Using Storm Tracking to Evaluate E3SM
Comparison of extreme precipitation between IMERG (OBS) and E3SM hindcast. (a/g) 99th percentile values are based on hourly IMERG/E3SM precipitation in 2011. (b–f) Accumulated precipitation, which contribute to the top 1% precipitation hours from each storm types from IMERG (OBS), and (h–l) ones from E3SM Day 2 hindcast.
Wen-Ying Wu, Lawrence Livermore National Laboratory
Heavy precipitation, often associated with weather phenomena such as tropical cyclones, extratropical cyclones (ETCs), atmospheric rivers (ARs), and mesoscale convective systems (MCSs), can cause significant socio-economic loss. It remains challenging for conventional Earth System Models (ESMs) to realistically simulate the magnitude and timing of extreme precipitation events, which are associated with a variety of storm types, for reasons that include biases in model parameterizations and insufficient spatial resolutions. In this study, scientists at Lawrence Livermore National Laboratory, with collaborators, applied atmospheric feature trackers to quantify the contributions of these storm types in observational data sets and climate model short-range hindcasts to address questions of (1) how much global annual mean precipitation is associated with these four major heavy-precipitating storm systems? (2) what are the spatiotemporal characteristics of storm-associated precipitation? and (3) How well does the DOE Energy Exascale Earth System Model (E3SM) simulate the storm-associated precipitation?
This phenomenon-based approach provides a better understanding of precipitation characteristics and can lead to enhanced model evaluation by revealing underlying problems in model physics related to precipitation processes associated with heavy-precipitating storms.
The analysis from observations show that these four storm types account for 67% of global annual mean precipitation and 82% of the top 1% precipitation extremes, with MCSs mainly over the tropics and ARs and ETCs over the mid-latitudes. The percentage of precipitation contributions from these storms also shows strong seasonality over many geographical locations. The tracking results were further applied to the E3SM short-range hindcasts to evaluate the performance of these storms. The evaluation shows that E3SM, with ∼1° resolution, significantly underestimates all storm-associated precipitation totals and extremes, especially for MCSs in the tropics. The analysis also suggests that the model fails to capture the correct mean diurnal phases and amplitude of MCS precipitation, which usually dominates the peak hours in the mean diurnal cycle of precipitation.