Fig. 4: Remote sensing modeling of surface PM2.5 based on GTW-RF on December 1, 2019. | npj Climate and Atmospheric Science

Fig. 4: Remote sensing modeling of surface PM2.5 based on GTW-RF on December 1, 2019.

From: Deriving PM2.5 from satellite observations with spatiotemporally weighted tree-based algorithms: enhancing modeling accuracy and interpretability

Fig. 4

a Estimated PM2.5 concentrations across China; b PM2.5 measurements of ground sites; c Number of training samples for each location; d Distribution of the most important variable, 1–8 represent AOD, PS, UW, VW, T, HCHO, NO2, and CAMS-PM2.5; e AOD-PM2.5 and NO2-PM2.5 relationships within region #A; f AOD-PM2.5 relationships within region #B (except for region #C) and region #C. r stands for correlation coefficient, * and *** means the 95% and 99.9% confidence level, respectively.

Back to article page