Earth Systems

Machine Learning Assisted Bayesian Calibration of Model Physics Parameters for Wetland Methane Emissions: A Case Study at a FLUXNET-CH4 Site

Abstract: Methane (CH4) possesses a notably higher warming potential than carbon dioxide despite its lower atmospheric concentration, making it integral to global climate dynamics. Wetlands stand out as the predominant natural contributor to global methane emissions. Accurate modeling of methane emissions from wetlands is crucial for understanding and predicting climate change dynamics. However, such modeling efforts are often constrained by the inherent uncertainties in model parameters.

Our work leverages machine learning (ML) to calibrate five physical parameters of the Energy Exascale Earth System Model (E3SM) land model (ELM) to improve the model’s accuracy in simulating wetland methane emissions. Unlike traditional deterministic calibration methods that target a single set of optimal values for each parameter, Bayesian calibration takes a probabilistic approach and enables capturing the inherent uncertainties in complex systems and providing robust parameter distributions for reliable predictions. However, Bayesian calibration requires numerous model runs and makes it computationally expensive. We employed an ML algorithm, Gaussian process regression (GPR), to emulate the ELM’s methane model, which dramatically reduced the computational time from 6 CPU hours to just 0.72 milliseconds per simulation. We exemplified the procedure at a representative FLUXNET-CH4 site (US-PFa) with the longest continuous methane emission data.

Results showed that the default values for two of the five parameters examined were not aligned well with their respective posterior distributions, suggesting that the model’s default parameter values might not always be optimal for all sites, and that site-specific analysis is warranted. In particular, analyses at sites with different vegetation types and wetland characteristics could reveal more useful insights for understanding methane emissions modeling.

A Large Ensemble Global Dataset for Climate Impact Assessments

Abstract: We present a self-consistent, large ensemble, high-resolution global dataset of long-term future climate, which accounts for the uncertainty in climate system response to anthropogenic emissions of greenhouse gases and in geographical patterns of climate change. The dataset is developed by applying an integrated spatial disaggregation (SD) - bias-correction (BC) method to climate projections from the MIT Integrated Global System Modeling (IGSM) framework. Four emissions scenarios are considered that represent energy and environmental policies and commitments of potential future pathways, namely, Reference, Paris Forever, Paris 2°C and Paris 1.5°C. The dataset contains nine key meteorological variables on a monthly scale from 2021 to 2100 at a spatial resolution of 0.5°x 0.5°, including precipitation, air temperature (mean, minimum and maximum), near-surface wind speed, shortwave and longwave radiation, specific humidity, and relative humidity.

We demonstrate the dataset’s ability to represent climate-change responses across various regions of the globe.

This dataset can be used to support regional-scale climate-related impact assessments of risk across different applications that include hydropower, water resources, ecosystem, agriculture, and sustainable development.

Emissions of CFC-11, a chlorofluorocarbon once frequently used in cooling and insulation systems to improve the quality of life, can also endanger life. Upon entry into the stratosphere where solar ultraviolet radiation is strong, CFC-11 decomposes, resulting in the release of chlorine, which degrades the ozone layer that shields life from harmful UV rays. In 2018, a team of scientists discovered an alarming upward spike in global CFC-11 emissions from 2013 to 2017.

Probing the atmosphere, protecting the biosphere
On the Influence of Ocean Sinks and Hydroxyl Radical Changes on Estimated Lifetimes and Emissions of Halogenated Greenhouse Gases

Abstract: Halogenated greenhouse gases (such as HCFCs, HFCs, PFCs, and SF6) have global warming potentials thousands to tens of thousands of times greater than carbon dioxide on a per kilogram basis. Estimating the emissions of these gases on a global scale is challenging since direct measurements are unavailable. Instead, they are inferred using measured global atmospheric concentrations and knowledge of their lifetimes. The ocean uptake for halogenated species can impact their lifetimes, but this process has been assumed to be largely negligible in the past. Further, reaction with hydroxyl radicals (OH) is a major atmospheric loss pathway for HCFCs and HFCs. Emission estimations usually assume OH is constant over time, but recent chemistry-climate models suggest OH increased after 1980, implying underestimated emissions. Here, we use a coupled atmosphere-ocean model to explore how the inferred lifetimes and emissions of certain HCFCs, HFCs, PFCs, and SF6 can be affected by ocean processes and time-varying OH. We show that by including the ocean uptake, the lifetimes are shortened by 2 – 15% for HCFCs and HFCs, and 20 – 40% for PFC-14 and SF6. Certain HCFCs and HFCs can be further destroyed in the ocean due to microbial activity; this could lead to up to an another 8 – 25% decrease in their lifetimes. We also show that increases in modeled OH imply an additional underestimation in HCFC and HFC emissions by ~10% near their respective peak emissions. These species are considered under the Montreal Protocol and its amendments and the Paris Agreement. Evaluating the success of these global agreements requires accurate knowledge of contributions to global warming from these gases and consideration of these processes.

Evaluation of the APCEMM intermediate-fidelity contrail model using LIDAR observations

Abstract: Contrails are aircraft-induced ice clouds that are estimated to account for 57% of aviation’s anthropogenic climate impact. However, an individual contrail's impacts are highly uncertain, and accurate models of individual contrails are needed in order to accurately predict and optimize the effects of different mitigation efforts. Existing high-fidelity (e.g. LES) contrail models are computationally expensive and therefore infeasible to use for large-scale simulation, while faster zero-dimensional models must necessarily rely on parameterizations of contrail properties which may not apply in all circumstances. The APCEMM model attempts to bridge this gap as an intermediate-fidelity model that features binned microphysics and 2-D advection/diffusion, while still being fast enough to run at scale. Here we evaluate the accuracy of APCEMM in predicting the shape, optical properties, and size of contrails observed in satellite LIDAR observations which have been attributed to specific flights. We classify differences into those due to our estimate of the ambient meteorology and those due to the APCEMM model’s assumptions about the physics. Using this data, we establish the degree to which accurate modeling of the contrail cross-section is necessary - or unnecessary - to understand and predict individual contrail climate impacts under different mitigation scenarios.

Pages

Subscribe to Earth Systems