Global Model Research Articles

Ammonium-derived nitrous oxide is a global source in streams

Global riverine nitrous oxide (N2O) emissions have increased more than 4-fold in the last century. It has been estimated that the hyporheic zones in small streams alone may contribute approximately 85% of these N2O emissions. However, the mechanisms and pathways controlling hyporheic N2O production in stream ecosystems remain unknown. Here, we report that ammonia-derived pathways, rather than the nitrate-derived pathways, are the dominant hyporheic N2O sources (69.6 ± 2.1%) in agricultural streams around the world. The N2O fluxes are mainly in positive correlation with ammonia. The potential N2O metabolic pathways of metagenome-assembled genomes (MAGs) provides evidence that nitrifying bacteria contain greater abundances of N2O production-related genes than denitrifying bacteria. Taken together, this study highlights the importance of mitigating agriculturally derived ammonium in low-order agricultural streams in controlling N2O emissions. Global models of riverine ecosystems need to better represent ammonia-derived pathways for accurately estimating and predicting riverine N2O emissions.

Properties of the ionization region in a magnetron plasma at gas-aggregation-source-relevant pressure regime explored using a global model

Abstract A global (volume averaged) model is developed for the ionization region of a gas aggregation source (GAS) plasma. The case of using argon gas and a copper target is considered. The model describes the densities of thermal and hot electrons, argon and copper ions, copper atoms and argon atoms in different excited states, the temperature of thermal electrons, the kinetic energies of the ions with which they bombard the target, the sheath width near the target cathode and the energy fluxes by different plasma species to a planar probe in the ionization region. Also, the fraction of input power is estimated which is dissipated to energize the thermal electrons in the ionization region. The gas discharge properties are analysed for different pressures and discharge currents under conditions corresponding to the experimental conditions (Gauter S et al 2018 J. Appl. Phys. 124 073301). The calculated pressure- and current-dependences for the GAS properties are used to explain the measured dependences for the deposition rate and the energy flux. It is found that the deposition rate increases with increasing discharge current because of the growth of currents of copper atoms and ions. With increasing pressure, the rate decreases due to drop of the densities of copper atoms and ions because of decreasing the kinetic energies of the ions with which they bombard the target. The model indicates that in the gas-aggregation-source relevant pressure regime, the energy flux by ions dominates over the energy fluxes of other plasma species.

NGMD: next generation malware detection in federated server with deep neural network model for autonomous networks

Distributed denial-of-service (DDoS) attacks persistently proliferate, impacting individuals and Internet Service Providers (ISPs). Deep learning (DL) models are paving the way to address these challenges and the dynamic nature of potential threats. Traditional detection systems, relying on signature-based techniques, are susceptible to next-generation malware. Integrating DL approaches in cloud-edge/federated servers enhances the resilience of these systems. In the Internet of Things (IoT) and autonomous networks, DL, particularly federated learning, has gained prominence for attack detection. Unlike conventional models (centralized and localized DL), federated learning does not require access to users’ private data for attack detection. This approach is gaining much interest in academia and industry due to its deployment on local and global cloud-edge models. Recent advancements in DL enable training a quality cloud-edge model across various users (collaborators) without exchanging personal information. Federated learning, emphasizing privacy preservation at the cloud-edge terminal, holds significant potential for facilitating privacy-aware learning among collaborators. This paper addresses: (1) The deployment of an optimized deep neural network for network traffic classification. (2) The coordination of federated server model parameters with training across devices in IoT domains. A federated flowchart is proposed for training and aggregating local model updates. (3) The generation of a global model at the cloud-edge terminal after multiple rounds between domains and servers. (4) Experimental validation on the BoT-IoT dataset demonstrates that the federated learning model can reliably detect attacks with efficient classification, privacy, and confidentiality. Additionally, it requires minimal memory space for storing training data, resulting in minimal network delay. Consequently, the proposed framework outperforms both centralized and localized DL models, achieving superior performance.

Assessment of Suitable Habitat of the Demoiselle Crane (Anthropoides virgo) in the Wake of Climate Change: A Study of Its Wintering Refugees in Pakistan

The inevitable impacts of climate change have reverberated across ecosystems and caused substantial global biodiversity loss. Climate-induced habitat loss has contributed to range shifts at both species and community levels. Given the importance of identifying suitable habitats for at-risk species, it is imperative to assess potential current and future distributions, and to understand influential environmental factors. Like many species, the Demoiselle crane is not immune to climatic pressures. Khyber Pakhtunkhwa and Balochistan provinces in Pakistan are known wintering grounds for this species. Given that Pakistan is among the top five countries facing devastating effects of climate change, this study sought to conduct species distribution modeling under climate change using data collected during 4 years of field surveys. We developed a Maximum Entropy distribution model to predict the current and projected future distribution of the species across the study area. Future habitat projections for 2050 and 2070 were carried out using two representative concentration pathways (RCP 4.5 and RCP 8.5) under three global circulation models, including HADGEM2-AO, BCC-CSM1-1, and CCSM4. The most influential factors shaping Demoiselle Crane habitat suitability included the temperature seasonality, annual mean temperature, terrain ruggedness index, and human population density, all of which contributed significantly to the suitability (81.3%). The model identified 35% of the study area as moderately suitable (134,068 km2) and highly suitable (27,911 km2) habitat for the species under current climatic conditions. Under changing climate scenarios, our model predicted a major loss of the species’ current suitable habitat, with shrinkage and shift towards western–central areas along the Pakistan–Afghanistan boarder. The RCP 8.5, which is the extreme climate change scenario, portrays particularly severe consequences, with habitat losses reaching 65% in 2050 and 85% in 2070. This comprehensive study provides useful insights into the Demoiselle Crane habitat’s current and future dynamics in Pakistan.

Short term forecasting of base metals prices using a LightGBM and a LightGBM - ARIMA ensemble

AbstractBase metals are key materials for various industrial sectors such as electronics, construction, manufacturing, etc. Their selling price is important both for the profitability of the mining and metallurgical companies that produce and trade them, as well as for the countries whose economies rely on their exports or tax revenues as a means for national income. Prices are also critical for companies that use base metals as inputs to fabricate end products. The prediction of prices’ future movements can serve as a tool for risk mitigation and better budget planning. In this study, the logarithmic returns of base metals are forecasted using an autoregressive Light Gradient Boosting Machine (LightGBM) as well as an ensemble comprising the aforementioned algorithm and a classical time series forecasting model (i.e., ARIMA). The two models are then compared to three simpler benchmark models, namely a global mean model, an exponential smoothing model and an ARIMA model. When comparing using RMSE, the autoregressive LightGBM model outperformed the three univariate benchmark models (and the ensemble) for forecasting 6 months ahead for aluminum and nickel returns, while copper and zinc returns were forecasted better by the ensemble. Neither of the proposed models performed better than an ARIMA model when it comes to forecasting lead and tin returns.

The role and deployment timing of direct air capture in Saudi Arabia’s net-zero transition

Abstract The Kingdom of Saudi Arabia (KSA) has pledged to achieve net-zero greenhouse gas (GHG) emissions by 2060. Direct air carbon capture and storage (DACCS) is critical for the country to meet its net-zero target given its reliance on fossil fuels and limited options for carbon dioxide removal (CDR). However, the role of DACCS in KSA’s national climate change mitigation has not been studied in the literature. In this study, we aim to understand the potential role of DACCS and the effect of its deployment timing in KSA’s transition toward its net-zero target using GCAM-KSA, which is a version of Global Change Analysis Model (GCAM) with KSA split out as an individual region. We find that the annual DACCS CO2 sequestration in KSA reaches 0.28 to 0.33 Gt/yr by 2060 depending on its deployment timing. Early DACCS deployment, driven by its early and rapid cost reduction worldwide, could bring significant savings (~420 billion USD during 2020¬¬–2060) in the cost of climate change mitigation in KSA, approximately 17% reduction relative to delayed DACCS deployment. Our study suggests a strong role for KSA to proactively invest in the R&D of DACCS, initiate early DACCS deployment, and explore a broad suite of CDR options.

PODE: privacy-enhanced distributed federated learning approach for origin-destination estimation

The statewide consumer transportation demand model analyzes consumers’ transportation needs and preferences within a particular state. It involves collecting and analyzing data on travel behavior, such as trip purpose, mode choice, and travel patterns, and using this information to create models that predict future travel demand. Naturalistic research, crash databases, and driving simulations have all contributed to our knowledge of how modifications to vehicle design affect road safety. This study proposes an approach named PODE that utilizes federated learning (FL) to train the deep neural network to predict the truck destination state, and in the context of origin-destination (OD) estimation, sensitive individual location information is preserved as the model is trained locally on each device. FL allows the training of our DL model across decentralized devices or servers without exchanging raw data. The primary components of this study are a customized deep neural network based on federated learning, with two clients and a server, and the key preprocessing procedures. We reduce the number of target labels from 51 to 11 for efficient learning. The proposed methodology employs two clients and one-server architecture, where the two clients train their local models using their respective data and send the model updates to the server. The server aggregates the updates and returns the global model to the clients. This architecture helps reduce the server’s computational burden and allows for distributed training. Results reveal that the PODE achieves an accuracy of 93.20% on the server side.

Studying long term relationship between carbon Emissions, Soil, and climate Change: Insights from a global Earth modeling Framework

The persistent increase in greenhouse gas (GHG) emissions, notably carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O), since the mid-20th century has been a key driver of significant climate alterations. This study investigates the complex feedback mechanisms that both influence and are influenced by global climate dynamics, soil processes, and GHG emissions. Our statistical approach incorporates correlation measures, highlighting the limitations of such analyses, namely their inability to confirm causality, sensitivity to outliers, and the exclusive capture of linear relationships. Geographically Weighted Regression (GWR) models reveal spatial variations in the relationship between environmental factors and GHGs, while Path Analysis aids in delineating direct and indirect influences among variables. The research pinpoints significant spatial heterogeneity in the impacts of economic and environmental factors on GHGs, underscoring the necessity of localized strategies for climate change mitigation and sustainable land management. This study also identifies potential threats to agricultural productivity due to soil degradation, which hinder climate adaptation efforts. Our findings advocate for a concerted global response to reduce GHG emissions and address the challenges posed by the interplay of climate change, soil dynamics, and GHG emissions.

Machine learning-enabled weather forecasting for real-time radioactive transport and contamination prediction

Effective responses to nuclear accidents require accurate and fast predictions of radionuclide transport in the atmosphere. However, real-time forecasting has been challenging, since predicting the transport requires meteorological forecasting, which relies on computationally intensive numerical weather simulations. To address this challenge, we have developed statistical emulators for the Weather Research and Forecasting model (WRF) to forecast wind and temperature fields around the source location. The emulator is constructed based on pre-computed WRF simulation results and their boundary conditions from global circulation models (GCM), enabling the rapid prediction of local-scale wind/temperature fields by ingesting the GCM-predicted boundary conditions available online for the next several days. We have explored two methods: (1) deep neural network and (2) a vector autoregressive model with exogenous variables (VARX). A Gaussian plume model is then used to take these field as input and to predict the distribution of radioactive material/dose in the environment. The results show that both emulators are capable of forecasting the wind field for several days in the future in real-time, while not compromising accuracy. The autoregressive model has shown a strong performance of predicting the temperature field, owing to the strong temporal autocorrelations in the temperature data. This framework offers a powerful tool for informed decision-making during emergencies such as enabling meteorology-informed timing of containment venting, and real-time guidance of evacuation strategies.

Sensitivities of Spectral Optical Properties of Dust Aerosols to Their Mineralogical and Microphysical Properties

AbstractDust aerosol is a key component in global radiative forcing and climate modeling. We developed a mineralogy‐resolved spectral optical property model for dust aerosols. The model incorporates nine mineral groups and applies the effective medium methods to obtain the spectral complex refractive indices of inhomogeneous dust particles. This study considers 5,000 sets of dust mineral composition mixtures along with the use of the bimodal particle size distribution. We calculate the bulk properties of these samples in a wide spectral range from 0.2 to 50 μm using a comprehensive database of hexahedral dust single‐scattering properties. Through sensitivity analyses, this study reveals that the illite and hematite components substantially affect the optical properties of dust. Furthermore, the effective radii of both the fine‐mode and the coarse‐mode dust particles have significant impacts on the bulk optical properties, albeit to varying degrees.

Fine Cloud Structures on a Hard Snowball Earth

AbstractEarth may have been globally ice‐covered at least two times during the Cryogenian Period (720–635 Ma). Previous studies showed that clouds could strongly warm a hard snowball Earth and promote its deglaciation. However, the understanding of clouds was largely based on coarse‐resolution global simulations with parameterized convection and clouds, or high‐resolution cloud‐resolving simulations with explicit convection and clouds but limited to a small domain without the effects of large‐scale circulation. Here we present the first global non‐hydrostatic high‐resolution (30 km) simulation to investigate snowball clouds and their radiative effects. We show that spatial‐temporal cloud distributions exhibit clear meanders (patterns of curved lines rather than straight lines along the west‐east direction) and strong variabilities, which result from the tight interactions among the Inter‐Tropical Convergence Zone, Hadley cells, and baroclinic instability. The net cloud radiative effect is around 20 W m−2 in the tropics, consistent with previous global general circulation model simulations.

PFedBASC: Personalized Federated Learning with Blockchain-Assisted Semi-Centralized Framework

As network technology advances, there is an increasing need for a trusted new-generation information management system. Blockchain technology provides a decentralized, transparent, and tamper-proof foundation. Meanwhile, data islands have become a significant obstacle for machine learning applications. Although federated learning (FL) ensures data privacy protection, server-side security concerns persist. Traditional methods have employed a blockchain system in FL frameworks to maintain a tamper-proof global model database. In this context, we propose a novel personalized federated learning (pFL) with blockchain-assisted semi-centralized framework, pFedBASC. This approach, tailored for the Internet of Things (IoT) scenarios, constructs a semi-centralized IoT structure and utilizes trusted network connections to support FL. We concentrate on designing the aggregation process and FL algorithm, as well as the block structure. To address data heterogeneity and communication costs, we propose a pFL method called FedHype. In this method, each client is assigned a compact hypernetwork (HN) alongside a normal target network (TN) whose parameters are generated by the HN. Clients pull together other clients’ HNs for local aggregation to personalize their TNs, reducing communication costs. Furthermore, FedHype can be integrated with other existing algorithms, enhancing its functionality. Experimental results reveal that pFedBASC effectively tackles data heterogeneity issues while maintaining positive accuracy, communication efficiency, and robustness.

On the frozen F-theory landscape

We study 6d \\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\mathcal{N}$$\\end{document} = (1, 0) supergravity theories arising in the frozen phase of F-theory. For each of the known global models, we construct an F-theory compactification in the unfrozen phase with an identical non-abelian gauge algebra and massless matter content. Two such low energy effective theories are then distinguished through gauge enhancements in moduli space. We study potentially new global models obtained via compact embeddings of a plethora of 6d \\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\mathcal{N}$$\\end{document} = (1, 0) superconformal field theories and little string theories constructed using frozen 7-branes. In some cases, these provably do not exist, and in other cases, we explicitly construct a compact embedding, yielding 6d supergravity theories with new massless spectra. Finally, by using gravitational anomaly cancellation, we conjecture the existence of localized neutral hypermultiplets along frozen 7-brane loci.

Future malaria environmental suitability in Africa is sensitive to hydrology

Changes in climate shift the geographic locations that are suitable for malaria transmission because of the thermal constraints on vector Anopheles mosquitos and Plasmodium spp. malaria parasites and the lack of availability of surface water for vector breeding. Previous Africa-wide assessments have tended to solely represent surface water using precipitation, ignoring many important hydrological processes. Here, we applied a validated and weighted ensemble of global hydrological and climate models to estimate present and future areas of hydroclimatic suitability for malaria transmission. With explicit surface water representation, we predict a net decrease in areas suitable for malaria transmission from 2025 onward, greater sensitivity to future greenhouse gas emissions, and different, more complex, malaria transmission patterns. Areas of malaria transmission that are projected to change are smaller than those estimated by precipitation-based estimates but are associated with greater changes in transmission season lengths.

A comparison of global flood models using Sentinel-1 and a change detection approach

AbstractAdvances in numerical algorithms, improvement of computational power and progress in remote sensing have led to the development of global flood models (GFMs), which promise to be a useful tool for large-scale flood risk management. However, performance and reliability of GFMs, especially in data-scarce regions, is still uncertain, as they are difficult to validate. Here we aim at contributing to develop alternative, more flexible, and consistent methods for GFM validation by applying a change detection analysis on synthetic aperture radar (CD-SAR) imagery obtained from the Sentinel-1 imagery, on a cloud-based geospatial analysis platform. The study addresses two main objectives. First, to validate four widely adopted GFMs with flood maps generated through the proposed CD-SAR approach. This exercise was conducted for eight different large river basins on four continents, to account for a diverse range of hydro-climatic environments. Second, to compare CD-SAR-derived flood maps with those obtained from alternative remote sensing sources. These comparative results offer valuable insights into the reliability of CD-SAR data as a validation tool, more specifically how it stacks up against flood maps generated by other remote sensing techniques.

Tectonic processes and the evolution of the continental crust

The Earth is the only known planet where plate tectonics operates. We review features of Archaean and early Proterozoic geology that constrain tectonic environments and inform discussions of the onset of plate tectonics. There is the question of scale, and how the results of individual case studies are put into a wider global context. Global models may be difficult to test, and we seek to integrate evidence for plate tectonics being active with ancient records of subduction. We explore evidence for when the continental crust became rigid enough to facilitate plate tectonics, based on the occurrence of widespread dyke swarms and large sedimentary basins, relatively widespread granulite facies metamorphism, and evidence for crustal thickening. We argue that it remains difficult to constrain tectonic settings from contemporaneous metamorphic events without spatial control. Archaean cratons stabilised at different times in different areas from 3.1–2.5 Ga, juvenile continental crust changed from mafic to more intermediate compositions, there was a reduction in crustal growth at ca. 3 Ga, and increasing evidence for lateral movement of crustal fragments. These with the other changes at the end of the Archaean are taken to reflect the onset of plate tectonics as the dominant global regime.

Federated 3D multi-organ segmentation with partially labeled and unlabeled data.

This paper considers a new problem setting for multi-organ segmentation based on the following observations. In reality, (1) collecting a large-scale dataset from various institutes is usually impeded due to privacy issues; (2) many images are not labeled since the slice-by-slice annotation is costly; and (3) datasets may exhibit inconsistent, partial annotations across different institutes. Learning a federated model from these distributed, partially labeled, and unlabeled samples is an unexplored problem. To simulate this multi-organ segmentation problem, several distributed clients and a central server are maintained. The central server coordinates with clients to learn a global model using distributed private datasets, which comprise a small part of partially labeled images and a large part of unlabeled images. To address this problem, a practical framework that unifies partially supervised learning (PSL), semi-supervised learning (SSL), and federated learning (FL) paradigms with PSL, SSL, and FL modules is proposed. The PSL module manages to learn from partially labeled samples. The SSL module extracts valuable information from unlabeled data. Besides, the FL module aggregates local information from distributed clients to generate a global statistical model. With the collaboration of three modules, the presented scheme could take advantage of these distributed imperfect datasets to train a generalizable model. The proposed method was extensively evaluated with multiple abdominal CT datasets, achieving an average result of 84.83% in Dice and 41.62mm in 95HD for multi-organ (liver, spleen, and stomach) segmentation. Moreover, its efficacy in transfer learning further demonstrated its good generalization ability for downstream segmentation tasks. This study considers a novel problem of multi-organ segmentation, which aims to develop a generalizable model using distributed, partially labeled, and unlabeled CT images. A practical framework is presented, which, through extensive validation, has proved to be an effective solution, demonstrating strong potential in addressing this challenging problem.

Spatial distribution and environmental/biological co-regulation mechanism of dimethyl sulfur compounds in the eastern Indian Ocean

Dimethyl sulfur compounds including dimethylsulfoniopropionate (DMSP), dimethyl sulfide (DMS), and dimethyl sulfoxide (DMSO), play a crucial part in global sulfur cycling. The eastern Indian Ocean (EIO), characterized by its remarkable diversity of biomes and climate dynamics, is integral to global climate regulation. However, the regulation mechanism of DMS (P, O) in the EIO remains to be elucidated in detail. This paper presented a field survey aimed at investigating the spatial distribution of DMS (P, O) and their relationships with environmental and biological factors in the EIO. The surface concentrations of DMS, DMSPt, and DMSOt varied from 0.07 to 7.37 nmol/L, 0.14 to 9.17 nmol/L, and 0.15 to 3.32 nmol/L, respectively, and their distributions are attributed to high Chl-a concentration near Sri Lanka and the influence of ocean currents (Wyrtki jets, Bay of Bengal runoff). Higher concentrations of DMS (P) and DMSOt were predominantly observed in water columns shallower than 75m and deeper than 75m deep, respectively. The monthly DMS fluxes in the study area peaked in August. Temperature and Dissolved Silica Index (DSI) were the key environmental determinants for DMS distribution, while nitrate (NO3-) was the primary factor for both DMSPt and DMSOt. In terms of biological factors, Prochlorococcus and Synechococcus were significant contributors to DMS (P, O) dynamics. Synechococcus was the dominant influence on the DMS source and DMSPt sink, whereas Prochlorococcus primarily consumed DMSOt. Furthermore, the structural equation modeling (SEM) revealed the relationship between DMS, DMSPt, DMSOt, and the key environmental/biological factors, as well as among them, and together they formed a co-regulatory network in the EIO. This contributes significantly to the advancement of global ecosystem models for DMS (P, O).

Leveraging regional mesh refinement to simulate future climate projections for California using the Simplified Convection-Permitting E3SM Atmosphere Model Version 0

Abstract. The spatial heterogeneity related to complex topography in California demands high-resolution (< 5 km) modeling, but global convection-permitting climate models are computationally too expensive to run multi-decadal simulations. We developed a 3.25 km California climate modeling framework by leveraging regional mesh refinement (CARRM) using the U.S. Department of Energy (DOE)'s global Simple Cloud-Resolving E3SM Atmosphere Model (SCREAM) version 0. Four 5-year time periods (2015–2020, 2029–2034, 2044–2049, and 2094–2099) were simulated by nudging CARRM outside California to 1° coupled simulation of E3SMv1 under the Shared Socioeconomic Pathways (SSP)5-8.5 future scenario. The 3.25 km grid spacing adds considerable value to the prediction of the California climate changes, including more realistic high temperatures in the Central Valley and much improved spatial distributions of precipitation and snowpack in the Sierra Nevada and coastal stratocumulus. Under the SSP5-8.5 scenario, CARRM simulation predicts widespread warming of 6–10 °C over most of California, a 38 % increase in statewide average 30 d winter–spring precipitation, a near-complete loss of the alpine snowpack, and a sharp reduction in shortwave cloud radiative forcing associated with marine stratocumulus by the end of the 21st century. We note a climatological wet precipitation bias for the CARRM and discuss possible reasons. We conclude that SCREAM RRM is a technically feasible and scientifically valid tool for climate simulations in regions of interest, providing an excellent bridge to global convection-permitting simulations.

Quantifying the role of termite decomposition in a mesic savanna

AbstractSavanna systems are among the most sensitive to future climate and land‐use change, yet we lack robust, direct quantifications of savanna carbon cycling. Together with fire, decomposition is the main process by which the carbon and nutrients are recycled and made available again to plants. Decomposition is largely mediated by microbes and soil invertebrates. Using a novel large‐scale termite suppression experiment, we quantify, for the first time, the relative contribution of microbes, termites, and other invertebrates to the decomposition of wood (fresh native and dry non‐native), dry dung, and grass in a mesic savanna. We found that termites were responsible for two thirds of the mass loss from dry wood and a third of the mass loss from fresh native wood, dry dung, and dry grass. Microbes were wholly responsible for the difference as there was no evidence of other invertebrates contributing to decomposition, even with fresh wood. Using multiple substrates in savanna decomposition studies is important where a mixture of contrasting life forms occur because both the rates of decomposition and the dominant agent varied considerably. In addition, including both a dry non‐native and fresh native wood cast light on possible explanatory variables such as wood density, green‐ness and the presence of bark, and the necessity of teasing these variables apart in future studies. Termites stand apart from all other insects in their impact on decomposition within savannas and should be acknowledged alongside microbes and fire as the primary agents of wood, grass, and dry dung turnover in global carbon models.