Abstract. The Paris Agreement of December 2015 stated a goal to pursue efforts to keep global temperatures below 1.5 °C above pre-industrial levels and well below 2 °C. The IPCC was charged with assessing climate impacts at these temperature levels, but fully coupled equilibrium climate simulations do not currently exist to inform such assessments. In this study, we produce a set of scenarios using a simple model designed to achieve long term 1.5 °C and 2 °C temperatures in a stable climate. These scenarios are then used to produce century scale ensemble simulations using the Community Earth System Model, providing impact-relevant long term climate data for stabilization pathways at 1.5 °C and 2 °C levels and an overshoot 1.5 °C case, which are freely available to the community. Here we describe the design of the simulations and key aspects of their impact-relevant climate response. Exceedance of historical record temperature occurs with 60 percent greater frequency in the 2 °C climate than in a 1.5 °C climate aggregated globally, and with twice the frequency in equatorial and arid regions. Extreme precipitation intensity is statistically significantly higher in a 2.0 °C climate than a 1.5 °C climate in several regions. The model exhibits large differences in the Arctic which is ice-free with a frequency of 1 in 3 years in the 2.0 °C scenario, and only 1 in 40 years in the 1.5 °C scenario.
Conformal Prediction (CP) is a method that can be used for complementing the bare predictions produced by any traditional machine learning algorithm with measures of confidence. CP gives good accuracy and confidence values, but unfortunately it is quite computationally inefficient. This computational inefficiency problem becomes huge when CP is coupled with a method that requires long training times, such as Neural Networks. In this paper we use a modification of the original CP method, called Inductive Conformal Prediction (ICP), which allows us to construct a Neural Network confidence predictor without the massive computational overhead of CP. The method we propose accompanies its predictions with confidence measures that are useful in practice, while still preserving the computational efficiency of its underlying Neural Network.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.