Department of Earth System Science

Understanding the global hydrogen (H2) budget is critical as H2 is expected to play an important role in future energy systems. Tropospheric H2 sources include direct emissions and atmospheric production via chemical reactions, while sinks are soil uptake and removal by hydroxyl radical (OH). Large uncertainties remain in quantifying the atmospheric production and loss of H2 due to large uncertainties in H2 uptake in the soil and the lack of global-scale knowledge of the abundance of OH. We use a suite of global three-dimensional Atmospheric Chemistry Models (ACM) to evaluate key reactive species involved in atmospheric production and loss-formaldehyde (HCHO), nitrogen dioxide (NO2), and carbon monoxide (CO). A box model is then used to simulate the evolution of global mean tropospheric H2 from pre-industrial to present day; to test different relative contributions in atmospheric production from methane and Volatile Organic Compounds (VOC); and to assess atmospheric loss with different OH concentrations. Isotopic compositions of the different sources and sinks are further used to constrain these terms and assess the potential importance of geological sources to the H2 budget. Models generally match each other for HCHO, though model diversity exists for NO2 and CO. From model evaluations and box model constraints, we estimate atmospheric H2 production of 37–60 Tgyr^-1, and atmospheric losses of 15–30 Tgyr^-1, suggesting that some top-down literature estimates may overestimate production. Box model results suggest an upper bound of 9 Tgyr^-1 for geological sources, considerably lower than the 23 Tgyr^-1 proposed previously. We recommend more isotopic observations and targeted measurement campaigns to further refine the budget.

Controlled Environment Agriculture (CEA) offers high-yield, climate-resilient food production, but high energy and resource demands challenge its sustainability. This paper synthesizes technologies that can improve outcomes across six categories—energy, CO2 utilization, building envelope, hardware, water, and process—plus colocation strategies. We evaluate 80 technologies and define ten implementation pathways bundling complementary technologies to reduce energy use, optimize water consumption, and minimize emissions. Regional application is demonstrated through five U.S. case studies spanning different climates. A logic framework guides pathway selection for case studies based on climate, infrastructure, and regulatory context, informing context-sensitive technology deployment. Results show energy intensity reductions of 3–55 %, ranging from energy management programs to comprehensive lighting retrofits; water savings of 20–40 % through closed-loop recirculation; and emissions reductions of 3–100 %, with strategic energy management achieving 3–5 % and renewable electricity paired with electrified heating achieving up to 100 %. Text mining revealed that energy, hardware, and process technologies account for 91 % of literature coverage. Water, building envelope, and CO2 utilization remain underexplored, indicating priorities for future research. This integrative approach to technology assessment supports growers, developers, and policymakers in aligning CEA system design with local conditions, improving resource efficiency and addressing gaps in cross-domain technology coverage.

Tropical forests are highly diverse and productive ecosystems and store nearly 60% of vegetation biomass. Across the tropics, forests span large gradients of soil fertility, from vast regions situated on highly weathered, ancient geological formations to others on young, nutrient-rich landscapes. Tropical forest productivity, mortality, and biomass all vary systematically across these gradients in soil fertility. Aboveground forest productivity tends to increase with higher soil fertility, while counterintuitively, aboveground biomass does not increase proportionally. This disconnect is likely due to coinciding increases in mortality with higher soil fertility. However, we know relatively little about the mechanisms underlying how soil fertility regulates productivity or mortality - two critical determinants of forest biomass - and even less about how these relationships will shape tropical forest responses to global change. Here, we present a mechanistic framework for understanding how soil fertility affects productivity through photosynthesis, carbon allocation, and carbon-use efficiency; how it modulates mortality via direct and indirect interactions with various drivers of mortality; and how global change may alter these relationships. Based on this synthesis, we outline a roadmap to advance our understanding of tropical forests and improve predictions of their responses to global change.

Drought has consequences for microbial decomposition rates, including indirect effects through changes in plant litter chemistry. Here, we studied the impact of a decade-long drought on plant litter chemistry and microbial decomposition traits in a semi-arid ecosystem during an 18-month litter bag experiment. We investigated litter sourced from four conditions: grass and shrub vegetation under ambient and reduced precipitation. We hypothesized that litter chemistry drives microbial decomposition capabilities and enzyme activity due to vegetation differences and drought effects on litter chemistry. We found that carbohydrate-rich grass litter had a higher abundance of decomposition genes detected using metagenomics and enzyme activity than more recalcitrant shrub litter, which was richer in lignin and lipids; these patterns were related to substrate supply. Drought decreased some carbohydrate fractions in grass litter but did not change the lignin fraction in grass and shrub litter, suggesting that drought does not make litter more recalcitrant. Most decomposition genes and enzyme activities were not significantly affected by drought, thereby maintaining decomposition rates. Microbial community succession patterns-decreasing fungal abundance and increasing bacterial abundance with time-corresponded with decreasing chitin gene abundance and increasing peptidoglycan gene abundance over time, indicating microbial necromass recycling. We demonstrate minimal litter chemistry-mediated effects of drought but show significant changes in community composition and their decomposition capabilities over time, highlighting that complex microbial-chemical interactions under climate change can influence ecosystem-scale processes. IMPORTANCE: Climate change is causing more severe and frequent droughts in semi-arid ecosystems, affecting soil microbes breaking down plant litter. Our research focuses on understanding the less studied pathway of drought impact on microbes via changes in plant litter chemistry. Drought can alter the plant litter chemistry by changing the composition and physiology of plants, which can alter microbial decomposition and ecosystem-level carbon cycling. We investigated litter decomposition traits of microbial communities in grass and shrub litter under long-term drought. There were significant changes in litter chemistry under drought but no increase in lignin fraction. Despite this, microbial communities maintained their decomposition capabilities under drought, highlighting the ability of microbes to adapt and continue functioning. We also demonstrate unique microbial community succession patterns and dead biomass recycling, which can have implications for carbon cycling rates in the ecosystem. This study sheds light on the complex microbial interactions that affect ecosystem functioning under climate change.

Hydrogen has gained attention as a key component of future low-carbon energy and industrial systems, with demand projected to increase from 100 Mt per year in 2024 to up to 1,370 Mt per year in 2050. However, there are concerns about the climate impact of the associated increase in hydrogen emissions. In this Review, we discuss the sources and sinks of atmospheric hydrogen and the resulting climate effects, including the use of metrics and implications for policy. Despite not being a direct greenhouse gas, hydrogen raises levels of methane, tropospheric ozone and stratospheric water vapour, which are potent greenhouse gases. Estimates of the 100-year global warming potential of hydrogen converge at 12 ± 6 (90% CI). Such metrics are sufficiently robust to inform policy and business decision-making and to be included in climate impact assessments to maximize the benefits of a future hydrogen economy. However, substantial uncertainties remain, particularly in quantifying various sources of atmospheric hydrogen from human and natural systems (estimates have a range of 55–141 Tg per year) and understanding soil uptake rates (32–90 Tg per year). Future research should focus on improved quantification of hydrogen emissions, especially across the hydrogen system value chain, better constraints on the soil sink and a more complete representation of hydrogen chemistry.

Accurate and spatially explicit forest Aboveground Biomass (AGB) mapping through remote sensing is critical for quantifying terrestrial carbon stocks and informing effective forest management strategies. However, AGB estimation in dense forests with complex terrain remains challenging due to satellite sensor signal saturation problem (saturation issue occurs in high biomass forests), structural complexity, and limited ground truth for calibration. This study presents a novel framework that integrates multi-temporal Sentinel-2 optical imagery, ALOS PALSAR-2 Synthetic Aperture Radar (SAR) data, and topographic variables with explainable Machine Learning to map AGB across mountainous forests within subtropical and temperate oceanic climate zones of Mexico. We evaluate the effects of temporal granularity and sensor synergy by comparing multiple temporal inputs and sensor configurations (Sentinel-2, PALSAR-2, and their fusion), and assess model performance using two reference datasets: NASA GEDI LiDAR-derived biomass and Mexico’s National Forest and Soil Inventory (INFyS). Our results showed that models trained on INFyS consistently outperformed those trained on GEDI, highlighting limitations in GEDI’s reliability in biomass estimates within this study region. Furthermore, the integration of Sentinel-2 and PALSAR-2 provided improved predictions compared to single-sensor models, particularly when combined with temporally explicit yearly statistics. The best-performing model, which was trained on INFyS data, and considered both Sentinel-2 and PALSAR-2 yearly statistics, as well as topographic variables, achieved an R2 of 0.64, RMSE of 51.10 Mg/ha, and relative RMSE (rRMSE) of 58.69%. Explainable ML analysis identified Sentinel-2 spectral indices and topographic features as key predictors, while PALSAR-2 metrics provided complementary information, partially mitigating saturation effects in high-biomass areas. Specifically, integrating both sensors substantially improved AGB estimation in high biomass forest (≥200 Mg/ha), yielding 98% gains over optical-only model, with resulting estimates exceeding GEDI L4B by 29% and ESA-CCI-BIOMASS by 174%. Terrain-stratified analysis indicated close agreement with GEDI in low-slope areas, with increasing divergence as slope steepness increased, while estimates remained consistently higher than ESA-CCI-BIOMASS across all slope classes. The proposed approach advances multi-sensor fusion and temporal feature engineering for AGB mapping using open-access satellite datasets, providing a scalable and reproducible framework for annual biomass monitoring in topographically complex mountainous forests. The resulting 25 m resolution biomass product has the potential to provide spatially detailed information for forest monitoring and may support applications in carbon accounting and forest management.

Reliable predictions of dryland carbon fluxes require understanding the persistence and turnover of soil organic carbon (SOC). We measure radiocarbon to quantify the age of SOC and CO₂ released from soil respiration at 97 dryland sites across six continents. Here we show that bulk SOC contains little C fixed in the past 60 years, while respired CO₂ originates from both bomb-derived recent C and millennia-old C, challenging the idea that old C is chemically or physically protected. Radiocarbon suggests mean ages of ~2100 years for bulk SOC and ~520 years for respired CO₂, the latter far older than machine-learning (<50 years) or Earth system models predict. Aridity, net primary productivity, and SOC content are dominant predictors for radiocarbon signatures, with abrupt shifts to older C beyond an aridity threshold of ~0.87. Our findings underscore the need to incorporate the vulnerability of older carbon into models and land management strategies.

Abstract The Energy Exascale Earth System Model version 3 (E3SMv3) represents the latest advancement in Earth system modeling developed by the U.S. Department of Energy (DOE). Building upon previous versions, E3SMv3 introduces significant updates across its coupled components to enhance capability and improve fidelity. The atmosphere component incorporates advancements in chemistry, aerosol‐cloud interactions, convection, and microphysics. The ocean features a new time‐stepping scheme and a higher‐resolution unstructured mesh with sub‐ice‐shelf cavities, while the sea ice model integrates advanced snow and ice physics for more realistic cryospheric simulations. The land model introduces prognostic vegetation dynamics and a new sub‐grid topographic treatment of solar radiation. A new tri‐grid configuration harmonizes the horizontal grids of the land and river components for improved process coupling. It is enabled by a new non‐linear remapping between the atmosphere and land. E3SMv3 underwent extensive testing through a comprehensive simulation campaign, including pre‐industrial control, idealized experiments, and historical simulations spanning 1850–2024. The model demonstrates significant improvements in simulating the evolution of the historical surface temperature, particularly addressing the “pothole cooling” bias in earlier versions. Reduced aerosol‐related forcing contributes to more realistic radiative forcing and better alignment with the observational record. Ocean heat content (OHC) and sea ice trends are also improved as a result. Plain Language Summary The Energy Exascale Earth System Model version 3 (E3SMv3), developed by the U.S. Department of Energy, is a state‐of‐the‐science tool designed to advance our understanding of the Earth and energy systems. This model simulates interactions between the atmosphere, land, rivers, oceans, and sea ice to predict Earth system changes and their impacts. E3SMv3 includes major improvements, such as enhanced representations of atmospheric chemistry, clouds, aerosols, sea ice, and vegetation dynamics. It also introduces refined marine and land grids to improve the accuracy of how these components interact. E3SMv3 was evaluated through simulations of past and historical conditions, totaling over 5,000 years. The model successfully resolved a major issue leading to unrealistic cooling trends during the mid‐20th century. E3SMv3 now provides more accurate predictions of surface temperatures, ocean heat content, and sea ice changes, closely matching observed data. Key Points E3SMv3 introduces major updates to the atmosphere, sea ice, and land components, enhancing capability and improving fidelity E3SMv3 increases resolution in the ocean‐sea ice mesh and unifies land and river grids for tighter coupling within and between components Reduced aerosol forcing in E3SMv3 resolves mid‐20th century cooling biases, aligning historical temperature trends with observations

Abstract In recent years, machine‐learning (ML) models trained on reanalysis data have rivaled physics‐based forecast models in terms of performance skill for global weather forecasting. With increased rollout stability, the question of how these models perform for subseasonal to seasonal (S2S, week 3–8) forecasting has emerged. In this study we run a large set of subseasonal hindcasts over 2004–2023 to evaluate two ML weather forecast models at the S2S time scale, SFNO‐HENS (Nvidia, fully ML) and NeuralGCM (Google Research, hybrid). Corresponding hindcasts from the European Centre for Medium‐Range Weather Forecasts (ECMWF) are used as a baseline for comparison to a physics‐based model. Because our focus is on predicting moisture transport over the Western United States between October and March, we evaluate the models' prediction skill for the Madden‐Julian Oscillation (MJO) and its associated teleconnections in the North Pacific. We find that both ML models are competitive with the ECWMF model, with comparable skill in predicting the North Pacific large‐scale circulation and the MJO at week 3 and beyond. Even though overall the mid‐latitude subseasonal prediction skill remains low, the ML models exhibit interesting behavior such as a realistic propagation of the MJO across the Maritime Continent and realistic teleconnections. A SFNO‐HENS sensitivity experiment with altered initial conditions in the tropics demonstrates the stability of the model, and it illustrates the capability of ML models to represent important physical processes of the atmosphere at the S2S time scale. Plain Language Summary Predicting weather patterns and precipitation a few weeks in advance (subseasonal time scale) is of great interest for stakeholders such as water managers in the Southwest United States (US), where arid conditions prevail. Subseasonal forecasts from traditional weather forecast models exhibit low skill in the region, limiting their applicability. Here we examine whether the recent breakthrough in weather forecasting made with machine learning/artificial intelligence models can translate to improved subseasonal forecasts. Recently‐developed machine learning models exhibit comparable skill to a state‐of‐the‐art physics‐based model for predicting weather patterns in the North Pacific/North America region, and associated moisture transport. The same applies to their skill in predicting the tropical pattern, the Madden‐Julian Oscillation, and its important remote perturbations over the midlatitude East Pacific and Southwest US. Additionally, a perturbation experiment carried out with one of the machine learning models illustrates their ability to not only predict the evolution of atmospheric fields, but also to learn and represent physical processes such as tropics‐extratropics Rossby wave propagation. Key Points Two machine learning weather forecast models exhibit state‐of‐the‐art prediction skill at the subseasonal time scale in the Pacific sector The models equal ECWMF in terms of Madden‐Julian oscillation (MJO) prediction skill, and they accurately predict the MJO propagation and associated teleconnections The two machine‐learning models represent key physical processes for subseasonal prediction, despite being trained for weather forecasting