ClimDetect: A Benchmark Dataset for Climate Change Detection and Attribution
This provides a benchmark dataset for climate scientists to improve detection and attribution methods, though it is incremental as it standardizes existing data rather than solving the core detection problem.
The authors tackled the challenge of detecting and attributing climate change signals by introducing ClimDetect, a standardized dataset of 1.17M daily climate snapshots from CMIP6 simulations and reanalysis datasets, which enhances model accuracy and enables consistent comparisons across studies.
Detecting and attributing temperature increases driven by climate change is crucial for understanding global warming and informing adaptation strategies. However, distinguishing human-induced climate signals from natural variability remains challenging for traditional detection and attribution (D&A) methods, which rely on identifying specific "fingerprints" -- spatial patterns expected to emerge from external forcings such as greenhouse gas emissions. Deep learning offers promise in discerning these complex patterns within expansive spatial datasets, yet the lack of standardized protocols has hindered consistent comparisons across studies. To address this gap, we introduce ClimDetect, a standardized dataset comprising 1.17M daily climate snapshots paired with target climate change indicator variables. The dataset is curated from both CMIP6 climate model simulations and real-world observation-assimilated reanalysis datasets (ERA5, JRA-3Q, and MERRA-2), and is designed to enhance model accuracy in detecting climate change signals. ClimDetect integrates various input and target variables used in previous research, ensuring comparability and consistency across studies. We also explore the application of vision transformers (ViT) to climate data -- a novel approach that, to our knowledge, has not been attempted before for climate change detection tasks. Our open-access data serve as a benchmark for advancing climate science by enabling end-to-end model development and evaluation. ClimDetect is publicly accessible via Hugging Face dataset repository at: https://huggingface.co/datasets/ClimDetect/ClimDetect.