Accurate precipitation simulations for various climate scenarios are critical for understanding and predicting the impacts of climate change. This study employs a Cycle-generative adversarial network (CycleGAN) to improve global 3-hour-average precipitation fields predicted by a coarse grid (200~km) atmospheric model across a range of climates, morphing them to match their statistical properties with reference fine-grid (25~km) simulations. We evaluate its performance on both the target climates and an independent ramped-SST simulation. The translated precipitation fields remove most of the biases simulated by the coarse-grid model in the mean precipitation climatology, the cumulative distribution function of 3-hourly precipitation, and the diurnal cycle of precipitation over land. These results highlight the potential of CycleGAN as a powerful tool for bias correction in climate change simulations, paving the way for more reliable predictions of precipitation patterns across a wide range of climates.

Coarse-grid weather and climate models rely particularly on parameterizations of cloud fields, and coarse-grained cloud fields from a fine-grid reference model are a natural target for a machine-learned parameterization. We machine-learn the coarsened-fine cloud properties as a function of coarse-grid model state in each grid cell of NOAA’s FV3GFS global atmosphere model with 200 km grid spacing, trained using a 3 km fine-grid reference simulation with a modified version of FV3GFS. The ML outputs are coarsened fine fractional cloud cover and liquid and ice cloud condensate mixing ratios, and the inputs are coarse model temperature, pressure, relative humidity, and ice cloud condensate. The predicted fields are skillful and unbiased, but somewhat under-dispersed, resulting in too many partially-cloudy model columns. When the predicted fields are applied diagnostically (offline) in FV3GFS’s radiation scheme, they lead to small biases in global-mean top-of-atmosphere (TOA) and surface radiative fluxes. An unbiased global mean TOA net radiative flux is obtained by setting to zero any predicted cloud with grid cell mean cloud fraction less than a threshold of 6.5%; this does not significantly degrade the ML prediction of cloud properties. The diagnostic, ML-derived radiative fluxes are  far more accurate than those obtained with the existing cloud parameterization in the nudged coarse-grid model, as they leverage the accuracy of the fine-grid reference simulation’s cloud properties. 

The use of machine learning (ML) for the online correction of coarse-resolution atmospheric models has proven effective in reducing biases in near-surface temperature and precipitation rate. However, this often introduces biases in the upper atmosphere and improvements are not always reliable across ML-corrective models trained with different random seeds. Furthermore, ML corrections can feed back on the baseline physics of the atmospheric model and produce profiles that are outside the distribution of samples used in training, leading to low confidence in the predicted corrections. This study introduces the use of a novelty detector to mask the predicted corrections when the atmospheric state is deemed out-of-sample. The novelty detector is trained on profiles of temperature and specific humidity in a semi-supervised fashion using samples from the coarsened reference fine-resolution simulation. Offline, the novelty detector determines more columns to be out-of-sample in simulations which are known, using simple metrics like mean bias, to drift further from the reference simulation. Without novelty detection, corrective ML leads to the development of undesirably large climate biases for some ML random seeds but not others. Novelty detection deems about 21% of columns to be novelties in year-long simulations. The spread in the root mean square error (RMSE) of time-mean spatial patterns of surface temperature and precipitation rate across a random seed ensemble is sharply reduced when using novelty detection. In particular, the random seed with the worst RMSE is improved by up to 60% (depending on the variable) while the best seed maintains its low RMSE.

Can the current successes of global machine learning-based weather simulators be generalized beyond two-week forecasts to stable and accurate multiyear runs? The recently developed AI2 Climate Emulator (ACE) suggests this is feasible, based upon 10-year simulations trained on a realistic global atmosphere model using a grid spacing of approximately 110~km and forced by a repeating annual cycle of sea-surface temperature. Here we show that ACE, without modification, can be trained to emulate another major atmospheric model, EAMv2, run at a comparable grid spacing for at least ten years with similarly small climate biases. ACE accurately reproduces EAMv2’s frequency distribution of daily-mean precipitation, its time-mean spatial pattern of precipitation, and its space-time structure of tropical precipitation, including the Madden-Julian Oscillation. Moreover, ACE’s climate biases with respect to EAMv2 are substantially smaller than EAMv2’s own biases compared to the observed historical average surface precipitation rate and top-of-atmosphere radiative fluxes.

One approach to improving the accuracy of a coarse-grid global climate model is to add machine-learned state-dependent corrections to the prognosed model tendencies, such that the climate model evolves more like a reference fine-grid global storm-resolving model (GSRM). Our past work demonstrating this approach was trained with short (40-day) simulations of GFDL’s X-SHiELD GSRM with 3 km global horizontal grid spacing. Here, we extend this approach to span the full annual cycle by training and testing our machine learning (ML) using a new year-long GSRM simulation. Our corrective ML models are trained by learning the state-dependent tendencies of temperature and humidity and surface radiative fluxes needed to nudge a closely related 200~km grid coarse model, FV3GFS, to the GSRM evolution. Coarse-grid simulations adding these learned ML corrections run stably for multiple years. Compared to a no-ML baseline, the time-mean spatial pattern errors with respect to the fine-grid target are reduced by 6-25% for land surface temperature and 9-25% for land surface precipitation. The ML-corrected simulations develop other biases in climate and circulation that differ from, but have comparable amplitude to, the no-ML baseline simulation.

Bretherton et al. (2022, https://doi.org/10.1029/2021MS002794) demonstrated a successful approach for using machine learning (ML) to help a coarse-resolution global atmosphere model with real geography (a ~200 km version of NOAA’s FV3GFS) evolve more like a fine-resolution model. This study extends that work for application in multiple climates and multi-year ML-corrected simulations. Here four fine-resolution (~25 km) two-year reference simulations are run using FV3GFS with climatological sea surface temperatures perturbed uniformly by -4 K, 0 K, +4 K, and +8 K. A dataset of state-dependent corrective tendencies is then derived through nudging the ~200 km model to the coarsened state of the fine-resolution simulations in each climate. Along with the surface radiative fluxes, the nudging tendencies of temperature and specific humidity are machine-learned as functions of the column state. ML predictions for the fluxes and corrective tendencies are applied in 5.25 year ~200 km resolution simulations in each climate, and improve the spatial pattern errors of land precipitation by 17% to 30% and land surface temperature by 20% to 23% across the four climates. The ML has a neutral impact on the pattern error of oceanic precipitation.

Global atmospheric ‘storm-resolving’ models with horizontal grid spacing of less than 5~km resolve deep cumulus convection and flow in complex terrain. They promise to be reference models that could be used to improve computationally affordable coarse-grid global climate models across a range of climates, reducing uncertainties in regional precipitation and temperature trends. Here, machine learning of nudging tendencies as functions of column state is used to correct the physical parameterization tendencies of temperature, humidity, and optionally winds, in a real-geography coarse-grid model (FV3GFS with a 200~km grid) to be closer to those of a 40-day reference simulation using X-SHiELD, a modified version of FV3GFS with a 3~km grid. Both simulations specify the same historical sea-surface temperature fields. This methodology builds on a prior study using a global observational analysis as the reference. The coarse-grid model without machine learning corrections has too little cloud, causing too much daytime heating of land surfaces that creates excessive surface latent heat flux and rainfall. This bias is avoided by learning downwelling radiative flux from the fine-grid model. The best configuration uses learned nudging tendencies for temperature and humidity but not winds. Neural nets slightly outperform random forests. Forecasts of 850 hPa temperature gain 18 hours of skill at 3–7 day leads and time-mean precipitation patterns are improved 30\% by applying the ML correction. Adding machine-learned wind tendencies improves 500 hPa height skill for the first five days of forecasts but degrades time-mean upper tropospheric temperature and zonal wind patterns thereafter.