Currently, in the modeling of various atmospheric pollutants, the simulation of independent trace gases is constrained by the insufficient resolution of key remote sensing products, resulting in insufficient simulation reliability.In this study, spatial sampling and parameter convolution are combined to optimize LightGBM by utilizing ground observations, remote sensing products, meteorological data, assistance data, and random ID.Through the above techniques and an sequentialsimulation of air pollutants, we produce seamless daily 1-km-resolution products of SO2for most parts of China from 2015 to 2018.Through random sampling, random site sampling, area-specific validation, comparisons of different models, and a cross-sectional comparison of different studies, we verified that our simulations of the spatial distribution of multiple atmospheric pollutants are reliable and effective.
The data used in this study include daily ground monitoring data for SO2 in China. Additionally, remote sensing data,meteorological data, and auxiliary data are used.
A general machine learning model for multiple pollutants based on random ID, spatial adoption, parameter convolution, and other methods is used to improve the consideration of multiple factors in the prediction of changes in atmospheric pollutant concentrations and optimize estimates of the spatial distributions of pollutants. We evaluate the model results using CV and visual qualitative analysis. LightGBM,LSTM, and RF-Ps are compared to our model to assess its performance. Finally, SHAP is used to try to interpret the output of the model.
The data quality is good.
This work is licensed under
CC BY 4.0 (Creative Commons Attribution 4.0 International License).
CSTR:11738.11.NCDC.ZENODO.DB4167.2024
10.5281/zenodo.7312179
