Intercomparison of four airborne imaging DOAS systems for tropospheric NO2 mapping - the AROMAPEX campaign

We present an intercomparison study of four airborne imaging DOAS instruments, dedicated to the retrieval and high resolution mapping of tropospheric nitrogen dioxide (NO 2 ) vertical column densities (VCDs). The AROMAPEX campaign took place in Berlin, Germany in April, 2016 with the primary objective to test and intercompare the performance of experimental airborne imagers. The imaging DOAS instruments were operated simultaneously from two manned aircraft, performing synchronised flights: APEX (VITO/BIRA-IASB) was operated from DLR's DO-228 D-CFFU aircraft at 6.2 km altitude, while AirMAP (IUP-Bremen), SWING (BIRA-IASB) and SBI (TNO/TU Delft/KNMI) were operated from the FUB Cessna 207T D-EAFU at 3.1 km. Two synchronised flights took place on 21 April 2016. NO 2 slant columns were retrieved by applying differential optical absorption spectroscopy (DOAS) in the visible wavelength region and converted to VCDs by the computation of appropriate air mass factors (AMFs). Finally, the NO 2 VCDs were georeferenced and mapped at high spatial resolution. For the sake of harmonising the different data sets, efforts were made to agree on a common set of parameter settings, AMF LUT and gridding algorithm. The NO 2 horizontal distribution, observed by the different DOAS imagers, shows very similar spatial patterns. The NO 2 field is dominated by two large plumes related to industrial compounds, crossing the city from west to east. The major highways A100 and A113 are also identified as line sources of NO 2 . Retrieved NO 2 VCDs range between 1 × 10 15  molec cm −2 upwind of the city and 20 × 10 15  molec cm −2 in the dominant plume, with a mean of 7.3 ± 1.8 × 10 15  molec cm −2 for the morning flight and between 1 and 23 × 10 15  molec cm −2 with a mean of 6.0 ± 1.4 × 10 15  molec cm −2 for the afternoon flight. The mean NO 2 VCD retrieval errors are in the range of 22 to 36 % for all sensors. The four data sets are in good agreement with Pearson correlation coefficients better than 0.9, while the linear regression analyses show slopes close to unity and generally small intercepts.