Find in Library

<p>Airborne imaging remote sensing is increasingly used to map the spatial distribution of nitrogen dioxide (<span class="inline-formula">NO₂</span>) in cities. Despite the small ground-pixel size of the sensors, the measured <span class="inline-formula">NO₂</span> distributions are much smoother than one would expect from high-resolution model simulations of <span class="inline-formula">NO₂</span> over cities. This could partly be caused by 3D radiative transfer effects due to observation geometry, adjacency effects and effects of buildings. Here, we present a case study of imaging a synthetic <span class="inline-formula">NO₂</span> distribution for a district of Zurich using the 3D MYSTIC (Monte carlo code for the phYSically correct Tracing of photons In Cloudy atmospheres) solver of the libRadtran radiative transfer library. We computed <span class="inline-formula">NO₂</span> slant column densities (SCDs) using the recently implemented 3D-box air mass factors (3D-box AMFs) and a new urban canopy module to account for the effects of buildings. We found that for a single ground pixel (50 m <span class="inline-formula">×</span> 50 m) more than 50 % of the sensitivity is located outside of the pixel, primarily in the direction of the main optical path between sun, ground pixel, and instrument. Consequently, <span class="inline-formula">NO₂</span> SCDs are spatially smoothed, which results in an increase over roads when they are parallel to the optical path and a decrease otherwise. When buildings are included, <span class="inline-formula">NO₂</span> SCDs are reduced on average by 5 % due to the reduced sensitivity to <span class="inline-formula">NO₂</span> in the shadows of the buildings. The effects of buildings also introduce a complex pattern of variability in SCDs that would show up in airborne observations as an additional noise component (about 12 <span class="inline-formula">µmol m⁻²</span>) similar to the magnitude of typical measurement uncertainties. The smearing of the SCDs cannot be corrected using 1D-layer AMFs that assume horizontal homogeneity and thus remains in the final <span class="inline-formula">NO₂</span> map. The 3D radiative transfer effects by including buildings need to be considered to compute more accurate AMFs and to reduce biases in <span class="inline-formula">NO₂</span> vertical columns obtained from high-resolution city-scale <span class="inline-formula">NO₂</span> remote sensing.</p>