Fig. 3: The positive and negative correlations between aerosol optical parameters and NCCN, as well as the importance of aerosol optical parameters. | npj Climate and Atmospheric Science

Fig. 3: The positive and negative correlations between aerosol optical parameters and NCCN, as well as the importance of aerosol optical parameters.

From: Interpretable ensemble learning unveils main aerosol optical properties in predicting cloud condensation nuclei number concentration

Fig. 3

In predicting NCCN, the SHAP values for each feature are denoted as follows: ae represent SGP, GUC, ASI, ENA, and MOS. The letters B, G, and R following these parameters represent measurements at three specific wavelengths: Blue (464 nm), Green (529 nm), and Red (648 nm). For instance, \({\sigma }_{{\rm{sp}}}\)_B represents the \({\sigma }_{{\rm{sp}}}\) at the blue wavelength. The AE parameter with letters indicates that it is computed from the total scattering at two wavelengths. For example, AE_BR denotes the AE calculated using the blue and red total scattering values. The mean of the absolute values of the SHAP values indicates the importance of each variable to NCCN prediction. In each plot, feature importance is arranged from top to bottom, with the width of the bars indicating the sample size. The color of the points represents the value of the corresponding variable, with warmer colors indicating higher values and cooler colors indicating lower values. Full variable abbreviations can be found in Supplementary Tables 15.

Back to article page