An Enhanced Geographically and Temporally Weighted Neural Network for Remote Sensing Estimation of Surface Ozone

Surface ozone (O ₃ ) pollution is a severe environmental problem that endangers human health. It is necessary to obtain high spatiotemporal resolution O ₃ data to provide support for pollution monitoring and prevention. For this purpose, this study makes comprehensive use of remote sensing data, reanalysis data, and ground station observations and develops an enhanced geographically and temporally weighted neural network (EGTWNN) model to acquire high spatial and temporal resolutions of O ₃ data. The EGTWNN model is nested by two neural networks (NNs). The first NN automatically learns the spatiotemporal proximity relationship to obtain spatiotemporal weights for the samples, and the spatiotemporal weights are then inputted into the second NN to conduct weighted modeling of the relationship between O ₃ and influencing variables. The contribution of the proposed model is that the first NN replaces the traditional empirical weighting method and represents the spatiotemporal proximity relationship more accurately to improve estimation accuracy. Results indicate that the cross-validation $R^{2}$ and the root mean square error (RMSE) of EGTWNN are 0.81 and $21.24 \ \mu \text{g}$ /m ³ , respectively, which are increased by 0.02 and decreased by $\sim 1 \ \mu \text{g}$ /m ³ relative to those of the traditional empirical weighting method-based geographically and temporally weighted NN model. The results also show that, compared with the geographically and temporally weighted regression model, the proposed model achieves superior performance. In addition, the spatiotemporal weights obtained by the first NN of EGTWNN are highly consistent with those obtained by the traditional empirical weighting method, indicating that the results of NNs are highly interpretable.