The objective of this project is to perform a process-based multi-timescale diagnostic of CMIP5 and CMIP6-era Earth System Models using a weather-typing dynamical approach. The proposed
work focuses on how accurately extreme rainfall events, both wet and dry, are represented over the US in CMIP5/6 models. Although the project will emphasize the present and next generation
of NOAA/GFDL models, to guarantee robustness other available models will also be diagnosed. This project will develop process-informed cross-timescale tools to diagnose CMIP5/6 historical and climate-change projections over North America based on the methodology of large-scale recurrent, persistent weather types (WTs), also known as large-scale meteorological patterns (LSMPs). These regimes provide a dynamically informative intermediary between the large- scale drivers of climate variability and change from sub-seasonal to decadal timescales, and mid- latitude high-impact weather events, through the mechanism of synoptic control. The proposed work will provide an urgently-needed process-level understanding on rainfall extremes in CMIP5/6 simulations, and develop standard metrics that model developers and users can apply to these models easily. These will allow model developers to quickly assess the impacts of changes in parameters, and will enable users to better assess confidence levels on projections of return intervals of extreme rainfall events.
The proposed work will build on recent previous work by the team demonstrating the effectiveness of the approach to both (1) cross-timescale diagnostics of rainfall over North and South America, and (2) diagnose GCM model performance in a suite of GFDL forecast models.
Expected deliverables of the project include (a) general open-source software package to perform weather-type based cross-timescale diagnostics of climate models, including new process-based metrics that can be added to the MAPP diagnostics Task Force framework, and documentation for the software package; and (b) an online “diagnostic atlas” containing the process-based metrics (e.g., WT spatial patterns and frequencies of occurrences at different timescales, extreme rainfall composite analysis for different thresholds, anomaly correlations to climate drivers) available via the IRI Data Library.