Detecting Climate Signals Using Explainable AI With Single‐Forcing Large Ensembles. (14th June 2021)

• Record Type: Journal Article
• Authors: Labe, Zachary M.; Barnes, Elizabeth A.
• Is Part Of: Journal of advances in modeling earth systems. Volume 13:Number 6(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-06-14
• Subjects: explainable AI -- neural networks -- climate models -- large ensembles -- climate forcings -- climate change; Geological modeling -- Periodicals; Climatology -- Periodicals; Geochemical modeling -- Periodicals; 551.5011
• DOI: 10.1029/2021MS002464
• Languages: English
• ISSNs: 1942-2466

Abstract

Abstract: It remains difficult to disentangle the relative influences of aerosols and greenhouse gases on regional surface temperature trends in the context of global climate change. To address this issue, we use a new collection of initial‐condition large ensembles from the Community Earth System Model version 1 that are prescribed with different combinations of industrial aerosol and greenhouse gas forcing. To compare the climate response to these external forcings, we adopt an artificial neural network (ANN) architecture from previous work that predicts the year by training on maps of near‐surface temperature. We then utilize layer‐wise relevance propagation (LRP) to visualize the regional temperature signals that are important for the ANN's prediction in each climate model experiment. To mask noise when extracting only the most robust climate patterns from LRP, we introduce a simple uncertainty metric that can be adopted to other explainable artificial intelligence (AI) problems. We find that the North Atlantic, Southern Ocean, and Southeast Asia are key regions of importance for the neural network to make its prediction, especially prior to the early‐21st century. Notably, we also find that the ANN predictions based on maps of observations correlate higher to the actual year after training on the large ensemble experiment with industrial aerosols held fixed to 1920 levels. This work illustrates the sensitivity of regional temperature signals to changes in aerosol forcing in historical simulations. By using explainable AI methods, we have the opportunity to improve our understanding of (non)linear combinations of anthropogenic forcings in state‐of‐the‐art global climate models. Plain Language Summary: Using a machine learning method called artificial neural networks (ANNs), we explore how human‐caused climate drivers can affect regional patterns of surface temperature. Here we use a climate model with different combinations of greenhouse gases and industrial aerosols (particles in the atmosphere) to understand their influence on climate change and variability. By employing visualization tools to see how the ANN makes its predictions, we can better recognize how these climate drivers influence global temperature in the past, present, and future. For instance, we find that aerosols emitted in the 20th century and early 21st century have influenced global warming temperature trends in some areas of the world, such as over the North Atlantic Ocean. Machine learning accompanied by new visualization methods have the potential to bring new insights into understanding the effects of global climate change in observations and models. Key Points: Using explainable artificial intelligence methods with artificial neural networks (ANN) reveals climate patterns in large ensemble simulations Predictions from an ANN trained using a large ensemble without time‐evolving aerosols show the highest correlation with actual observations A metric is proposed for quantifying the uncertainty of an ANN visualization method that extracts signals from different external forcings