ClimateNet: an expert-labeled open dataset and deep learning architecture for enabling high-precision analyses of extreme weather

Abstract

Abstract. Identifying, detecting, and localizing extreme weather events is a crucial first step in understanding how they may vary under different climate change scenarios. Pattern recognition tasks such as classification, object detection, and segmentation (i.e., pixel-level classification) have remained challenging problems in the weather and climate sciences. While there exist many empirical heuristics for detecting extreme events, the disparities between the output of these different methods even for a single event are large and often difficult to reconcile. Given the success of deep learning (DL) in tackling similar problems in computer vision, we advocate a DL-based approach. DL, however, works best in the context of supervised learning – when labeled datasets are readily available. Reliable labeled training data for extreme weather and climate events is scarce. We create “ClimateNet” – an open, community-sourced human-expert-labeled curated dataset that captures tropical cyclones (TCs) and atmospheric rivers (ARs) in high-resolution climate model output from a simulation of a recent historical period. We use the curated ClimateNet dataset to train a state-of-the-art DL model for pixel-level identification – i.e., segmentation – of TCs and ARs. We then apply the trained DL model to historical and climate change scenarios simulated by the Community Atmospheric Model (CAM5.1) and show that the DL model accurately segments the data into TCs, ARs, or “the background” at a pixel level. Further, we show how the segmentation results can be used to conduct spatially and temporally precise analytics by quantifying distributions of extreme precipitation conditioned on event types (TC or AR) at regional scales. The key contribution of this work is that it paves the way for DL-based automated, high-fidelity, and highly precise analytics of climate data using a curated expert-labeled dataset – ClimateNet. ClimateNet and the DL-based segmentation method provide several unique capabilities: (i) they can be used to calculate a variety of TC and AR statistics at a fine-grained level; (ii) they can be applied to different climate scenarios and different datasets without tuning as they do not rely on threshold conditions; and (iii) the proposed DL method is suitable for rapidly analyzing large amounts of climate model output. While our study has been conducted for two important extreme weather patterns (TCs and ARs) in simulation datasets, we believe that this methodology can be applied to a much broader class of patterns and applied to observational and reanalysis data products via transfer learning.

Increases in Future AR Count and Size: Overview of the ARTMIP Tier 2 CMIP5/6 Experiment

“…Recent work involving manual identification of ARs by experts (Prabhat et al, 2021;O'Brien, Risser, et al, 2020) suggests that the spread in AR algorithm behavior is linked to differences in opinion about what does and does not constitute an AR. O' Brien, Risser, et al (2020) show that this spread in subjective opinion projects directly on to quantitative differences in the sign of the correlation coefficient between an El Niño index and global AR count.…”

AtmoDist: Self-supervised Representation Learning for Atmospheric Dynamics

“…We also want to explore other downstream tasks, e.g. the classification and prediction of hurricanes (Prabhat et al, 2021) or extreme events (Racah et al, 2017). Interesting would also be to explore transfer learning for AtmoDist, e.g.…”

Atmospheric river, a term encompassing different meteorological patterns

“…Therefore, there is no need to include any-subjective-threshold criteria for the detection method. This facilitates objective detection, and allows for comprehensive comparison between different large climate datasets (e.g., Muszynski et al, 2019;Prabhat et al, 2021;Xu et al, 2020). These techniques have also proven to be useful for improving ARs forecasts as shown in Chapman et al (2019).…”