Search for: All records

The ocean has absorbed about 25% of the carbon emitted by humans to date. To better predict how much climate will change, it is critical to understand how this ocean carbon sink will respond to future emissions. Here, we examine the ocean carbon sink response to low emission (SSP1-1.9, SSP1-2.6), intermediate emission (SSP2-4.5, SSP5-3.4-OS), and high emission (SSP5-8.5) scenarios in CMIP6 Earth System Models and in MAGICC7, a reduced-complexity climate carbon system model. From 2020–2100, the trajectory of the global-mean sink approximately parallels the trajectory of anthropogenic emissions. With increasing cumulative emissions during this century (SSP5-8.5 and SSP2-4.5), the cumulative ocean carbon sink absorbs 20%–30% of cumulative emissions since 2015. In scenarios where emissions decline, the ocean absorbs an increasingly large proportion of emissions (up to 120% of cumulative emissions since 2015). Despite similar responses in all models, there remains substantial quantitative spread in estimates of the cumulative sink through 2100 within each scenario, up to 50 PgC in CMIP6 and 120 PgC in the MAGICC7 ensemble. We demonstrate that for all but SSP1-2.6, approximately half of this future spread can be eliminated if model results are adjusted to agree with modern observation-based estimates. Considering the spatial distribution of air-sea CO₂fluxes in CMIP6, we find significant zonal-mean divergence from the suite of newly-available observation-based constraints. We conclude that a significant portion of future ocean carbon sink uncertainty is attributable to modern-day errors in the mean state of air-sea CO₂fluxes, which in turn are associated with model representations of ocean physics and biogeochemistry. Bringing models into agreement with modern observation-based estimates at regional to global scales can substantially reduce uncertainty in future role of the ocean in absorbing anthropogenic CO₂from the atmosphere and mitigating climate change.

 

Large volcanic eruptions drive significant climate perturbations through major anomalies in radiative fluxes and the resulting widespread cooling of the surface and upper ocean. Recent studies suggest that these eruptions also drive important variability in air‐sea carbon and oxygen fluxes. By simulating the Earth system using two initial‐condition large ensembles, with and without the aerosol forcing associated with the Mt. Pinatubo eruption in June 1991, we isolate the impact of this volcanic event on physical and biogeochemical properties of the ocean. The Mt. Pinatubo eruption forced significant anomalies in surface fluxes and the ocean interior inventories of heat, oxygen, and carbon. Pinatubo‐driven changes persist for multiple years in the upper ocean and permanently modify the ocean's heat, oxygen, and carbon inventories. Positive anomalies in oxygen concentrations emerge immediately post‐eruption and penetrate into the deep ocean. In contrast, carbon anomalies intensify in the upper ocean over several years post‐eruption, and are largely confined to the upper 150 m. In the tropics and northern high latitudes, the change in oxygen is dominated by surface cooling and subsequent ventilation to mid‐depths, while the carbon anomaly is associated with solubility changes and eruption‐generated El Niño—Southern Oscillation variability. We do not find significant impact of Pinatubo on oxygen or carbon fluxes in the Southern Ocean; but this may be due to Southern Hemisphere aerosol forcing being underestimated in Community Earth System Model 1 simulations.

 

The El Niño‐Southern Oscillation (ENSO) in the equatorial Pacific is the dominant mode of global air‐sea carbon dioxide (CO₂) flux interannual variability (IAV). Air‐sea CO₂fluxes are driven by the difference between atmospheric and surface ocean pCO₂, with variability of the latter driving flux variability. Previous studies found that models in Coupled Model Intercomparison Project Phase 5 (CMIP5) failed to reproduce the observed ENSO‐related pattern of CO₂fluxes and had weak pCO₂IAV, which were explained by both weak upwelling IAV and weak mean vertical dissolved inorganic carbon (DIC) gradients. We assess whether the latest generation of CMIP6 models can reproduce equatorial Pacific pCO₂IAV by validating models against observations‐based data products. We decompose pCO₂IAV into thermally and non‐thermally driven anomalies to examine the balance between these competing anomalies, which explain the total pCO₂IAV. The majority of CMIP6 models underestimate pCO₂IAV, while they overestimate sea surface temperature IAV. Insufficient compensation of non‐thermal pCO₂to thermal pCO₂IAV in models results in weak total pCO₂IAV. We compare the relative strengths of the vertical transport of temperature and DIC and evaluate their contributions to thermal and non‐thermal pCO₂anomalies. Model‐to‐observations‐based product comparisons reveal that modeled mean vertical DIC gradients are biased weak relative to their mean vertical temperature gradients, but upwelling acting on these gradients is insufficient to explain the relative magnitudes of thermal and non‐thermal pCO₂anomalies.

 

Weather forecasts made with imperfect models contain state‐dependent errors. Data assimilation (DA) partially corrects these errors with new information from observations. As such, the corrections, or “analysis increments,” produced by the DA process embed information about model errors. An attempt is made here to extract that information to improve numerical weather prediction. Neural networks (NNs) are trained to predict corrections to the systematic error in the National Oceanic and Atmospheric Administration's FV3‐GFS model based on a large set of analysis increments. A simple NN focusing on an atmospheric column significantly improves the estimated model error correction relative to a linear baseline. Leveraging large‐scale horizontal flow conditions using a convolutional NN, when compared to the simple column‐oriented NN, does not improve skill in correcting model error. The sensitivity of model error correction to forecast inputs is highly localized by vertical level and by meteorological variable, and the error characteristics vary across vertical levels. Once trained, the NNs are used to apply an online correction to the forecast during model integration. Improvements are evaluated both within a cycled DA system and across a collection of 10‐day forecasts. It is found that applying state‐dependent NN‐predicted corrections to the model forecast improves the overall quality of DA and improves the 10‐day forecast skill at all lead times.

 

In large‐eddy simulations, subgrid‐scale (SGS) processes are parameterized as a function of filtered grid‐scale variables. First‐order, algebraic SGS models are based on the eddy‐viscosity assumption, which does not always hold for turbulence. Here we apply supervised deep neural networks (DNNs) to learn SGS stresses from a set of neighboring coarse‐grained velocity from direct numerical simulations of the convective boundary layer at friction Reynolds numbersRe_τup to 1243 without invoking the eddy‐viscosity assumption. The DNN model was found to produce higher correlation between SGS stresses compared to the Smagorinsky model and the Smagorinsky‐Bardina mixed model in the surface and mixed layers and can be applied to different grid resolutions and various stability conditions ranging from near neutral to very unstable. The DNN model can capture key statistics of turbulence ina posteriori(online) tests when applied to large‐eddy simulations of the atmospheric boundary layer.

 

Accurate prediction of precipitation intensity is crucial for both human and natural systems, especially in a warming climate more prone to extreme precipitation. Yet, climate models fail to accurately predict precipitation intensity, particularly extremes. One missing piece of information in traditional climate model parameterizations is subgrid-scale cloud structure and organization, which affects precipitation intensity and stochasticity at coarse resolution. Here, using global storm-resolving simulations and machine learning, we show that, by implicitly learning subgrid organization, we can accurately predict precipitation variability and stochasticity with a low-dimensional set of latent variables. Using a neural network to parameterize coarse-grained precipitation, we find that the overall behavior of precipitation is reasonably predictable using large-scale quantities only; however, the neural network cannot predict the variability of precipitation ( R 2 ∼ 0.45) and underestimates precipitation extremes. The performance is significantly improved when the network is informed by our organization metric, correctly predicting precipitation extremes and spatial variability ( R 2 ∼ 0.9). The organization metric is implicitly learned by training the algorithm on a high-resolution precipitable water field, encoding the degree of subgrid organization. The organization metric shows large hysteresis, emphasizing the role of memory created by subgrid-scale structures. We demonstrate that this organization metric can be predicted as a simple memory process from information available at the previous time steps. These findings stress the role of organization and memory in accurate prediction of precipitation intensity and extremes and the necessity of parameterizing subgrid-scale convective organization in climate models to better project future changes of water cycle and extremes. 

Abstract The process of evapotranspiration transfers liquid water from vegetation and soil surfaces to the atmosphere, the so-called latent heat flux ( Q LE ), and modulates the Earth’s energy, water, and carbon cycle. Vegetation controls Q LE by regulating leaf stomata opening (surface resistance r s in the Big Leaf approach) and by altering surface roughness (aerodynamic resistance r a ). Estimating r s and r a across different vegetation types is a key challenge in predicting Q LE . We propose a hybrid approach that combines mechanistic modeling and machine learning for modeling Q LE . The hybrid model combines a feed-forward neural network which estimates the resistances from observations as intermediate variables and a mechanistic model in an end-to-end setting. In the hybrid modeling setup, we make use of the Penman–Monteith equation in conjunction with multi-year flux measurements across different forest and grassland sites from the FLUXNET database. This hybrid model setup is successful in predicting Q LE , however, this approach leads to equifinal solutions in terms of estimated physical parameters. We follow two different strategies to constrain the hybrid model and therefore control for the equifinality that arises when the two resistances are estimated simultaneously. One strategy is to impose an a priori constraint on r a based on mechanistic assumptions (theory-driven strategy), while the other strategy makes use of more observational data and adds a constraint in predicting r a through multi-task learning of both latent and sensible heat flux ( Q H ; data-driven strategy) together. Our results show that all hybrid models predict the target variables with a high degree of success, with R 2 = 0.82–0.89 for grasslands and R 2 = 0.70–0.80 for forest sites at the mean diurnal scale. The predicted r s and r a show strong physical consistency across the two regularized hybrid models, but are physically implausible in the under-constrained hybrid model. The hybrid models are robust in reproducing consistent results for energy fluxes and resistances across different scales (diurnal, seasonal, and interannual), reflecting their ability to learn the physical dependence of the target variables on the meteorological inputs. As a next step, we propose to test these heavily observation-informed parameterizations derived through hybrid modeling as a substitute for ad hoc formulations in Earth system models.