Detection and Attribution of Climate Change Using a Neural Network

A new detection and attribution method is presented and applied to the global mean surface air temperature (GSAT) from 1900 to 2014. The method aims at attributing the climate changes to the variations of greenhouse gases, anthropogenic aerosols, and natural forcings. A convolutional neural network (CNN) is trained using the simulated GSAT from historical and single‐forcing simulations of 12 climate models. Then, we perform a backward optimization with the CNN to estimate the attributable GSAT changes. Such a method does not assume additivity in the effects of the forcings. The uncertainty in the attributable GSAT is estimated by sampling different starting points from single‐forcing simulations and repeating the backward optimization. To evaluate this new method, the attributable GSAT changes are also calculated using the regularized optimal fingerprinting (ROF) method. Using synthetic non‐additive data, we first find that the neural network‐based method estimates attributable changes better than ROF. When using GSAT data from climate model, the attributable anomalies are similar for both methods, which might reflect that the influence of forcing is mainly additive for the GSAT. However, we found that the uncertainties given both methods are different. The new method presented here can be adapted and extended in future work, to investigate the non‐additive changes found at the local scale or on other physical variables. In order to design effective adaptation policies, it is essential to have reliable estimates of the effect of anthropogenic activities on the climate. For that purpose, a new attribution method based on a neural network is designed and evaluated. The method estimates the past global mean surface air temperatures anomalies caused by the changes in the greenhouse gases concentration, the variation of anthropogenic aerosols, and the variations driven by naturally occurring phenomena. To build this estimation, the data from observations and climate models are used. This methodology is compared with another state‐of‐the‐art method. The results of both methods are evaluated and discussed. The proposed method provides better estimations in the case of large non‐additivity of the causes of climate change and can be applied to other physical variables or at the regional scale. In the case of the global mean surface air temperature, the method presented provides estimation similar to other methods. We present a non linear method based on neural netw