Despite recent improvements, many global climate models (GCMs) still show strong biases
in the representation of tropical cyclone (TC) activity, especially its frequency and intensity.
These GCM biases limit the reliability of TC sub-seasonal and seasonal predictions and future
projections. A lack of diagnostics that could provide insights into process-level errors in the
model representation of TCs has slowed model improvement.
We propose a project focused on the diagnosis of process-level errors in the model
representation of TC genesis and intensification. Our proposed project will build upon the
success of our ongoing project, during which we have developed process-oriented diagnostics
for TCs by adapting diagnostics that were originally developed for the Madden-Julian Oscillation
(MJO) and convective self-aggregation. No widely-accepted such process-based diagnostics for
global TC modeling existed before our current project. The diagnostics we developed have been
applied to a limited number of high- and low-resolution GCMs, which has allowed us to identify
processes that are key to TC intensification and genesis. In the proposed work, we will extend
the development and implementation of these diagnostics so that they may be fully utilized as
a community tool and may guide NOAA model development:
• First, we will examine key processes associated with TC genesis and intensification in
long term satellite observations and reanalysis products, using multiple observations
to quantify uncertainty. This will provide a “reference” version of our diagnostics
against which the model representation of the same processes can be validated. This
process-level evaluation against observations is crucial to model improvement.
• Second, while the diagnostics developed have been applied to a limited number of
model simulations that have been obtained in an opportunity-based manner, the
diagnostics and comparison with observational results need to be applied to a wider
group of models. We will take advantage of the upcoming model intercomparison
projects (CMIP6/HighResMIP/PRIMAVERA) to evaluate and identify biases in the
• Third, we will use the new NOAA model (GFDL AM4/CM4), which has been already
developed and shows decent capability in simulating TCs globally, to perform targeted
experiments guided by the results of our ongoing project and the proposed
observational analyses. The result of the targeted simulations will help improve the
NOAA model, and also would provide useful information to other modeling groups;
• Lastly, we will translate the existing codes and scripts to open source languages and
implement it into the NOAA MDTF diagnostics package to maximize the accessibility
of the diagnostics.
This project fits well within the MAPP Competition entitled “Addressing Key Issues in
CMIP6-era Earth System Models”, by developing and using process-oriented diagnostics to
identify the source of GCM biases in TC simulation and by providing paths toward the model
improvement. This advancement in our understanding of TC simulation will be extremely
valuable for improving the next generation of climate models, which is vital for making robust
projections of future TC activity and its impacts; a key component of NOAA’s long-term goals.