To better understand the regional changes in summertime temperatures across the conterminous United States (CONUS), we adopt a recently developed machine learning framework that can be used to reveal the timing of emergence of forced climate signals from the noise of internal climate variability. Specifically, we train an artificial neural network (ANN) on seasonally averaged temperatures across the CONUS and then task the ANN to output the year associated with an individual map. In order to correctly identify the year, the ANN must therefore learn time-evolving patterns of climate change amidst the noise of internal climate variability. The ANNs are first trained and tested on data from large ensembles and then evaluated using observations from a station-based data set. To understand how the ANN is making its predictions, we leverage a collection of ad hoc feature attribution methods from explainable artificial intelligence (XAI). We find that anthropogenic signals in seasonal mean minimum temperature have emerged by the early 2000s for the CONUS, which occurred earliest in the Eastern United States. While our observational timing of emergence estimates are not as sensitive to the spatial resolution of the training data, we find a notable improvement in ANN skill using a higher resolution climate model, especially for its early twentieth century predictions. Composites of XAI maps reveal that this improvement is linked to temperatures around higher topography. We find that increases in spatial resolution of the ANN training data may yield benefits for machine learning applications in climate science.
Eischeid, Jon K., Martin P Hoerling, Xiao-Wei Quan, Arun Kumar, Joseph J Barsugli, Zachary M Labe, Kenneth E Kunkel, Carl J Schreck, David R Easterling, , John Uehling, and Xiangdong Zhang, October 2023: Why has the summertime central U.S. warming hole not disappeared?Journal of Climate, 36(20), DOI:10.1175/JCLI-D-22-0716.17319-7336. Abstract
A cooling trend in summer (May–August) daytime temperatures since the mid-twentieth century over the central United States contrasts with strong warming of the western and eastern United States. Prior studies based on data through 1999 suggested that this so-called warming hole arose mainly from internal climate variability and thus would likely disappear. Yet it has prevailed for two more decades, despite accelerating global warming, compelling reexamination of causes that in addition to natural variability could include anthropogenic aerosol–induced cooling, hydrologic cycle intensification by greenhouse gas increases, and land use change impacts. Here we present evidence for the critical importance of hydrologic cycle change resulting from ocean–atmosphere drivers. Observational analysis reveals that the warming hole’s persistence is consistent with unusually high summertime rainfall over the region during the first decades of the twenty-first century. Comparative analysis of large ensembles from four different climate models demonstrates that rainfall trends since the mid-twentieth century as large as observed can arise (although with low probability) via internal atmospheric variability alone, which induce warming-hole-like patterns over the central United States. In addition, atmosphere-only model experiments reveal that observed sea surface temperature changes since the mid-twentieth century have also favored central U.S cool/wet conditions during the early twenty-first century. We argue that this latter effect is symptomatic of external radiative forcing influences, which, via constraints on ocean warming patterns, have likewise contributed to persistence of the U.S. warming hole in roughly equal proportion to contributions by internal variability. These results have important ramifications for attribution of extreme events and predicting risks of record-breaking heat waves in the region.
