Search

Previous work identified an anthropogenic fingerprint pattern in TAC(x, t), the amplitude of the seasonal cycle of mid- to upper-tropospheric temperature (TMT), but did not explicitly consider whether fingerprint identification in satellite TAC(x, t) data could have been influenced by real-world multidecadal internal variability (MIV). We address this question here using large ensembles (LEs) performed with five climate models. LEs provide many different sequences of internal variability noise superimposed on an underlying forced signal. Despite differences in historical external forcings, climate sensitivity, and MIV properties of the five models, their TAC(x, t) fingerprints are similar and statistically identifiable in 239 of the 240 LE realizations of historical climate change. Comparing simulated and observed variability spectra reveals that consistent fingerprint identification is unlikely to be biased by model underestimates of observed MIV. Even in the presence of large (factor of 3–4) intermodel and inter-realization differences in the amplitude of MIV, the anthropogenic fingerprints of seasonal cycle changes are robustly identifiable in models and satellite data. This is primarily due to the fact that the distinctive, global-scale fingerprint patterns are spatially dissimilar to the smaller-scale patterns of internal TAC(x, t) variability associated with the Atlantic multidecadal oscillation and El Niño–Southern Oscillation. The robustness of the seasonal cycle detection and attribution results shown here, taken together with the evidence from idealized aquaplanet simulations, suggest that basic physical processes are dictating a common pattern of forced TAC(x, t) changes in observations and in the five LEs. The key processes involved include GHG-induced expansion of the tropics, lapse-rate changes, land surface drying, and sea ice decrease.

Climate-model simulations exhibit approximately two times more tropical tropospheric warming than satellite observations since 1979. The causes of this difference are not fully understood and are poorly quantified. Here, we apply machine learning to relate the patterns of surface-temperature change to the forced and unforced components of tropical tropospheric warming. This approach allows us to disentangle the forced and unforced change in the model-simulated temperature of the midtroposphere (TMT). In applying the climate-model-trained machine-learning framework to observations, we estimate that external forcing has produced a tropical TMT trend of 0.25 ± 0.08 K⋅decade −1 between 1979 and 2014, but internal variability has offset this warming by 0.07 ± 0.07 K⋅decade −1 . Using the Community Earth System Model version 2 (CESM2) large ensemble, we also find that a discontinuity in the variability of prescribed biomass-burning aerosol emissions artificially enhances simulated tropical TMT change by 0.04 K⋅decade −1 . The magnitude of this aerosol-forcing bias will vary across climate models, but since the latest generation of climate models all use the same emissions dataset, the bias may systematically enhance climate-model trends over the satellite era. Our results indicate that internal variability and forcing uncertainties largely explain differences in satellite-versus-model warming and are important considerations when evaluating climate models.

Current knowledge of the spatiotemporal patterns of changes in soil moisture-based terrestrial aridity has considerable uncertainty. Using Standardized Soil Moisture Index (SSI) calculated from multi-source merged data sets, we find widespread drying in the global midlatitudes, and wetting in the northern subtropics and in spring between 45°N–65°N, during 1971–2016. Formal detection and attribution analysis shows that human forcings, especially greenhouse gases, contribute significantly to the changes in 0–10 cm SSI during August–November, and 0–100 cm during September–April. We further develop and apply an emergent constraint method on the future SSI’s signal-to-noise (S/N) ratios and trends under the Shared Socioeconomic Pathway 5-8.5. The results show continued significant presence of human forcings and more rapid drying in 0–10 cm than 0–100 cm. Our findings highlight the predominant human contributions to spatiotemporally heterogenous terrestrial aridification, providing a basis for drought and flood risk management.

The path toward realizing the potential of seasonal forecasting and its socioeconomic benefits depends heavily on improving general circulation model based dynamical forecasting systems. To improve dynamical seasonal forecast, it is crucial to set up forecast benchmarks, and clarify forecast limitations posed by model initialization errors, formulation deficiencies, and internal climate variability. With huge cost in generating large forecast ensembles, and limited observations for forecast verification, the seasonal forecast benchmarking and diagnosing task proves challenging. In this study, we develop a probabilistic deep neural network model, drawing on a wealth of existing climate simulations to enhance seasonal forecast capability and forecast diagnosis. By leveraging complex physical relationships encoded in climate simulations, our probabilistic forecast model demonstrates favorable deterministic and probabilistic skill compared to state-of-the-art dynamical forecast systems in quasi-global seasonal forecast of precipitation and near-surface temperature. We apply this probabilistic forecast methodology to quantify the impacts of initialization errors and model formulation deficiencies in a dynamical seasonal forecasting system. We introduce the saliency analysis approach to efficiently identify the key predictors that influence seasonal variability. Furthermore, by explicitly modeling uncertainty using variational Bayes, we give a more definitive answer to how the El Nino/Southern Oscillation, the dominant mode of seasonal variability, modulates global seasonal predictability.

Although anthropogenic climate change is expected to have caused large shifts in temperature and rainfall, the detection of human influence on global drought has been complicated by large internal variability and the brevity of observational records. Here we address these challenges using reconstructions of the Palmer drought severity index obtained with data from tree rings that span the past millennium. We show that three distinct periods are identifiable in climate models, observations and reconstructions during the twentieth century. In recent decades (1981 to present), the signal of greenhouse gas forcing is present but not yet detectable at high confidence. Observations and reconstructions differ significantly from an expected pattern of greenhouse gas forcing around mid-century (1950-1975), coinciding with a global increase in aerosol forcing. In the first half of the century (1900-1949), however, a signal of greenhouse-gas-forced change is robustly detectable. Multiple observational datasets and reconstructions using data from tree rings confirm that human activities were probably affecting the worldwide risk of droughts as early as the beginning of the twentieth century.