First results of the CINDI-2 semi-blind MAX-DOAS intercomparison K. Kreher1, M. Van Roozendael2, F. Hendrick2, A. Apituley3, U. Frieß4, J Lampel4, A. Piters3, A. Richter5, T. Wagner6,

and the CINDI-2 Team

1BK Scientific, Mainz, Germany; 2IASB-BIRA, Brussels, Belgium; 3KNMI, de Bilt, The Netherlands; 4Institute of Environmental Physics, University of Heidelberg, Germany; 5Institute of Environmental Physics, University of Bremen, Germany; 6MPI for Chemistry, Mainz, Germany

The CINDI-2 Campaign The second Cabauw Intercomparison campaign for Nitrogen Dioxide measuring Instruments (CINDI-2) took place at the Cabauw Experimental Site for Atmospheric Research (CESAR, Utrecht area, The Netherlands) from 25 August until 7 October 2016.

The goals of this intercomparison campaign are:

To characterise and better understand the differences between a large number of MAX-DOAS and DOAS instruments and analysis methods.

To contribute to a harmonisation of the measurement settings and retrieval methods.

To support the creation of high-quality ground-based data sets (e.g. to provide reliable long-term time series for trend analysis and satellite data validation).

To apply recommendations from previous intercomparisons by Roscoe et al. (1999), Vandaele et al. (2005) and Roscoe et al. (2010):

• Synchronisation of measurements by applying a very detailed measurement protocol • Exact alignment of the elevation angles using regular horizon scans & lamp measurements • More clearly prescribed viewing directions (7 for 2d instruments, see Figure 1) • Emphasis on profiling techniques

Figure 1: Azimuthal directions for the 2D-MAXDOAS instruments (North is 0° and 287° is the main direction without any obstructions down to the horizon).

Semi-blind MAX-DOAS intercomparison (12-28 September 2016)

As part of CINDI-2, a semi-blind intercomparison was held under the umbrella of NDACC (Network for the Detection of Atmospheric Composition Change) over 17 days in September 2016.

A strict measurement protocol was applied to ensure optimal time and spatial colocation (common viewing geometries and synchronization of measurements to within 1 min).

The scanning accuracy of the MAX-DOAS instruments was also systematically monitored by means of active light source measurements and horizon scans.

Daily meetings and reports were undertaken following the semi-blind comparison protocol as laid out by NDACC. During the campaign period, the measurements for the previous day were submitted and statistical analyses and comparisons of the submitted slant column data sets were discussed in the daily meetings.

Final data sets were submitted on 18th of October 2016 for a post-campaign analysis. The approach adopted is largely motivated by previous intercomparisons, in particular CINDI and MADCAT. It is based on the systematic analysis of regression plots between each measurement and a reference value, which for CINDI-2 was obtained through calculation of the median from all available measurements. Considering the large number of instruments included in the CINDI-2 intercomparison exercise (36), the median value is expected to be a good proxy for the unknown “true” value.

Participating groups & their instruments

Fig. 2 shows the container layout at the campaign site were all the MAX-DOAS instruments were setup. An example of slant column time series measured during the intercomparison is shown in Fig. 3.

Table 2: Overview of the participating institutes (2. column) and technical details for the 36 instruments (4.-8. column) which took part in the CINDI-2 semi-blind intercomparison.

Horizon scan analysis

Fig. 4: The left panel shows the horizon scans measured on 24 September in the visible wavelength region. The right panel shows an overview of the average horizon elevation and of the field of view determined from the scans for all instruments.

Certification matrix

Fig. 7: Certification matrix for all 36 instruments and 8 data products for MAX-DOAS and 4 data products for zenith-sky mode. White indicates when data sets were not measured.

Criteria for performance assessment

Fig. 6: Examples of the assessment plots for MAX-DOAS and zenith-sky data. The top left panel shows the regression analysis for MAX-DOAS NO2 in the visible wavelength region and the panel beneath shows the corresponding overview plot (bottom left). The bottom middle and right panels show the same overview plots for MAX-DOAS O4 measured in the UV and zenith-sky NO2 measured in the visible region. The performance criteria shown in the top middle and right panel were applied (green shading). Red dots mark the parameters of the instruments which don’t meet the criteria. The orange ovals mark instruments which fulfil the majority of the assessment criteria while the red ovals mark instruments which don’t. The summary of all the assessments is shown below in Fig. 7.

Summary & outlook Achievements of the CINDI-2 semi-blind intercomparison campaign:

• The application of a detailed and prescriptive acquisition protocol did lead to noticeably improved synchronisation of the measure-ments.

• Regular horizon scans & lamp measurements clearly improved the alignment of the elevation angles.

• Greater emphasis was also placed on profiling techniques (see posters listed in the panel left below).

• The assessment is based on previously applied methods which were further refined and substantially extended (see Fig. 6 & 7).

• Based on the performance criteria, a certification table was developed for MAX-DOAS and zenith sky data products (Fig. 7).

The purpose of the certification process is to endorse groups & their instruments for contributing to the NDACC data base and/or to satellite validation programmes (in particular S5P and the future atmospheric Sentinel missions).

• A CINDI-2 workshop was held at KNMI, De Bilt, The Netherlands during 3-5 April 2017 to discuss results to date with a second workshop planned for Feb. 2018.

• A substantial list of planned publications was circulated. • Further data analysis is underway using a) sequential references and b) direct sun measurements.

Acknowledgments The CINDI-2 campaign and this study have been funded by the ESA projects CINDI-2 and FRM4DOAS (contract No. 4000118533/16/I-Sbo and 4000118181/16/I-EF, respectively). The CINDI-2 campaign was hosted by the Royal Netherlands Meteorological Institute (KNMI), de Bilt, The Netherlands.

List of companion posters related to the CINDI-2 campaign:

• The Second Cabauw Intercomparison Campaign for Nitrogen Dioxide Measuring Instruments — CINDI-2 — Overview, Arnoud Apituley et al., EGU-2017-10177 (oral)

• Comparison of MAX-DOAS profiling algorithms during CINDI-2 - Part 1: aerosols, Udo Friess et al., EGU2017-8368 (poster X5.380)

• Comparison of MAX-DOAS profiling algorithms during CINDI-2 - Part 2: trace gases, Francois Hendrick et al., EGU2017-8484 (poster X5.381)

• The ESA FRM4DOAS project: Towards a quality-controlled MAXDOAS Centralized Processing System, Francois Hendrick et al. , EGU2017-11999 (poster X5.395)

Green label All criteria fulfilled Orange label Majority of criteria fulfilled Red label Majority of criteria not fulfilled Ranking: In each category (green, orange, red), groups are sorted by increasing values of the median DOAS fit RMS

Data product Typical wavelengths

NO2 (VIS range) 425 – 490 nm

NO2 (VIS range small) 411 – 445 nm

NO2 (UV range) 338 – 370 nm

O4 (VIS range) 425 – 490 nm

O4 (UV range) 338 – 370 nm

HCHO 336.5 – 359 nm

O3 (Chappuis bands) 450 – 520 nm

O3 (Huggins bands) 320 – 340 nm

Relative intensity 340, 380, 440, 500 nm

# Instrument ID Type Spectral range Resolution Detector type T° 1 BIRA 4 MAXDOAS (2D) 300-390 nm/ 400-560 nm 0.4/ 0.6 nm CCD -50°

2 AUTH 3 PHAETON (2D) 300-450 nm 0.4 nm CCD 5°

3 AIOFM 1 MAXDOAS (2D) 290-380 nm 0.4 nm CCD -30°

4 IUPH 19 EnviMes (2D) 300-460 nm/ 440-580 nm 0.6/ 0.5 nm CCD Room T°

5 IUPB 18 MAXDOAS (2D) 305-390 nm/ 405-580 nm 0.5/ 0.9 nm CCD -35°

6 IUPB 37 I-DOAS (2D) 400-580 nm 0.5 nm CCD -30°

7 BOKU 6 MAXDOAS (2D) 405-580 nm 0.9 nm CCD -30°

8 CMA 7 Hoffmann (1D) 300-450 nm 0.7 nm PDA Room T°

9 CMA 8 Hoffmann (1D) 400-710 nm 0.7 nm PDA Room T°

10 CHIBA-U 9 MAXDOAS (1D) 310-515 nm 0.4 nm CCD 0-40°

11 CSIC 10 MAXDOAS (1D) 300-500 nm 0.5 nm CCD Room T°

12 CU-Boulder 11 MAXDOAS (2D) 325-470 nm/ 430-680 nm 0.7/ 1.2 nm CCD -30°

13 CU-Boulder 12 MAXDOAS (1D) 300-465 nm/ 380-490 nm 0.8/ 0.5 nm CCD -30°/ 0°

14 DLR-USTC 13 EnviMes (2D) 300-460 nm/ 450-600 nm 0.6/ 0.6 nm CCD Room T°

15 DLR-USTC 14 EnviMes (2D) 300-460 nm/ 450-600 nm 0.6/ 0.6 nm CCD Room T°

16 IISERM 16 Hoffmann (1D) 320-470 nm 1.0 nm CCD Room T°

17 INTA 17 MAXDOAS (2D) 400-550 nm 0.5 nm CCD -20°

18 KNMI 21 Hoffmann (1D) 290-430 nm 0.5 nm PDA Room T°

19 KNMI 22 Hoffmann (1D) 400-600 nm 0.6 nm PDA Room T°

20 KNMI 23 Pandora (2D) 285-530 nm 0.6 nm CCD 20°

21 LUFTB 26 Pandora-2S (2D) 280-540 nm 0.6 nm CCD 15°

22 LUFTB 260 Pandora-2S (2D) 400-900 nm 1.1 nm CCD 15°

23 LUFTB 27 Pandora-2S (2D) 280-540 nm 0.6 nm CCD 15°

24 LUFTB 270 Pandora-2S (2D) 400-900 nm 1.1 nm CCD 15°

25 MPIC 28 Tube-DOAS (1D) 315-475 nm 0.6 nm CCD 10°

26 NASA 31 Pandora (2D) 285-530 nm 0.6 nm CCD 20°

27 NASA 32 Pandora (2D) 285-530 nm 0.6 nm CCD 20°

28 NIWA 29 EnviMes (1D) 305-460 nm/ 410-550 nm 0.7 nm CCD 20°

29 NIWA 30 MAXDOAS (1D) 290-365 nm/ 400-460 nm 0.6 nm CCD -20°

30 NUST 33 Hoffmann (1D) 320-465 nm 0.7 nm CCD Room T°

31 LMU-MIM 35 EnviMes 2D 300-460 nm/ 450-600 nm 0.6 nm CCD 20°

32 U-Toronto 36 MAXDOAS (2D) 300-500 nm 0.5 nm CCD 20°

33 AMOIAP 2 2-port DOAS (1D) 420-490 nm 0.5 nm CCD -40°

34 LATMOS 24 SAOZ (ZS) 270-640 nm 1.3 nm PDA Room T°

35 LATMOS 25 Mini-SAOZ (ZS) 270-820 nm 0.7 nm CCD Room T°

36 BSU 5 MARSB (1D) 300-500 nm 0.4 nm CCD -40°

Horizon scans: pointing accuracy

Each group’s ability to measure the products listed in Table 1 is assessed as part of the semi-blind intercomparison and results are shown in (Figures 4 - 7).

Table 1: Data products included in the semi-blind intercomparison exercise.

The 5 criteria for the assessment of MAX-DOAS data sets:

- Duty rate (percent number of valid points): 50% or higher - Pointing accuracy (horizon elevation): <=0.5° - Slope, offset and RMS from the regression

Fig. 5: The left panel shows an overview of the pointing accuracy of each instrument derived from the horizon scans. The right panel shows the relative difference between the individual data sets against the median of all data sets for O4.

Product Bias (%) Offset (molec/cm2)

RMS (molec/cm2)

NO2vis (*) 5 1.5 E15 8.0 E15

NO2visSmall 5 1.5 E15 8.0 E15

NO2uv 6 2.0 E15 1.0 E16

O4vis 5 0.7 E42 3.0 E42

O4uv 6 0.8 E42 4.0 E42

HCHO 10 5.0 E15 2.0 E16

O3uv 4 0.2 E18 1.0 E18

O3vis (*) 4 1.0 E18 4.0 E18

The 4 criteria for the zenith-sky case:

- Slope, offset and RMS from the regression analysis

- Duty rate: 50% or higher

- Same limits used as for MAX-DOAS

Fig. 2: CINDI-2 intercomparison site Fig. 3: O4 and NO2 slant columns measured at 30o elevation angle, HCHO measured at 5o elevation.

For some instruments, regular horizon scans helped to improve data quality quite dramatically during the first view days (Fig. 5, left panel). The corresponding improvement in the data quality can, for example, be seen in O4 slant columns measured in the UV wavelength range at a 1o elevation angle (Fig 5., right panel).

- (*) NO2 and O3 criteria adapted from previous NDACC campaigns - Other criteria derived from analysis of the distribution of the bias, offset and RMS parameters