower elevation angles, the amount of H 2 O is generally high; thus, the fitting is generally poor.It should be noted that the H 2 O DSCD for 425-450 nm is consistent with that for 460-690 nm (figure not shown; a correlation coefficient (r) for EL = 3 • was 0.98).We also investigated other relationships and found a negative correlation between H 2 O DSCD and the Ring effect (not shown), suggesting that the Ring effect also contributed to the poor fitting (Fig. 6c).
We next employed an ultraviolet (UV) fitting window (338-370 nm), which is completely free from absorption by H 2 O. Here, we considered O 4 for 338-370 nm to derive NO 2 VMR.Although the retrieval errors were higher than in the case of 460-490 nm (Table 2; see Sect.4.2) due in part to the lower intensity (particularly in the morning/evening), there is generally good agreement between NO 2 (460-490 nm) and NO 2 (338-370 nm) concentrations for concentrations higher than ∼0.2 ppbv (Fig. 8).This result suggests that our MAX-DOAS can at least detect NO 2 concentrations as low as ∼0.2 ppbv for the 0-1 km layer.The results also indicate that the background level over the western Pacific and Indian Ocean during the cruise was less than ∼0.2 ppbv for the 0-1 km layer.The 460-490 nm standard fitting window of JM1 seems to be particularly useful for the retrieval of low NO 2 concentrations over the ocean and in H 2 O-rich areas.Introduction

Conclusions References
Tables Figures

Back Close
Full

Sensitivity analysis and error estimates
We next conducted a DOAS analysis using the NO 2 cross-section of 220 K as a sensitivity test of cross-section temperature dependence, revealing a strong correlation between NO 2 concentration using 220 and 294 K cross-sections (R = 0.99) but NO 2 concentration for 220 K systematically underestimates NO 2 for 294 K of ∼30 % (Fig. 9).
Because the actual temperature at 0-1 km is unlikely to have been as cold as 220 K, the sensitivity to the actual temperature variation would have been much smaller than that obtained from the DOAS analysis.Sanders (1996) and more recently Boersma et al. (2004) reported the temperature dependence of trace gas in tropospheric NO 2 retrieval.Subsequently, A. Richter et al. (personal communication, 2011) indicated that the retrieved NO 2 concentration using a 425-450 nm window shows a linear increase with applying a warmer cross-section, because the NO 2 cross-section at high temperatures is smaller than that at low temperatures.In their analysis, the temperature dependence is ∼0.36 % K −1 for the 425-450 nm fitting window, which is similar to our result of ∼30 % (294-220 K) −1 = 0.4 % K −1 , despite the different fitting window used in the two studies.Random and systematic errors for each MAX-DOAS measurement were estimated following Irie et al. (2011) and Takashima et al. (2011).The random error was estimated from the residual in the fitting of the NO 2 DSCD, and the systematic error was estimated assuming an additional 30 % change in aerosol optical depth (AOD), for which A box varies accordingly.
The estimated random and systematic errors in the NO 2 concentration for the 460-490 nm standard window during the KY09-01 cruise over the ocean were 0.009 ppbv (7.6 %) and 0.015 ppbv (12.6 %), respectively (Table 2).The total error was as small as ∼15 % (∼0.18 ppbv), even in the case of low background values over the remote ocean.These errors were of a similar order to those for the retrieval with the 220 K NO 2 cross-section.Note that the coldest and warmest temperatures below 1 km during KY09-01 were about −20 K (with respect to 294 K) near Japan in February and +3 K Introduction

Conclusions References
Tables Figures

Back Close
Full in the tropics.These correspond to systematic errors of −8 % and +1 %, respectively, based on the derived cross-section temperature dependence of 0.4 % K −1 .The range of this systematic error (∼9 % for 23 K) is of a smaller order than the total error.For UV, the errors were higher than those for 460-490 nm, due in part to lower intensity in our observation system, particularly in the morning/evening.The errors obtained near land (in the Japan region) for 460-490 nm are similar to those for 425-450 nm.
The errors over the ocean during KY09-01 are largely consistent with those reported previously for a remote island (Okinawa Island, Japan; for which the systematic and random errors were 12.8 % and 13.0 %, respectively) by Takashima et al. (2011), although these error values (i.e. of the present and previous studies) are much smaller than the background value over the ocean (<∼0.2 ppbv).Note that over the remote ocean, variability in NO 2 concentrations was less than ∼0.1 ppbv at 0-1 km, as assessed from a time series (not shown) and from Fig. 10.Therefore, this level (∼0.1 ppbv) may be close to the detection limit of NO 2 concentrations over the remote ocean.

NO 2 variations over the remote ocean
We next discuss NO 2 variations over the remote ocean retrieved by JM1 with a standard 460-490 nm fitting window for concentrations >0.2 ppbv.Over the remote ocean during cruise KY09-01, NO 2 concentrations were generally very low (<0.5 ppbv at 0-1 km; Fig. 10).The probability density function (PDF) for ≤1 ppbv during KY09-01 (Fig. 11) yields a mode of ∼ 0.1 ± 0.1 ppbv, which appears to correspond to the background level over the remote ocean (<0.2 ppbv).This value is similar to those obtained by direct measurements by airplane during the TRACE-P observation campaign, which yielded NO 2 concentrations of less than ∼0.05 ppbv over the western Pacific in the marine boundary layer during P3-B Flight 08 (data are available from the NASA TRACE-P Website at http://www-gte.larc.nasa.gov.Figure 11 shows a skewed distribution of NO 2 concentrations with some enhancements, including NO 2 concentrations (∼0.2-0.4 ppbv) higher than the background level, in which the fitting residual for NO 2 Introduction

Conclusions References
Tables Figures

Back Close
Full concentrations of 0.2-0.4ppbv was as small as ∼8.1 × 10 −4 (the median value; an example of the fitting is shown in Fig. 6b).These enhancements over the remote ocean are also apparent in Fig. 10, with spatial variability.In some cases, the air mass was affected by polluted air (e.g.southwest of Guam, where the air mass was advected from the direction of Guam), but this was infrequently observed.We also tested for the effect of emissions from the research vessel.To avoid such contamination, we analyzed wind data recorded on the vessel and removed potentially contaminated NO 2 data before repeating the analysis; however, no significant difference was observed in the PDF compared with the entire dataset.The enhancement may also reflect emissions from ships over the ocean or long-range rapid transport from polluted areas, as suggested by Takashima et al. (2011).Additional measurements over the ocean are required to quantify the background levels and the nature of spatial-temporal variations over the ocean.

Summary
NO 2 measurements by ship-borne MAX-DOAS with a compact/low-power spectrometer were conducted during two ocean cruises, around Japan and Japan-Bali (Indonesia)-Indian Ocean.The telescope was mounted on an active gimbal to ensure it was kept horizontal; it was successfully kept within ±0.2 • of the target elevation angle for ∼60 % of the time.
To test the sensitivity of the fitting window for NO 2 retrieval, focusing on low NO 2 concentration over the ocean, we considered windows of 425-450, 460-490, and 338-Introduction

Conclusions References
Tables Figures

Back Close
Full 425-450 nm at mid-latitudes.Because the NO 2 DSCD for 425-450 nm has a negative correlation with the H 2 O DSCD (Fig. 7), the poor fitting appears to be due to the H 2 O interference in the fitting.We analyzed a 338-370 nm fitting window, which is completely free from absorption by H 2 O, and found a good agreement between NO 2 VMRs retrieved from 460-490 nm and 338-370 nm, even at low-latitudes, at NO 2 VMRs higher than ∼0.2 ppbv (Fig. 8).Consequently, the 460-490 nm fitting window seems to be useful for the retrieval of low NO 2 concentrations over the ocean.
We also performed a sensitivity analysis using different NO 2 cross-sections (294 K and 220 K) for the 460-490 nm fitting window.The correlation between the two NO 2 concentrations was reasonably good, but NO 2 for 220 K systematically underestimates NO 2 for 294 K by ∼30 % (Fig. 9).
The results indicate that the background value of NO 2 over the western Pacific and Indian Ocean during the cruises was less than 0.2 ppbv, but occasional enhancement to values of ∼0.2-0.4 ppbv was often observed, exceeding the background level.On rare occasions, the air mass was affected by polluted air.Introduction

Conclusions References
Tables Figures

Back Close
Full  Full  Full Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Figure3shows an example of the pitch and roll angles of the vessel and the telescope unit.The telescope unit was installed on the starboard-side (roll) direction of the viewing azimuth angle.The figure shows that the orientation of the telescope unit was generally controlled well even when the roll and pitch angles of the vessel reached >2 degrees.During the KY08-05 and KY09-01 cruises, the telescope was kept within ±0.2• of the target elevation angle for ∼60 % of the time.In general, in the case of a Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Fig. 6 .
Fig. 6.Example of the result of nonlinear least squares spectral fitting for O 4 and NO 2 observed on 14 April 2009 on the R/V Kaiyo at 141.7 • E, 11.1 • N.For the NO 2 fittings, we used three fitting windows (460-490, 425-450, and 336-370 nm, as shown in (b), (c), and (d), respectively).The red line shows the cross-section scaled to the spectrum (black) measured by DOAS.The spectra are plotted in terms of differential optical depth from the reference spectrum (elevation angle of 70 • ).

2 Measurements 2.1 Two ocean cruises on the Japanese R/V K aiyo
• E, 35.45 • N) to Bari (Indonesia; 115.21 • E, 8.74 . We used NO 2 absorption cross-section data at 294 K of Vandaele et al. (1998), O 4 data of Hermans et al. (http://spectrolab.aeronomie.be/o2.htm),H 2 O data of the year 2004 edition of the High-Resolution Transmission (HITRAN) database (fitting windows and absorbers fitted in DOAS analysis are shown in Table

Table 1 .
Fitting windows and absorbers fitted in DOAS analysis.The representative wavelength for each target component is the cross-section-weighted mean wavelength over the fitting window.