Si-Liang Li

Greenhouse gas emissions from temporarily flooded upland croplands remain poorly characterized in the global greenhouse gas budget. We combined field measurements from 11 maize‐field sites temporarily flooded across the North China Plain with a global synthesis of 23 flooded upland cropland sites to quantify diffusive emissions. Flooded croplands release high fluxes of nitrous oxide, carbon dioxide, and methane, with emission factors exceeding relevant Intergovernmental Panel on Climate Change (IPCC) defaults for rice paddies and newly flooded lands. Extrapolating site‐level fluxes from limited observations over the estimated global area of flooded upland cropland and mean flooding duration yields a mean of 63.3 teragrams (Tg) of carbon dioxide‐equivalent (CO ₂ ‐eq) emissions per year during the standing‐water flooding periods, comprising 7.2 (1.7–17.2), 34.6 (9.0–83.7), and 21.5 (6.2–45.6) Tg from nitrous oxide, carbon dioxide, and methane, respectively. Temporarily flooded upland croplands are therefore identified as a previously overlooked component of the global greenhouse gas budget, providing an empirical basis for incorporating flood‐driven emissions into future agricultural inventories.

Large methane (CH4) emissions during ice-melt periods have been frequently reported in lakes, but whether similar processes occur in ponds remains largely unknown. Here, we conducted observations in a temperate pond across two consecutive ice seasons (2022–2024), with daily sampling during ice melt, to characterize ice-melt CH4 dynamics, reveal potential CH4 production pathways, and estimate the contribution of ice-melt emissions to the annual CH4 budget. CH4 concentrations increased sharply prior to ice melt, resulting in a diffusive pulse during the ice-melt periods (20.1 ± 14.1 mg m–2 h–1), approximately 11 times higher than that during observed ice-free periods (1.8 ± 2.5 mg m–2 h–1). In contrast, CH4 ebullition during ice melt remained low (0.21 ± 0.15 mg m–2 h–1). Despite the brief duration (∼7 days), ice-melt CH4 emissions accounted for 8.7% of annual total CH4 emissions (diffusion + ebullition) and 22.3% of annual diffusive CH4 emissions from June 2022 to May 2023. Variations in water chemistry and sediment microbial communities suggest that both anaerobic and aerobic CH4 production may have contributed to these emission pulses. Particularly, CH4 concentration increased concurrently with chlorophyll a and dissolved oxygen before ice melt, suggesting possible phytoplankton-derived CH4 production under oxic conditions. Given the ubiquity of seasonal freeze-melt cycles in temperate ponds, incorporating ice-melt CH4 fluxes is necessary to refine global CH4 budget estimates.

Hydropower reservoirs, long considered as a clean energy solution, are increasingly recognized as major greenhouse gas (GHG) emissions hotspots. However, over 97% of these hydropower reservoirs remain unmonitored, creating major uncertainties in global GHG budgets. Here, machine learning models were employed to predict GHG fluxes from hydropower reservoirs lacking reported direct GHG measurements using 67 widely accessible environmental predictors covering climate, land use, soil properties, geology, and topography. The predicted GHG fluxes revealed marked spatiotemporal heterogeneity, characterized by distinct monthly variability and emission hotspots predominantly concentrated in tropical and arid zones, particularly across Africa. Moreover, scaling the predictions to 5,293 unmonitored hydropower reservoirs yielded total emissions of 304 ± 37 Tg CO ₂ eq year ⁻¹ , enabling a more geographically representative assessment of global reservoir GHG emissions. Methane notably dominates emissions (55.3%), followed by carbon dioxide (41.3%) and nitrous oxide (3.4%), with the latter often overlooked but gaining relevance during increasing eutrophication. In addition, our gas‐specific models identified temperature‐related variables as the primary drivers of GHG dynamics, emphasizing the central role of thermal regimes, especially in African reservoirs. Our findings call for a re‐evaluation of the climate costs of hydropower and suggest that future warming may further amplify reservoir‐based GHG emissions. The predictive capacity of our models provides a scalable tool for refining global GHG inventories and guiding low‐carbon hydropower development under a changing climate.

Automatic classification of remote sensing images for land use and land cover (LULC) applications encounter numerous challenges because of multi-resolution data, heterogeneous appearance and multi-spectral complexity. Existing deep learning models, especially conventional CNNs, struggle to include inherent features of objects because of their limited ability in modeling the long-range dependency and global contextual information. To address these limitations, this paper proposes a novel Heterogeneous Dual-Path Wide Residual Network (HDP-WRN) model for enhanced LULC classification. The initial feature extraction is performed using a post-activation residual block and two parallel paths using different convolutional operators with different attention mechanisms. Branch A integrates standard convolutions and Efficient Channel Attention (ECA) to extract fine-grained local textures, and Branch B merges Ghost convolutions and Convolutional Block Attention Module (CBAM) to boost global spatial channel feature extraction effectively and their complementary representations are fused for obtaining discriminative joint feature space effectively. Extensive experiments show that HDP-WRN achieves competitive performance on multiple benchmark datasets, with accuracy of 98.67% on EuroSAT, 95.11% on RSSCN, 96.13% on SIRI-WHU, and 97.24% on UC-Merced. Under our experimental setup, the proposed model outperforms the reported results of several existing methods, though direct comparisons should be interpreted with caution due to differences in training protocols across studies. Results verify the model for accurate and robust LULC classification for diverse spatial resolutions and remote sensing images.

Inland waters constitute vital components in the global carbon (C) cycle. Nevertheless, the regional patterns of aquatic carbon transfer across China remain poorly quantified, and their impacts on the terrestrial carbon budget remain unclear. Here, we quantified the three primary aquatic carbon output fluxes and evaluated their implications for China's terrestrial carbon budget. We showed that the total aquatic carbon output flux was 165 ± 20 Tg C yr ⁻¹ , counteracting 29% of China's terrestrial net ecosystem productivity (NEP). Carbon dioxide (CO ₂ ) emissions from river networks (86 ± 19 Tg C yr ⁻¹ ) dominated the aquatic carbon transfer, accounting for 52% ± 13% of China's aquatic carbon output flux. Carbon burial in reservoirs (20.2 ± 3.4 Tg C yr ⁻¹ ) represented a significant anthropogenic redistribution of terrestrial carbon equivalent to 46% ± 8% of the downstream carbon export (43.6 ± 1.8 Tg C yr ⁻¹ ) across China. Furthermore, the positive correlations between aquatic carbon yield and precipitation and terrestrial NEP indicated a direct climatic control on aquatic carbon dynamics, whereas human activities have amplified these effects. Considerable spatial variations were observed in the net landscape carbon balance (NLCB) across China. Specifically, the NLCB in low‐productivity regions was sensitive to aquatic carbon transfer, whereas terrestrial ecosystem productivity exerted the dominant control in high‐productivity ecosystems. Our findings elucidate the spatial patterns and regulatory mechanisms governing aquatic carbon transfer across China, establishing a robust framework to refine regional carbon balance assessments and inform targeted climate management strategies.

Inland waters are significant sources of atmospheric nitrous oxide (N ₂ O), and their emissions are expected to rise rapidly due to human activities. However, current estimates remain highly uncertain, partly because sparse field measurements and existing models fail to capture fine‐scale variations in N ₂ O emissions, especially due to the interactions between fluid dynamics and biogeochemical drivers. This study developed an N ₂ O module in the Environmental Fluid Dynamics Code (EFDC), a comprehensive numerical model capable of simulating hydrodynamics and water quality in different inland aquatic systems, by incorporating key nitrogen (N) processes that influence N ₂ O production and emissions. The developed model was applied to a typical mountainous reservoir in China, and reliably reproduced the temporal and spatial variations in N ₂ O concentrations and fluxes. Seasonal patterns of N ₂ O fluxes were primarily driven by temperature, stratification, and nutrient availability, with autumn exhibiting the peak values. In contrast, spatial variations were mainly regulated by fluid dynamics and nutrient availability, with the highest N ₂ O fluxes observed in the main part of the reservoir. Furthermore, the whole‐reservoir N ₂ O budget indicated that sediments were the primary source of N ₂ O, which is mainly produced via denitrification under low dissolved oxygen conditions, underscoring the importance of understanding the sedimentary N cycle. Overall, our model enables the quantification of significant spatiotemporal variations in N ₂ O emissions from inland waters and unravels the underlying mechanisms, thereby enhancing our mechanistic understanding of N ₂ O dynamics and improving the quantification accuracy of N ₂ O emissions at broader scales.

China is the largest emitter of cropland gaseous reactive nitrogen (Ngr, including NH ₃ , N ₂ O, and NOx), which greatly affects regional air quality, climate change, and human health. Despite the substantial spatial and temporal variations in cropland Ngr emissions, their driving mechanisms and net climate change forcings (either cooling or warming) remain unclear. Using revised emission estimates and bottom–up mass flow-based approaches, we evaluated the spatiotemporal changes in and key drivers of China’s cropland Ngr emissions from 1978 to 2023. We assessed the relative contribution of China’s cropland Ngr emissions to climate change via multiple metrics (global warming and temperature potentials). China’s cropland Ngr emissions increased by approximately 2.3-fold from 1978 to 2023. The pace of this increase decelerated over time, from 0.17 Tg N·y ⁻² before 1996 to 0.15 Tg N·y ⁻² during 1997–2005 and −0.05 Tg N·y ⁻² after 2005. Population growth and N-intensive animal-sourced food production accounted for 31.7% and 23.2% of this historical emission change, respectively. The net climate change forcing of Ngr varies with scale. Currently, China’s cropland Ngr emissions in 2017 at the county level impart a net cooling effect ascribed to short-lived NH ₃ and NOx emissions, but this forcing shifts in sign and magnitude to a warming effect over a one-hundred-year time horizon, primarily through long-lived N ₂ O emissions. Our findings suggest that a combination of metrics can comprehensively and adequately capture short-, medium-, and long-term climate impacts and that achieving climate targets will conspire against Ngr mitigation efforts for environmental protection goals.

The contribution of terrestrial dissolved organic matter (DOM) to the ocean has been an enigma for decades. Tracking terrestrial DOM in the ocean has proven challenging due to factors such as the instability of terrestrial biomarkers, indistinguishable carbon isotopes from biogeochemical fractionation, and similar chemical composition between terrestrial and oceanic DOM. Here, we show that the terrestrial contribution to oceanic organic carbon pools is 1.7 to 2.5 times higher than previously assumed, highlighting the need to adjust global carbon cycle models. We derive these estimates by bridging high-performance liquid chromatography with ultra-high resolution mass spectrometry to investigate the presence of terrestrial molecules that are transported from rivers to the ocean and estimate their contribution to oceanic DOM. We identified 269 molecular formulae (MF) that are likely transported from land to the ocean. These formulae exhibited resistance to biological and photochemical degradation in incubation experiments, and were widely distributed in global rivers, marginal seas, and open oceans, suggesting that they are ubiquitous in inland and ocean waters and have a similar source. By relating the abundances of terrestrially derived MF to dissolved organic carbon concentrations, and radiocarbon measurements, we estimated that 16.7 to 25.0% of oceanic DOM is likely derived from rivers.

The response of fluvial nitrate (NO₃^-) export to climate warming remains poorly constrained in high-elevation basins. Here, we integrate multi-decadal NO₃^- flux records with multi-year measurements of NO₃^- concentrations and dual-isotope signatures (δ¹⁵N-NO₃^- and δ¹⁸O-NO₃^-) from two contrasting rivers on the Qinghai-Tibetan Plateau, the permafrost-rich Jinsha River (JSR) and the permafrost-scarce Yalong River (YLR), to evaluate climate-driven shifts in NO₃^- dynamics. In the JSR, where sustained warming has driven widespread permafrost thaw, NO₃^- concentrations and fluxes increased by ∼190 and ∼220%, respectively, from 1993 to 2018. In contrast, the NO₃^- export in the YLR was primarily controlled by hydrological variability rather than temperature. In-soil nitrification is the dominant source of fluvial NO₃^-, with warming-enhanced nitrification emerging as the primary driver of increased NO₃^- flux, while runoff plays a minor role. Using the space-for-time substitution method, we infer that ongoing permafrost degradation may shift NO₃^- export from warming-driven to hydrology-driven controls. This transition is expected to enhance nitrogen leaching under warming and intensified precipitation, contributing to nitrogen limitation in high-elevation vegetation and weakening of CO₂ uptake. These findings reveal a previously underappreciated climate-warming nitrogen-carbon feedback with implications for Earth system models in permafrost-affected regions.