Climate-invariant machine learning.

Projecting climate change is a generalization problem: We extrapolate the recent past using physical models across past, present, and future climates. Current climate models require representations of processes that occur at scales smaller than model grid size, which have been the main source of model projection uncertainty. Recent machine learning (ML) algorithms hold promise to improve such process representations but tend to extrapolate poorly to climate regimes that they were not trained on. To get the best of the physical and statistical worlds, we propose a framework, termed "climate-invariant" ML, incorporating knowledge of climate processes into ML algorithms, and show that it can maintain high offline accuracy across a wide range of climate conditions and configurations in three distinct atmospheric models. Our results suggest that explicitly incorporating physical knowledge into data-driven models of Earth system processes can improve their consistency, data efficiency, and generalizability across climate regimes.

Muted extratropical low cloud seasonal cycle is closely linked to underestimated climate sensitivity in models.

A large spread in model estimates of the equilibrium climate sensitivity (ECS), defined as the global mean near-surface air-temperature increase following a doubling of atmospheric CO2 concentration, leaves us greatly disadvantaged in guiding policy-making for climate change adaptation and mitigation. In this study, we show that the projected ECS in the latest generation of climate models is highly related to seasonal variations of extratropical low-cloud fraction (LCF) in historical simulations. Marked reduction of extratropical LCF from winter to summer is found in models with ECS > 4.75 K, in accordance with the significant reduction of extratropical LCF under a warming climate in these models. In contrast, a pronounced seasonal cycle of extratropical LCF, as supported by satellite observations, is largely absent in models with ECS < 3.3 K. The distinct seasonality in extratropical LCF in climate models is ascribed to their different prevailing cloud regimes governing the extratropical LCF variability.

High-resolution precipitation monitoring with a dense seismic nodal array.

Accurate precipitation monitoring is crucial for understanding climate change and rainfall-driven hazards at a local scale. However, the current suite of monitoring approaches, including weather radar and rain gauges, have different insufficiencies such as low spatial and temporal resolution and difficulty in accurately detecting potentially destructive precipitation events such as hailstorms. In this study, we develop an array-based method to monitor rainfall with seismic nodal stations, offering both high spatial and temporal resolution. We analyze seismic records from 1825 densely spaced, high-frequency seismometers in Oklahoma, and identify signals from nine precipitation events that occurred during the one-month station deployment in 2016. After removing anthropogenic noise and Earth structure response, the obtained precipitation spatial pattern mimics the one from a nearby operational weather radar, while offering higher spatial (~ 300 m) and temporal (< 10 s) resolution. We further show the potential of this approach to monitor hail with joint analysis of seismic intensity and independent precipitation rate measurements, and advocate for coordinated seismological-meteorological field campaign design.

Regionally high risk increase for precipitation extreme events under global warming.

Daily precipitation extremes are projected to intensify with increasing moisture under global warming following the Clausius-Clapeyron (CC) relationship at about [Formula: see text]. However, this increase is not spatially homogeneous. Projections in individual models exhibit regions with substantially larger increases than expected from the CC scaling. Here, we leverage theory and observations of the form of the precipitation probability distribution to substantially improve intermodel agreement in the medium to high precipitation intensity regime, and to interpret projected changes in frequency in the Coupled Model Intercomparison Project Phase 6. Besides particular regions where models consistently display super-CC behavior, we find substantial occurrence of super-CC behavior within a given latitude band when the multi-model average does not require that the models agree point-wise on location within that band. About 13% of the globe and almost 25% of the tropics (30% for tropical land) display increases exceeding 2CC. Over 40% of tropical land points exceed 1.5CC. Risk-ratio analysis shows that even small increases above CC scaling can have disproportionately large effects in the frequency of the most extreme events. Risk due to regional enhancement of precipitation scale increase by dynamical effects must thus be included in vulnerability assessment even if locations are imprecise.

GoAmazon2014/5 campaign points to deep-inflow approach to deep convection across scales.

A substantial fraction of precipitation is associated with mesoscale convective systems (MCSs), which are currently poorly represented in climate models. Convective parameterizations are highly sensitive to the assumptions of an entraining plume model, in which high equivalent potential temperature air from the boundary layer is modified via turbulent entrainment. Here we show, using multiinstrument evidence from the Green Ocean Amazon field campaign (2014-2015; GoAmazon2014/5), that an empirically constrained weighting for inflow of environmental air based on radar wind profiler estimates of vertical velocity and mass flux yields a strong relationship between resulting buoyancy measures and precipitation statistics. This deep-inflow weighting has no free parameter for entrainment in the conventional sense, but to a leading approximation is simply a statement of the geometry of the inflow. The structure further suggests the weighting could consistently apply even for coherent inflow structures noted in field campaign studies for MCSs over tropical oceans. For radar precipitation retrievals averaged over climate model grid scales at the GoAmazon2014/5 site, the use of deep-inflow mixing yields a sharp increase in the probability and magnitude of precipitation with increasing buoyancy. Furthermore, this applies for both mesoscale and smaller-scale convection. Results from reanalysis and satellite data show that this holds more generally: Deep-inflow mixing yields a strong precipitation-buoyancy relation across the tropics. Deep-inflow mixing may thus circumvent inadequacies of current parameterizations while helping to bridge the gap toward representing mesoscale convection in climate models.

Tightening of tropical ascent and high clouds key to precipitation change in a warmer climate.

The change of global-mean precipitation under global warming and interannual variability is predominantly controlled by the change of atmospheric longwave radiative cooling. Here we show that tightening of the ascending branch of the Hadley Circulation coupled with a decrease in tropical high cloud fraction is key in modulating precipitation response to surface warming. The magnitude of high cloud shrinkage is a primary contributor to the intermodel spread in the changes of tropical-mean outgoing longwave radiation (OLR) and global-mean precipitation per unit surface warming (dP/dTs) for both interannual variability and global warming. Compared to observations, most Coupled Model Inter-comparison Project Phase 5 models underestimate the rates of interannual tropical-mean dOLR/dTs and global-mean dP/dTs, consistent with the muted tropical high cloud shrinkage. We find that the five models that agree with the observation-based interannual dP/dTs all predict dP/dTs under global warming higher than the ensemble mean dP/dTs from the ∼20 models analysed in this study.

Global warming precipitation accumulation increases above the current-climate cutoff scale.

Precipitation accumulations, integrated over rainfall events, can be affected by both intensity and duration of the storm event. Thus, although precipitation intensity is widely projected to increase under global warming, a clear framework for predicting accumulation changes has been lacking, despite the importance of accumulations for societal impacts. Theory for changes in the probability density function (pdf) of precipitation accumulations is presented with an evaluation of these changes in global climate model simulations. We show that a simple set of conditions implies roughly exponential increases in the frequency of the very largest accumulations above a physical cutoff scale, increasing with event size. The pdf exhibits an approximately power-law range where probability density drops slowly with each order of magnitude size increase, up to a cutoff at large accumulations that limits the largest events experienced in current climate. The theory predicts that the cutoff scale, controlled by the interplay of moisture convergence variance and precipitation loss, tends to increase under global warming. Thus, precisely the large accumulations above the cutoff that are currently rare will exhibit increases in the warmer climate as this cutoff is extended. This indeed occurs in the full climate model, with a 3 °C end-of-century global-average warming yielding regional increases of hundreds of percent to >1,000% in the probability density of the largest accumulations that have historical precedents. The probabilities of unprecedented accumulations are also consistent with the extension of the cutoff.

Rough parameter dependence in climate models and the role of Ruelle-Pollicott resonances.

Despite the importance of uncertainties encountered in climate model simulations, the fundamental mechanisms at the origin of sensitive behavior of long-term model statistics remain unclear. Variability of turbulent flows in the atmosphere and oceans exhibits recurrent large-scale patterns. These patterns, while evolving irregularly in time, manifest characteristic frequencies across a large range of time scales, from intraseasonal through interdecadal. Based on modern spectral theory of chaotic and dissipative dynamical systems, the associated low-frequency variability may be formulated in terms of Ruelle-Pollicott (RP) resonances. RP resonances encode information on the nonlinear dynamics of the system, and an approach for estimating them--as filtered through an observable of the system--is proposed. This approach relies on an appropriate Markov representation of the dynamics associated with a given observable. It is shown that, within this representation, the spectral gap--defined as the distance between the subdominant RP resonance and the unit circle--plays a major role in the roughness of parameter dependences. The model statistics are the most sensitive for the smallest spectral gaps; such small gaps turn out to correspond to regimes where the low-frequency variability is more pronounced, whereas autocorrelations decay more slowly. The present approach is applied to analyze the rough parameter dependence encountered in key statistics of an El-Niño-Southern Oscillation model of intermediate complexity. Theoretical arguments, however, strongly suggest that such links between model sensitivity and the decay of correlation properties are not limited to this particular model and could hold much more generally.

High regional climate sensitivity over continental China constrained by glacial-recent changes in temperature and the hydrological cycle.

The East Asian monsoon is one of Earth's most significant climatic phenomena, and numerous paleoclimate archives have revealed that it exhibits variations on orbital and suborbital time scales. Quantitative constraints on the climate changes associated with these past variations are limited, yet are needed to constrain sensitivity of the region to changes in greenhouse gas levels. Here, we show central China is a region that experienced a much larger temperature change since the Last Glacial Maximum than typically simulated by climate models. We applied clumped isotope thermometry to carbonates from the central Chinese Loess Plateau to reconstruct temperature and water isotope shifts from the Last Glacial Maximum to present. We find a summertime temperature change of 6-7 °C that is reproduced by climate model simulations presented here. Proxy data reveal evidence for a shift to lighter isotopic composition of meteoric waters in glacial times, which is also captured by our model. Analysis of model outputs suggests that glacial cooling over continental China is significantly amplified by the influence of stationary waves, which, in turn, are enhanced by continental ice sheets. These results not only support high regional climate sensitivity in Central China but highlight the fundamental role of planetary-scale atmospheric dynamics in the sensitivity of regional climates to continental glaciation, changing greenhouse gas levels, and insolation.

Considerations for parameter optimization and sensitivity in climate models.

Climate models exhibit high sensitivity in some respects, such as for differences in predicted precipitation changes under global warming. Despite successful large-scale simulations, regional climatology features prove difficult to constrain toward observations, with challenges including high-dimensionality, computationally expensive simulations, and ambiguity in the choice of objective function. In an atmospheric General Circulation Model forced by observed sea surface temperature or coupled to a mixed-layer ocean, many climatic variables yield rms-error objective functions that vary smoothly through the feasible parameter range. This smoothness occurs despite nonlinearity strong enough to reverse the curvature of the objective function in some parameters, and to imply limitations on multimodel ensemble means as an estimator of global warming precipitation changes. Low-order polynomial fits to the model output spatial fields as a function of parameter (quadratic in model field, fourth-order in objective function) yield surprisingly successful metamodels for many quantities and facilitate a multiobjective optimization approach. Tradeoffs arise as optima for different variables occur at different parameter values, but with agreement in certain directions. Optima often occur at the limit of the feasible parameter range, identifying key parameterization aspects warranting attention--here the interaction of convection with free tropospheric water vapor. Analytic results for spatial fields of leading contributions to the optimization help to visualize tradeoffs at a regional level, e.g., how mismatches between sensitivity and error spatial fields yield regional error under minimization of global objective functions. The approach is sufficiently simple to guide parameter choices and to aid intercomparison of sensitivity properties among climate models.

Rethinking convective quasi-equilibrium: observational constraints for stochastic convective schemes in climate models.

Convective quasi-equilibrium (QE) has for several decades stood as a key postulate for parametrization of the impacts of moist convection at small scales upon the large-scale flow. Departures from QE have motivated stochastic convective parametrization, which in its early stages may be viewed as a sensitivity study. Introducing plausible stochastic terms to modify the existing convective parametrizations can have substantial impact, but, as for so many aspects of convective parametrization, the results are sensitive to details of the assumed processes. We present observational results aimed at helping to constrain convection schemes, with implications for each of conventional, stochastic or 'superparametrization' schemes. The original vision of QE due to Arakawa fares well as a leading approximation, but with a number of updates. Some, like the imperfect connection between the boundary layer and the free troposphere, and the importance of free-tropospheric moisture to buoyancy, are quantitatively important but lie within the framework of ensemble-average convection slaved to the large scale. Observations of critical phenomena associated with a continuous phase transition for precipitation as a function of water vapour and temperature suggest a more substantial revision. While the system's attraction to the critical point is predicted by QE, several fundamental properties of the transition, including high precipitation variance in the critical region, need to be added to the theory. Long-range correlations imply that this variance does not reduce quickly under spatial averaging; scaling associated with this spatial averaging has potential implications for superparametrization. Long tails of the distribution of water vapour create relatively frequent excursions above criticality with associated strong precipitation events.