Science China Earth Sciences最新文献

The stable hydrogen isotope in precipitation is an effective environmental tracer for climatic and hydrologic studies. However, accurate and high-precision precipitation hydrogen isoscapes are currently unavailable in China. In this study, a data fusion method based on Convolutional Neural Networks (CNN) is used to fuse the hydrogen isotopic composition (δ²H_p) of observations and isotope-equipped general circulation model (iGCM) simulations. A precipitation hydrogen isoscape with a temporal resolution of monthly and a spatial resolution of 50–60 km is established for East China for the 1969–2017 period. Prior to building the isoscape, the performance of three data fusion methods (DFMs) and two bias correction methods (BCMs) is compared. The results indicate that the CNN fusion method performs the best with a correlation coefficient larger than 0.90 and root mean square error smaller than 10.5‰when using observation as a benchmark. The fusion methods based on back propagation and long short-term memory neural network perform similarly, while slightly outperforming the bias correction methods. Thus, the CNN method is used to generate the hydrogen isoscape, and the temporal and spatial distribution characteristics of the hydrogen isotope in precipitation are analyzed based on this dataset. The generated isoscape shows similar spatial and temporal distribution characteristics to observations. In general, the distribution pattern of δ²H_p is consistent with the temperature effect in northern China, and consistent with the precipitation amount effect in southern China. The trend of the δ²H_p time series is consistent with that of observed precipitation and temperature. Overall, the generated isoscape effectively reproduces the observations, and has the characteristics of time continuity and relative spatial regularity, which can provide valuable data support for tracking atmospheric and hydrological processes.

Millet agriculture, which originated in northern China, alongside rice agriculture, have nurtured the Chinese civilization. Prehistoric manuring practices likely promoted and maintained sustainable millet agricultural production in the loess area of northern China. However, ongoing controversy exists regarding the indicators of prehistoric manuring intensity of foxtail millet (Setaria italica) and common millet (Panicum miliaceum). Here, we present the results of pot and field fertilization experiments on two millet types. Our findings suggest that manuring can significantly increase the δ¹⁵N values of foxtail millet, and its δ¹⁵N values increase with increasing manuring levels. The δ¹⁵N values of foxtail millet leaves are systematically greater than those of grains by approximately 1.6‰. Manuring could have a long-term residual impact on increasing the δ¹⁵N values of soil and millet crops. Combined with previous crop fertilization experiment results, we propose that the impact of manuring on the δ¹⁵N values of non-nitrogen-fixing crops is roughly consistent. The δ¹⁵N value and amount of manure are key factors determining the extent of change in plant δ¹⁵N values. The millet grain δ¹⁵N values can serve as reliable indicators of manuring practices. Finally, we provide an interpretive framework for assessing the correlation between manuring levels and the δ¹⁵N values of archaeological millet remains. The δ¹⁵N values of ancient millet grains suggest widespread and intensive manuring practices in prehistoric millet agriculture in northern China, spanning from the early Yangshao period to the Longshan period.

Mineral prospectivity mapping (MPM) is designed to reduce the exploration search space by combining and analyzing geological prospecting big data. Such geological big data are too large and complex for humans to effectively handle and interpret. Artificial intelligence (AI) algorithms, which are powerful tools for mining nonlinear mineralization patterns in big data obtained from mineral exploration, have demonstrated excellent performance in MPM. However, AI-driven MPM faces several challenges, including difficult interpretability, poor generalizability, and physical inconsistencies. In this study, based on previous studies, we devised a novel workflow that aims to constructing more transparent and explainable artificial intelligence (XAI) models for MPM by embedding domain knowledge throughout the AI-driven MPM, from input data to model design and model output. This newly proposed approach provides strong geological and conceptual leads that guide the entire AI-driven MPM model training process, thereby improving model interpretability and performance. Overall, the development of XAI models for MPM is capable of embedding prior and expert knowledge throughout the modeling process, presenting a valuable and promising area for future research designed to improve MPM.

The middle and lower reaches of the Yangtze River, a primary region for freshwater lakes in China, have undergone significant transformations throughout the Holocene. These changes, driven by factors such as sea-level rise, climate change, and human activities, have led to the progressive elevation of water levels in this area. As a result, a floodplain has emerged, characterized by the formation of numerous shallow lakes along the river course. However, the pattern of water-level changes in the main channel of the Yangtze River during the Holocene remains unclear. This gap in knowledge poses challenges for understanding sediment transport dynamics, the interactions between the river and its adjacent lakes, and the prevention and control of flood disasters in the Yangtze River basin. To shed light on these issues, our study compiled data on the surface elevation and water depth of 81 lakes in the mid-lower reaches of the Yangtze River basin. Additionally, we analyzed historical water-level records from the 1900s to the 1970s at eight gauging stations from Shashi to Jiangyin along the river’s main stream. Our findings reveal that, particularly along the Jingjiang section, the basal elevation of most lakes is lower than the Yangtze River’s water level during the dry season. Conversely, the water level of the main stream exceeds that of both the floodplain and the lakes enclosed by the Jingjiang embankment. In the tidal reach, especially within the Taihu Lake basin, the basal elevation of lakes typically falls below sea level. Meanwhile, lakes located along the section from Chenglingji to Wuhu exhibit basal elevations that correspond with the Yangtze River’s annual average and dry season water levels. Given the widespread presence of lakes along the middle and lower reaches of the Yangtze River, our study introduces a new proxy for reconstructing the mean water level of the mid-lower Yangtze River in the Holocene. By analyzing sediments from Nanyi Lake and Chenyao Lake in the lower Yangtze River, we attempted to reconstruct the water level of the Yangtze River’s main channel since 8 ka BP.

The Ross, Filchner-Ronne, and Amery ice shelves are the three largest ice shelves in Antarctica, playing a crucial role in supporting the Antarctic ice sheet. However, current studies on the stability of the three largest ice shelves primarily focus on singular or limited factors, lacking a comprehensive assessment of multiple parameters. To systematically and in-depth study the stability and trend of the three largest ice shelves, we comprehensively collected and analyzed key parameters, including elevation changes, basal melting, surface meltwater, major rifts propagation rate, suture zones, ice front area change rate, grounding lines, ice velocity, and mass balance. Additionally, we selected the collapsed Larsen B Ice Shelf (LBIS), the rapidly changing and structurally weakened Pine Island Ice Shelf (PIIS), and the accelerating Totten Ice Shelf (TIS) as reference ice shelves. By comparing and analyzing the key parameters between these reference ice shelves and the three largest ice shelves, we find the status and trends in the stability of the latter. Our findings reveal that most key parameters of the three largest ice shelves present relatively minor variations compared to those of the reference ice shelves. Specifically, 50% of the parameters are smaller than those of the accelerating TIS, 88% are smaller than those of the rapidly changing PIIS, and all parameters are smaller than those of the collapsed LBIS. Furthermore, after analyzing parameters that are not smaller than those of the TIS, it is observed that they remain in a stable state. Hence, the three largest ice shelves are currently undergoing natural changes that do not threaten their stability in the short term. Nevertheless, the evolution of the ice shelves under global climate change remains uncertain, making long-term observation and monitoring essential to assess their impact on sea level rise.

As the western boundary of the Sichuan-Yunnan block (SYB), the Red River fault (RRF) is a major fault that controls deep crustal movement and deformation in the southeast margin of the Tibetan Plateau and regulates middle-lower crustal flow. Geophysical data suggest that the RRF is segmented and exhibits distinct variations in seismicity, velocity structure and crustal deformation from north to south. Seismic anisotropy reveals a complex pattern of lateral spatial and vertical stratified distributions. (1) From the perspective of crustal stratification, in the upper crust, the fast wave polarization in the north segment of the RRF is complex and possibly influenced by the Sanjiang lateral collision zone and adjacent faults with varying strikes. The fast wave polarization in the middle segment is in the NW-SE direction, indicating a localized area of closed down or locked up with consistent deformation. And in the south segment, it presents a disordered pattern, signifying complex deep tectonics and stress conditions at the wedged intersection zone. In the middle-lower crust in the north and south segments of the RRF, the azimuthal anisotropy is strong and consistent with the spatial strike of the weak zone characterized by low-velocity and high-conductivity. This suggests a connection between the anisotropy and the material migration. (2) In the whole crustal scale, the fast wave directions in two sides of the RRF are consistent with the NW-SE tectonic strike. It indicates that the RRF, as a large fault potentially cutting through the whole crust, strongly controls the surrounding media. (3) In the lithospheric scale, the fast wave polarizations are oriented nearly E-W and independent of the fault strike, consistent with the low P- and S-wave velocity structures and positive radial anisotropy in the upper mantle. The fast wave directions could be related to lithospheric olivine deformation and asthenospheric flow. This paper suggests a decoupling of deformation between the crust and the lithospheric mantle in the south of approximately 26°20′N near the RRF, which can potentially be attributed to the subduction and rollback of the Indian plate. Based on various geophysical observations and inversions, we can determine the detailed anisotropic structure in the crust and the upper mantle around the RRF. Denser geophysical arrays and more accurate records can be used to explore the intricate anisotropy in segmentation and stratification around the RRF, enhancing the understanding of its tectonic significance.

The North Pacific sea surface salinity (SSS) decadal variability (NPSDV) and its potential forcing were evaluated from 25 coupled models of the Coupled Model Intercomparison Project phase 6 (CMIP6) considering the prospects for decadal climate predictions. The results indicated that the CMIP6 models generally reproduced the spatial patterns of NPSDV. The large standard deviation of the SSS anomaly over the strong current regions, such as the Kuroshio-Oyashio Extension (KOE), North Pacific Current (NPC), California Current System (CCS), and Alaskan Coastal Current (ACC), is reflected in the two leading modes of NPSDV: a dipole with out-of-phase loadings in the KOE-NPC versus CCS-ACC and a monopole with positive loading over the KOE-NPC. The order of modes is sensitive to individual models that exhibit discrepancies, especially in temporal phases and power spectra. An autoregressive model of order-1 was used to reconstruct the NPSDV with several forcing terms. The generally weaker influence of forcings in an autoregressive model of order-1 is partly related to the overestimated response time of NPSDV relative to forcings. Most NPSDV variances originate from the persistence of SSS anomalies, but the dominant forcing factors are diverse among models. The model diversity for the NPSDV simulation mainly arises from the influence of the tropical El Niño-Southern Oscillation through teleconnection on the North Pacific Oscillation or Aleutian Low with timescale dependence. Conversely, models that can reproduce the NPSDV well are not dependent on those with larger impacts from the North Pacific oceanic processes.

The evolution of solar magnetic fields is significant for understanding and predicting solar activities. And our knowledge of solar magnetic fields largely depends on the photospheric magnetic field. In this paper, based on the spherical harmonic expansion of the photospheric magnetic field observed by Wilcox Solar Observatory, we analyze the time series of spherical harmonic coefficients and predict Sunspot Number as well as synoptic maps for Solar Cycle 25. We find that solar maximum years have complex short-period disturbances, and the time series of coefficient g₇⁰ is nearly in-phase with Sunspot Number, which may be related to solar meridional circulation. Utilizing Long Short-Term Memory networks (LSTM), our prediction suggests that the maximum of Solar Cycle 25 is likely to occur in June 2024 with an error of 8 months, the peak sunspot number may be 166.9±22.6, and the next solar minimum may occur around January 2031. By incorporating Empirical Mode Decomposition, we enhance our forecast of synoptic maps truncated to Order 5, validating their relative reliability. This prediction not only addresses a gap in forecasting the global distribution of the solar magnetic field but also holds potential reference value for forthcoming solar observation plans.