Understanding cirrus clouds using explainable machine learning
It addresses uncertainties in climate modeling for researchers and policymakers, but is incremental as it applies existing machine learning methods to new data in this domain.
This study tackled the problem of predicting cirrus cloud properties, such as ice water content and ice crystal number concentration, using meteorological and aerosol data, achieving an R² of 0.49 and identifying specific thresholds like a dust particle concentration of 2×10⁻⁴ mg m⁻³ that affects predictions.
Cirrus clouds are key modulators of Earth's climate. Their dependencies on meteorological and aerosol conditions are among the largest uncertainties in global climate models. This work uses three years of satellite and reanalysis data to study the link between cirrus drivers and cloud properties. We use a gradient-boosted machine learning model and a Long Short-Term Memory (LSTM) network with an attention layer to predict the ice water content and ice crystal number concentration. The models show that meteorological and aerosol conditions can predict cirrus properties with $R^2 = 0.49$. Feature attributions are calculated with SHapley Additive exPlanations (SHAP) to quantify the link between meteorological and aerosol conditions and cirrus properties. For instance, the minimum concentration of supermicron-sized dust particles required to cause a decrease in ice crystal number concentration predictions is $2 \times 10^{-4}$ mg m\textsuperscript{-3}. The last 15 hours before the observation predict all cirrus properties.