Toward Optimal Simulation Strategies for Understanding Model Biases and Sensitivities

Researchers carried out a comprehensive evaluation of various nudging methods in the E3SM atmosphere model.

The Science

For the development and application of numerical weather and climate models, it is often useful to constrain a numerical experiment so that the evolution of the meteorological conditions follows a specific pathway. One of the techniques to apply such constraints is called nudging. A team of scientists led by researchers at the U.S. Department of Energy’s Pacific Northwest National Laboratory performed and analyzed a large set of sensitivity experiments carried out with the Energy Exascale Earth System Model (E3SM) atmosphere model to identify implementations of nudging that can provide sufficient constraints without severely interfering with the simulations.

The Impact

Results show that carefully configured nudging can allow the model to reproduce the characteristic evolution of observed weather events while maintaining the statistical properties of the unconstrained climate simulations. Such nudged simulations can facilitate process-based evaluations of model fidelity and allow for detailed analysis without requiring years of simulation data, thus providing a valuable experimentation strategy for the development of high-resolution models.


Quantifying, attributing, and reducing biases in global climate models is a very difficult task. Process interactions lead to nonlinear variabilities in the observed and simulated atmospheric states; this causes noise that can hinder signal detection, hide compensating errors, and complicate the comparison between model results and observational data.

Constraining the simulated atmospheric circulation using methods like nudging can help alleviate these difficulties, providing an unprecedented opportunity to understand model biases and sensitivities at shorter time scales under specific meteorological conditions. However, if a model is constrained too hard, the simulated long-term mean results might not be representative of its own climate.

In this work, the researchers performed and analyzed sensitivity experiments with the E3SM atmosphere model (EAM) to identify best implementations of nudging that can provide skillful atmospheric hindcasts without severely interfering with the simulations. They showed that when the prescribed meteorological conditions are temporally interpolated to the model time to constrain the EAM’s horizontal winds at each time step, a nudged simulation can reproduce the characteristic evolution of the observed weather events (especially in middle and high latitudes) as well as the model’s long-term climatology. Compared to its predecessor model used in an earlier study, EAM is not as sensitive to temperature nudging but remains very sensitive to humidity nudging. Constraining humidity substantially improves the correlation between simulated and observed tropical precipitation but also leads to large changes in the long-term statistics of the simulated precipitation, clouds, and aerosol life cycles.

Principal Investigator(s)

Ruby Leung
Pacific Northwest National Laboratory

Hui Wan
Pacific Northwest National Laboratory


This research was supported by the Energy Exascale Earth System Model (E3SM) project and the “ACME-SM: A Global Climate Model Software Modernization Surge” project, funded by the Office of Biological and Environmental Research (BER), within the U.S. Department of Energy (DOE) Office of Science. This research used high-performance computing resources from the Pacific Northwest National Laboratory (PNNL) Research Computing (RC); the National Energy Research Scientific Computing Center (NERSC), a DOE Office of Science user facility supported under Contract DE-AC02-05CH11231; and the Laboratory Computing Resource Center at Argonne National Laboratory, provided by the BER Earth System Modeling program.


Sun, J., Zhang, K., Wan, H. et al. “Impact of nudging strategy on the climate representativeness and hindcast skill of constrained EAMv1 simulations.” Journal of Advances in Modeling Earth Systems 11(12), 3911–3933 (2019). [DOI:10.1029/2019MS001831]