Science Motivation
Extra-Tropical Cyclones (ETCs) are a key mechanism for the transport of energy and moisture in the atmosphere. ETCs have an impact on water supply and natural hazards (flooding caused because of high precipitation, storm surges, high winds, etc). The characterization of how ETC activity (both frequency and intensity) will change in the future under global warming scenarios is an important line of investigation.

Methods
In this work, we apply the Toolkit for Extreme Climate Analytics (TECA) for extracting ETC patterns in massive climate archives. We implement a pattern detection procedure outlined by Wang, et al. that identifies local pressure minima with a threshold on pressure gradient. We have successfully scaled this code to process the entire CMIP-5 archive in one shot on 750,000 cores on the Mira, BG/Q system. We extract storm track counts, and track densities from the resulting analysis. The output of the climate model simulations is compared against observed extra-tropical cyclone behaviour using the multiple linear regression analysis commonly used in detection and attribution studies. Various metrics of frequency and location are analysed, both on global and ocean-basin scales.

Datasets
We process all CMIP-5 models under the historicalNAT, historical and rcp8.5 scenarios. We also process major reanalysis products (ERA-Interim, 20CR, NCEP2, NCEP-CFSR-0.5 and NCEP-CFSR-2.5).