Learning Surface Ozone From Satellite Columns (LESO): A Regional Daily Estimation Framework for Surface Ozone Monitoring in China

Continuously monitoring surface ozone (O
₃
) spatial distribution and forecasting its variations are beneficial to improving air quality and ensuring public health in China, although achieving this goal faces challenges from currently available observations and retrieval techniques. Hence, we introduce a coupled surface O
₃
estimation framework (LESO) to address these challenges by integrating ground-level observing networks and satellite remote sensing. LESO features easy-to-use deep learning algorithms, independence on chemical transportation models (CTMs), and consistent performance using data from different satellites. LESO includes a deep forest 21 (DF21) model to interpolate O
₃
concentration by learning spatial patterns and a long short-term memory (LSTM) model to forecast O
₃
concentration by learning data from the past. We used sites of city-level
in situ
networks as the control sites to manifest short-distance O
₃
transportation. Satellite-based observations of O
₃
precursor indicators were incorporated to capture O
₃
photochemical reactions. DF21 explained a larger fraction of O
₃
variability (90%) with a mean bias error (MBE) of smaller than
$1~\mu \text {g/m}^{3}$
. We also investigated the impact of the number of training sites on the DF21 performance, which suggested that five training sites could ensure a good DF21 performance for the most areas (
$R^{2} > 0.85$
and bias
$< 2~\mu \text {g/m}^{3}$
). The forecast O
₃
concentration via LSTM showed a good and stable agreement (
$R^{2}\approx 0.85$
and bias
$< 5~\mu \text {g/m}^{3}$
) with ground-based measurements for 8-, 24-, 28-, and 72-h time periods, respectively. Overall, LESO aims to bring convenient functionality and reliable surface O
₃
estimates for broad users.