We investigate the representation of individual supercells and intriguing tornado-like vortices in a simplified, locally-refined global atmosphere model. The model, featuring grid stretching, can locally enhance the model resolution and economically reach the cloud-resolving scale. Given an unstable sheared environment, the model can simulate supercells realistically, with a near-ground vortex and funnel cloud at the center of a rotating updraft reminiscent of a tornado. An analysis of the vorticity budget suggests that the updraft core of the supercell tilts environmental horizontal vorticity into the tornado-like vortex. The updraft also acts to amplify the vortex through vertical stretching. Results suggest that the simulated vortex is dynamically similar to observed tornadoes and modeling studies at much higher horizontal resolution.

Though tropical cyclone (TC) models have been routinely evaluated against track and intensity observations, little work has been performed to validate modeled TC wind fields over land. In this paper, we present a simple framework for evaluating simulated low-level inland winds with in-situ observations and existing TC structure theory. The Automated Surface Observing Systems, Florida Coastal Monitoring Program, and best track data are used to generate a theory-driven wind profile that reasonably represents the observed radial distribution of TC wind speeds. We quantitatively and qualitatively evaluated the modeled inland TC wind fields, and described the model performance with a set of simple indicators. The framework was used to examine the performance of a high-resolution two-way nested Geophysical Fluid Dynamics Laboratory model on recent U.S. landfalling TCs. Results demonstrate the capacity of using this framework to assess the modeled TC low-level wind field in the absence of dense inland observations.

We present the System for High-resolution prediction on Earth-to-Local Domains (SHiELD), an atmosphere model coupling the nonhydrostatic FV3 Dynamical Core to a physics suite originally taken from the Global Forecast System. SHiELD is designed to demonstrate new capabilities within its components, explore new model applications, and to answer scientific questions through these new functionalities. A variety of configurations are presented, including short-to-medium-range and subseasonal-to-seasonal (S2S) prediction, global-to-regional convective-scale hurricane and contiguous US precipitation forecasts, and global cloud-resolving modeling. Advances within SHiELD can be seamlessly transitioned into other Unified Forecast System (UFS) or FV3-based models, including operational implementations of the UFS. Continued development of SHiELD has shown improvement upon existing models. The flagship 13-km SHiELD demonstrates steadily improved large-scale prediction skill and precipitation prediction skill. SHiELD and the coarser-resolution S-SHiELD demonstrate a superior diurnal cycle compared to existing climate models; the latter also demonstrates 28 days of useful prediction skill for the Madden-Julian Oscillation. The global-to-regional nested configurations T-SHiELD (tropical Atlantic) and C-SHiELD (contiguous United States) shows significant improvement in hurricane structure from a new tracer advection scheme and promise for medium-range prediction of convective storms, respectively.

Intense convection (updrafts exceeding 10 m∙s-1) plays an essential role in severe weather and Earth’s energy balance. Despite its importance, how the global pattern of intense convection changes in response to warmed climates remains unclear, as simulations from traditional climate models are too coarse to simulate intense convection. Here we take advantage of a kilometer-scale global storm resolving model and conduct year-long simulations of a control run, forced by analyzed sea surface temperature (SST), and one with a 4-K increase in SST for comparison. Comparisons show that the increased SST enhances the frequency of intense convection globally with large spatial and seasonal variations. Increases in the intense convection frequency do not necessarily reflect increases in convective available potential energy (CAPE). Results are also compared with traditional climate model projections. Changes in the spatial pattern of intense convection are associated with changes in planetary circulation.

The 13 km SHiELD (System for High-resolution prediction on Earth-to-Local Domains) global model that is under development at the Geophysical Fluid Dynamics Laboratory (GFDL) and runs in near real time, produced outstanding tropical cyclone track forecasts during the 2021 Atlantic hurricane season, compared to both the upgraded National Weather Service Global Forecast System (GFSv16), the Hurricane Weather Research and Forecasting (HWRF) model and the European Centre for Medium-Range Weather Forecast Integrated Forecasting System (IFS). SHiELD’s average track forecast errors were 10% and 15% less than the GFSv16 and HWRF, respectively, for the 3-5 day forecast lead times. SHiELD’s track forecast skill was comparable to the National Hurricane Center’s official forecast at several forecast lead times, and approached 70% skill relative to the Climatology and Persistence Model (CLIPER) at 3 and 4 days. Similar improvements were found in the western Pacific basin in 2021, with improvements also seen in the eastern Pacific at days 4 and 5. Improved performance was also found in the 2019 Atlantic hurricane season, with neutral performance in 2020, when SHiELD was run retrospectively from the GFSv16 initial conditions. Distribution of the spatial errors and biases showed that in both the 2021 Atlantic hurricane season and the previous two years, the largest track forecast errors from both SHiELD and GFSv16 occurred in the subtropical eastern Atlantic, associated with a distinct northeast bias. Analysis indicated that some of the excessive north bias in the GFSv16 is associated with lower geopotential height fields compared to those in SHiELD.

This paper documents the third version of the GFDL cloud microphysics scheme (GFDL MP v3) used in the System for High-resolution prediction on Earth-to-Local Domains (SHiELD) model. Compared to the GFDL MP v2, the GFDL MP v3 is entirely reorganized, optimized, and modularized by functions. In addition, the particle size distribution (PSD) of all cloud categories is redefined to mimic the latest observations, and the cloud condensation nuclei (CCNs) are calculated from the MERRA2 aerosol data. The GFDL MP has been redesigned so all processes use the redefined PSD to ensure overall consistency and easily permit introductions of new PSDs and microphysical processes. Analyses gathered from simulations by SHiELD with selected configurations are examined. Compared to the GFDL MP v2, the GFDL MP v3 significantly improves the predictions of geopotential height, air temperature, and specific humidity in the Troposphere, as well as the high, middle and total cloud fractions and the liquid water path. With the more realistic PSD implemented in GFDL MP v3, the predictions of geopotential height in the Troposphere, low and total cloud fractions are further improved. Furthermore, using climatological aerosol data to calculate CCNs leads to even better predictions of geopotential height, air temperature, and specific humidity in the Troposphere, high and middle cloud fractions, as well as the liquid and ice water paths. However, the upgrade of the GFDL MP shows little impact on the precipitation prediction. Degradation due to the scheme upgrade is also addressed and discussed to guide the future GFDL MP development.

We present the characteristics of rotating convective updrafts in the 2021 version of GFDL’s Experimental System for High-resolution prediction on Earth-to-Local Domains (X-SHiELD), a kilometer scale global storm resolving model (GSRM). Rotation is quantified using 2–5 km Updraft Helicity (UH) in a year-long integration forced by analyzed SSTs. Updrafts with UH magnitudes above 50 m2/s2; are common over the mid-latitude continents, especially in the warm seasons where they are associated with severe weather but are also common over most tropical ocean basins. In nearly all areas cyclonically rotating convection dominates, with larger UH values increasingly preferring cyclonic rotation. The ratio of cyclonic to anticyclonic updrafts is largest in the subtropical and mid-latitude oceans and is slightly lower over mid-latitude continents. The ratio of cyclonic to anticyclonic updrafts can be substantively explained by the mean storm-relative helicity (SRH) in convective regions, indicating the importance for environmental controls on the sense of storm rotation, although internal storm dynamics also plays a role in the generation of anticyclonic updrafts.

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.

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. The figure shows maps of 30-day time-mean precipitation pattern difference from the fine-grid reference for prognostic simulations: (a) 200 km baseline(no machine learning correction) (b) Using random forest correction and (c) neural net correction for temperature. humidity and surface radiation corrections. RMSE is the root mean squared precipitation difference from the reference, which is 30% less for the two machine-learning corrected simulations compared to the baseline. (d) Bar charts of the land-mean, ocean-mean and global-mean precipitation biases for these three configurations, showing the machine-learning corrected simulations remove a high bias of land surface precipitation in the 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.

We propose an integrated dynamics-physics coupling framework for weather and climate-scale models. Each physical parameterization would be advanced on its natural time scale, revised to include a moist thermodynamic relationship, and finally integrated into the relevant components of the dynamical core. We show results using a cloud microphysics scheme integrated within the dynamical core of the GFDL SHiELD weather model to demonstrate the promise of this concept. We call it the in-line microphysics as it is in-lined within the dynamical core. Statistics gathered from one year of weather forecasts show significantly better prediction skills when the model is upgraded to use the in-line microphysics. However, we do find that some biases are degraded with the in-line microphysics. The in-line microphysics also shows larger-amplitude and higher-frequency variations in cloud structures within a tropical cyclone than the traditionally-coupled microphysics. Finally, we discuss the prospects for further development of this integrated dynamics-physics coupling.