Nudging and Model Sensitivities
Because current global climate models contain a large number of uncertain parameters in representing processes unresolved by the computational mesh, quantifying these uncertain parameters is important for future climate projections. The chaotic nature of the atmospheric noise, however, often obscures the atmospheric response to changes in values of parameters (signals) used in global climate models. In the effort to overcome the effect of natural noise, conducting long or numerous simulations can be computationally expensive, or even impractical, for high-resolution models. The nudging method constraining meteorological fields toward weather reanalysis or baseline simulations may be used to reduce the effect of natural variability and distinguish signal from noise in shorter simulations.
This study further confirmed the conclusion from an earlier study that nudging temperature can have non-negligible detrimental impacts in the tropical regions. This study also provided a cautionary note to the climate modeling community that wind nudging with a 6-hour relaxation time scale can substantially modify the model’s sensitivity to parameter perturbations in regions with strong circulation feedback.
Even though nudging may cause unintentional side effects due to the adding of the artificial forcing term to the physical forcing term, the question of whether nudging can be used to quantify model sensitivities needs to be answered. Climate scientists at the Department of Energy’s Pacific Northwest National Laboratory evaluated the use of the nudging technique in quantifying model sensitivities in precipitation and cloud forcing. The method was evaluated in this study by comparing the traditional free-running simulations with the simulations that were conducted by nudging meteorology fields (wind and temperature). The study found that in the case of strong perturbation in convection, the nudged simulations cause undesired damaging effects because of the strong interactions between convection and resolved winds, while in the case of weak perturbation in large-scale cloud formation, the nudged simulations within a shorter simulation period can reasonably capture the signals obtained in long term free-running simulations.