Statistics of extreme events in climate models via coarse-scale simulations and machine learning correction operators based on nudged datasets

Alexis-Tzianni Charalampopoulos, Shixuan Zhang, Bryce Harrop, Ruby Leung, Themistoklis Sapsis

Keywords: machine learning, uncertainty quantification, climate, tropical cyclones

Abstract:

This work presents a systematic framework for improving the
predictions of statistical quantities for turbulent systems, with
a focus on correcting climate simulations obtained by coarsescale
models. Specifically, failure to incorporate all relevant
scales in climate simulations leads to discrepancies in the energy
spectrum as well as higher order statistics. While high
resolution simulations or reanalysis data are available, at least
for short periods, they cannot be directly used as training
datasets to machine learn a correction for the coarse-scale
climate model outputs, since chaotic divergence, inherent in
the climate dynamics, makes datasets from different resolutions
incompatible. To overcome this fundamental limitation
we employ coarse-resolution model (here we employ Energy
Exascale Earth System Model, E3SM) simulations nudged
towards high quality climate realizations, here in the form
of ERA5 reanalysis data. The nudging term is sufficiently
small to not “pollute” the coarse-scale dynamics over short
time scales, but also sufficiently large to keep the coarse-scale
simulations “close” to the ERA5 trajectory over larger time
scales. The result is a “compatible” pair of the ERA5 trajectory
(used as output training data) and the weakly nudged
coarse-resolution E3SM output that is used as input training
data to machine learn a correction operator. We emphasize
that the nudging step is used only for the training phase.
Once training is complete, we perform free-running coarsescale
E3SM simulations without nudging and use those as
input to the machine-learned correction operator to obtain
high-quality (corrected) outputs. The model is applied to atmospheric
climate data with the purpose of predicting global
and local statistics of various quantities of a time-period of a
decade. Using ERA5 datasets that are not employed for training,
we demonstrate that the produced datasets from the MLcorrected
coarse E3SM model have statistical properties that
closely resemble the observations. In particular, the corrected
coarse-scale E3SM output closely captures the non-Gaussian
statistics of quantities such as temperature, wind speed and
humidity, as well as the frequency of occurrence of extreme
events, such as tropical cyclones and atmospheric rivers. We
present thorough comparisons and discuss limitations of the
approach.