skip to main content
US FlagAn official website of the United States government
dot gov icon
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
https lock icon
Secure .gov websites use HTTPS
A lock ( lock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


This content will become publicly available on April 15, 2026

Title: Transferring climate change physical knowledge
Precise and reliable climate projections are required for climate adaptation and mitigation, but Earth system models still exhibit great uncertainties. Several approaches have been developed to reduce the spread of climate projections and feedbacks, yet those methods cannot capture the nonlinear complexity inherent in the climate system. Using a Transfer Learning approach, we show that Machine Learning can be used to optimally leverage and merge the knowledge gained from global temperature maps simulated by Earth system models and observed in the historical period to reduce the spread of global surface air temperature fields projected in the 21st century. We reach an uncertainty reduction of more than 50% with respect to state-of-the-art approaches while giving evidence that our method provides improved regional temperature patterns together with narrower projections uncertainty, urgently required for climate adaptation.  more » « less
Award ID(s):
2019625 2047418 2007719
PAR ID:
10611093
Author(s) / Creator(s):
; ; ; ; ; ; ;
Publisher / Repository:
PNAS
Date Published:
Journal Name:
Proceedings of the National Academy of Sciences
Volume:
122
Issue:
15
ISSN:
0027-8424
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; (, ICES Journal of Marine Science)
    Travers-Trolet, Morgane (Ed.)
    Abstract Projections of climate change impacts on living resources are being conducted frequently, and the goal is often to inform policy. Species projections will be more useful if uncertainty is effectively quantified. However, few studies have comprehensively characterized the projection uncertainty arising from greenhouse gas scenarios, Earth system models (ESMs), and both structural and parameter uncertainty in species distribution modelling. Here, we conducted 8964 unique 21st century projections for shifts in suitable habitat for seven economically important marine species including American lobster, Pacific halibut, Pacific ocean perch, and summer flounder. For all species, both the ESM used to simulate future temperatures and the niche modelling approach used to represent species distributions were important sources of uncertainty, while variation associated with parameter values in niche models was minor. Greenhouse gas emissions scenario contributed to uncertainty for projections at the century scale. The characteristics of projection uncertainty differed among species and also varied spatially, which underscores the need for improved multi-model approaches with a suite of ESMs and niche models forming the basis for uncertainty around projected impacts. Ensemble projections show the potential for major shifts in future distributions. Therefore, rigorous future projections are important for informing climate adaptation efforts. 
    more » « less
  2. ; ; ; ; ; ; ; (, Journal of Advances in Modeling Earth Systems)
    Abstract Assessing uncertainty in future climate projections requires understanding both internal climate variability and external forcing. For this reason, single‐model initial condition large ensembles (SMILEs) run with Earth System Models (ESMs) have recently become popular. Here we present a new 20‐member SMILE with the Energy Exascale Earth System Model version 1 (E3SMv1‐LE), which uses a “macro” initialization strategy choosing coupled atmosphere/ocean states based on inter‐basin contrasts in ocean heat content (OHC). The E3SMv1‐LE simulates tropical climate variability well, albeit with a muted warming trend over the twentieth century due to overly strong aerosol forcing. The E3SMv1‐LE's initial climate spread is comparable to other (larger) SMILEs, suggesting that maximizing inter‐basin ocean heat contrasts may be an efficient method of generating ensemble spread. We also compare different ensemble spread across multiple SMILEs, using surface air temperature and OHC. The Community Earth system Model version 1, the only ensemble which utilizes a “micro” initialization approach perturbing only atmospheric initial conditions, yields lower spread in the first ∼30 years. The E3SMv1‐LE exhibits a relatively large spread, with some evidence for anthropogenic forcing influencing spread in the late twentieth century. However, systematic effects of differing “macro” initialization strategies are difficult to detect, possibly resulting from differing model physics or responses to external forcing. Notably, the method of standardizing results affects ensemble spread: control simulations for most models have either large background trends or multi‐centennial variability in OHC. This spurious disequlibrium behavior is a substantial roadblock to understanding both internal climate variability and its response to forcing. 
    more » « less
  3. ; ; ; ; ; ; ; ; ; ; et al (, Atmospheric Chemistry and Physics)
    Abstract. We present here results from the Geoengineering Model Intercomparison Project (GeoMIP) simulations for the experiments G6sulfur and G6solar for six Earth system models participating in the Climate Model Intercomparison Project (CMIP) Phase 6. The aim of the experiments is to reduce the warming that results from a high-tier emission scenario (Shared Socioeconomic Pathways SSP5-8.5) to that resulting from a medium-tier emission scenario (SSP2-4.5). These simulations aim to analyze the response of climate models to a reduction in incoming surface radiation as a means to reduce global surface temperatures, and they do so either by simulating a stratospheric sulfate aerosol layer or, in a more idealized way, through a uniform reduction in the solar constant in the model. We find that over the final two decades of this century there are considerable inter-model spreads in the needed injection amounts of sulfate (29 ± 9 Tg-SO2/yr between 2081 and 2100), in the latitudinal distribution of the aerosol cloud and in the stratospheric temperature changes resulting from the added aerosol layer. Even in the simpler G6solar experiment, there is a spread in the needed solar dimming to achieve the same global temperature target (1.91 ± 0.44 %). The analyzed models already show significant differences in the response to the increasing CO2 concentrations for global mean temperatures and global mean precipitation (2.05 K ± 0.42 K and 2.28 ± 0.80 %, respectively, for SSP5-8.5 minus SSP2-4.5 averaged over 2081–2100). With aerosol injection, the differences in how the aerosols spread further change some of the underlying uncertainties, such as the global mean precipitation response (−3.79 ± 0.76 % for G6sulfur compared to −2.07 ± 0.40 % for G6solar against SSP2-4.5 between 2081 and 2100). These differences in the behavior of the aerosols also result in a larger uncertainty in the regional surface temperature response among models in the case of the G6sulfur simulations, suggesting the need to devise various, more specific experiments to single out and resolve particular sources of uncertainty. The spread in the modeled response suggests that a degree of caution is necessary when using these results for assessing specific impacts of geoengineering in various aspects of the Earth system. However, all models agree that compared to a scenario with unmitigated warming, stratospheric aerosol geoengineering has the potential to both globally and locally reduce the increase in surface temperatures. 
    more » « less
  4. (, Journal of climate)
    null (Ed.)
    The Arctic has experienced a warming rate higher than the global mean in the past decades, but previous studies show that there are large uncertainties associated with future Arctic temperature projections. In this study, near- surface mean temperatures in the Arctic are analyzed from 22 models participating in phase 6 of the Coupled Model Intercomparison Project (CMIP6). Compared with the ERA5 reanalysis, most CMIP6 models underestimate the observed mean temperature in the Arctic during 1979–2014. The largest cold biases are found over the Greenland Sea the Barents Sea, and the Kara Sea. Under the SSP1-2.6, SSP2-4.5, and SSP5-8.5 scenarios, the multimodel ensemble mean of 22 CMIP6 models exhibits significant Arctic warming in the future and the warming rate is more than twice that of the global/Northern Hemisphere mean. Model spread is the largest contributor to the overall uncertainty in projections, which accounts for 55.4% of the total uncertainty at the start of projections in 2015 and remains at 32.9% at the end of projections in 2095. Internal variability uncertainty accounts for 39.3% of the total uncertainty at the start of projections but decreases to 6.5% at the end of the twenty-first century, while scenario uncertainty rapidly increases from 5.3% to 60.7% over the period from 2015 to 2095. It is found that the largest model uncertainties are consistent cold bias in the oceanic regions in the models, which is connected with excessive sea ice area caused by the weak Atlantic poleward heat transport. These results suggest that large intermodel spread and uncertainties exist in the CMIP6 models’ simulation and projection of the Arctic near- surface temperature and that there are different responses over the ocean and land in the Arctic to greenhouse gas forcing. Future research needs to pay more attention to the different characteristics and mechanisms of Arctic Ocean and land warming to reduce the spread. 
    more » « less
  5. ; (, Earth's Future)
    Abstract Climate models disagree on the direction of precipitation change over about half of the Earth. Current characterizations of expected change use the ensemble mean, which systematically underestimates the magnitude and overestimates the time of emergence (ToE) of precipitation change in regions of high uncertainty. We develop a new approach to estimate both ToE and the potential to update uncertainty in precipitation over time with new observations. Further, we develop two new metrics that increase the usefulness of ToE for adaptation planning. The time of confidence estimates when projections of precipitation emergence will have high confidence. Second, the advance warning time (AWT) indicates how long policymakers will have to prepare for a new precipitation regime after they know change is likely to occur. Our approach uses individual model projections that show change before averaging across models to calculate ToE. It then applies a Bayesian method to constrain uncertainty from climate model ensembles using a perfect model approach. Results demonstrate the potential for widespread and decades‐earlier precipitation emergence, with potential for end‐of‐century emergence to occur across 99% of the Earth compared to 60% in previous estimates. Our method reduces uncertainty in the direction of change across 8% of the globe. We find positive estimates of AWT across most of the Earth; however, in 34% of regions there is potential for no advanced warning before new precipitation regimes emerge. These estimates can guide adaptation planning, reducing the risk that policymakers are unprepared for precipitation changes that occur earlier than expected. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email