Ozone Imager | V2 Upd

A neural network correction for surface albedo heterogeneity (training data from MODIS BRDF products). 4. Validation and Performance 4.1 In-Orbit Testing Deployed on a SpaceX Transporter-12 rideshare (January 2026). After 6 months of commissioning:

where (I) is Earth radiance, (I_0) solar irradiance, (\mu) air mass factor, (\Omega) total column ozone (Dobson Units, DU), and (\sigma) ozone absorption cross-section. A weighted least-squares fit over the five UV channels retrieves (\Omega) per pixel, with cloud screening using the 360 nm band (Rayleigh+surface reflectance). ozone imager v2

[ R(\lambda) = \fracI(\lambda)I_0(\lambda) \cdot \exp\left(-\mu \cdot \Omega \cdot \sigma(\lambda, T, P)\right) ] A neural network correction for surface albedo heterogeneity

Authors: A. Chen¹, M. Kapoor², L. Rodriguez¹ Affiliations: ¹Laboratory for Atmospheric Physics, ETH Zürich; ²Remote Sensing Division, NOAA After 6 months of commissioning: where (I) is

| Metric | OIv2 | OMI (Aura) | |--------|------|-------------| | Nadir resolution | 1.5 km | 13×24 km | | Swath width | 1200 km | 2600 km | | TCO RMSE vs. Dobson | 2.8% | 3.5% | | Signal-to-noise ratio (312 nm) | 450 | 180 | OIv2 resolved a 120 DU depletion zone within a pyrocumulonimbus plume over Tasmania – a feature smoothed out in OMI data. The 1.5 km imagery revealed filamentary ozone transport structures consistent with WRF-Chem simulations. 5. Discussion OIv2 demonstrates that fine-scale ozone variability is ubiquitous, particularly near urban centers, biomass burning, and mountain waves. The ability to resolve <5 km features challenges the assumption of spatial homogeneity in satellite ozone validation. Future work includes assimilation into GEOS-Chem and operational UV index forecasting.