Earth and Space Science最新文献

Accurate aerosol optical depth (AOD) prediction remains challenging due to complex aerosol-radiation interactions and highly variable spatio-temporal patterns. Three critical scientific issues motivate this work: understanding whether and how physical principles can enhance deep learning predictions, identifying which aerosol properties most strongly govern AOD variations, and improving the prediction of extreme AOD events critical for air quality management. Herein, utilizing MERRA-2 reanalysis data (1980–2024) over the Huaihe River Basin in eastern China, a Physics-Guided deep learning framework is presented for Aerosol Optical Depth (AOD) prediction. The model proposed integrates Convolutional Neural Networks (CNN), Long Short-TermMemory (LSTM) networks, and multi-head attention mechanisms to capture both spatio-temporal features and physical relationships of aerosol properties. Three key aspects are involved: First, a hybrid deep learning model is developed and evaluated, which combines CNNs for spatial correlation extraction, bidirectional LSTM for temporal dependency modeling, and multi-head attention for feature interaction learning. Second, a comprehensive feature importance analysis is conducted by examining the relationships between different aerosol properties (mass concentration, scattering coefficient, and Ångström exponent) and AOD prediction, offering physical insights into the model's decision-making process. Third, a specialized approach is proposed for extreme AOD event prediction, focusing on early detection and accurate forecasting of high-AOD episodes. Overall, the results demonstrate the model's efficacy in capturing both regular AOD variations and extreme events, with the Physics-Guided architecture showing superior performance compared to traditional methods. This integrated approach enhances AOD prediction accuracy and deepens insights into aerosol-radiation interactions, thereby improving atmospheric monitoring and air quality forecasting. While MERRA-2 has inherent temporal delays, this framework provides valuable capabilities for historical trend analysis, numerical model validation, and can be readily adapted for real-time applications through transfer learning with satellite observations.

Measuring and mapping vegetation structure is essential for understanding the functioning of terrestrial ecosystems and for informing environmental policies. Recent years have seen a growing demand for high-resolution data on vegetation structure, driving their prediction at fine resolutions (1–30 m) at state, continental, and global spatial extents by combining satellite data with machine learning. As these initiatives expand, it is crucial to actively discuss the quality and usability of these products. Here, we briefly summarize current efforts to map vegetation structure and show that continental-to-global canopy height models (CHMs) exhibit significant errors in canopy heights compared to national airborne laser scanning (ALS) data. We recommend that regions with abundant ALS data, such as Europe, prioritize using ALS-based canopy height metrics rather than relying on less accurate predictions from satellite products. Despite variations in ALS data characteristics, such as temporal inconsistencies and differences in acquisition characteristics and classification accuracy, the generation of spatially contiguous canopy height products in raster format at fine spatial resolution is necessary and feasible. This requires coordinating efforts for data and survey harmonization, developing standardized processing pipelines and continent-wide ALS products, and ensuring free access for research and environmental policy. We show that ALS data now cover most of Europe, with newer surveys achieving higher point densities, improving their suitability for vegetation mapping. Beyond numerous applications in forestry, ecology, and conservation, such data sets are crucial for calibrating future Earth Observation missions, making them essential for producing reliable and accurate global, fine-resolution vegetation structure data.

The fifth generation of ECMWF atmospheric reanalyses data (ERA5), operational observed data, S-band radar data and Large Eddy Simulation (LES) tool were employed to investigate the evolution of local small-scale convection in Ziyang City of Sichuan Province on 11 April 2022 in spring. The meso-β-scale vortex and 700-hPa wind shear in the Sichuan Basin horizontally merged and formed the Southwest China vortex. The local convection induced heavy rainfall, hail and thunderstorm in the northeast of the organizing meso-β-scale vortex. The well-developed meso-β-scale vortex became the center of Southwest China vortex. The horizontal scale and time scale of the meso-β-vortex were 20–50 km and 2–3 hr, respectively. The newly formed convections over the northeast of the meso-β-scale vortex developed in the middle troposphere and quickly separated from the main body of the meso-β-scale vortex with the guided airflow. The convective organization at the northeastern forepart of the meso-β-scale vortex was reproduced in LES results. The strong convection and meso-β-scale vortex were basically activated in the same period. The significant isobaric surfaces for strong convection formation were from 750 to 500 hPa. And cold air in the middle and lower troposphere rarely affected the convective process. Local convections were mainly triggered by strong vertically meridional wind shear.

The limits of detection for earthquake surface deformation in the spatial domain have improved with advances in remote sensing imagery data availability, resolution, and analysis. Sub-pixel correlation and digital elevation model (DEM) differencing from sub-meter, earthquake-spanning satellite optical imagery has enhanced surface rupture mapping and deformation measurements. However, knowledge of measurement accuracy and uncertainty is limited. To address this, we construct orthophotos and digital elevation models (DEMs) from repeat high resolution (∼0.5 m) satellite optical imagery along two sections of the Garlock fault, California with clear fault geomorphology and differing land cover. We deform later sets of DEMs and images with synthetic earthquakes containing both diffuse and discrete horizontal and vertical displacements. Sub-pixel image correlation and DEM differencing demonstrate how vegetation degrades recovered displacement accuracy. In barren land cover, horizontal displacements are detectable to an expected ∼1/10th-pixel size. With shrubs, trees, and grass, detectable displacements increase to >1/2-pixel size, and filtering results by correlation score and using elevation values as input rather than image values improves accuracy. Vertical displacement detection thresholds remain lower in vegetation, at >1-pixel size. Higher slope angles degrade displacement recovery, worsened by vegetation. Diminishing seasonal separation improves accuracy over vegetated regions, though not to the level achieved in barren environments. These results will inform research and operational efforts on the utility of high resolution satellite optical imagery for detecting deformation in varied land cover. Furthermore, they reveal where alternative measurements, such as from LiDAR or radar interferometry, are required to mitigate the effects of vegetation and capture fine-scale crustal deformation.

Accurate monitoring of the Earth's gravity field is crucial for understanding mass redistribution processes related to climate change, hydrology, and geodynamics. The Gravity Recovery and Climate Experiment (GRACE) and its successor, GRACE Follow-On (GRACE-FO), have provided invaluable satellite gravimetry data through low-low satellite-to-satellite tracking (LL-SST). However, the precision of gravity field recovery is significantly affected not only by data gaps in the accelerometer (ACC) measurements, but also by potential failures or limitations in their performance. To mitigate these issues, accelerometer data transplantation has been employed, leveraging the similarity in non-gravitational accelerations experienced by both satellites. This study presents an in-depth assessment of transplant noise and evaluates advanced accelerometer configurations, including Cold Atom Interferometry (CAI) accelerometers and hybrid electrostatic-quantum accelerometer setups for future satellite gravimetry missions. Through closed-loop LL-SST simulations, we compare four different accelerometer configurations, ranging from conventional electrostatic accelerometers (EAs) to fully hybrid CAI-EA setups. Results indicate that a dual hybrid accelerometer configuration offers the highest accuracy in gravity field recovery, while a transplant-based hybrid approach significantly enhances the performance of non-gravitational force modeling without requiring additional instrumentation. The findings underscore the potential of quantum accelerometery and transplant methodologies for future satellite gravimetry missions, offering a cost-effective solution to improve gravity field recovery, while benefitting from new sensor types.

Previous papers have outlined nowcasting methods to track the current state of earthquake hazard using only observed seismic catalogs. The basis for one of these methods, the “counting method,” is the Gutenberg-Richter (GR) magnitude-frequency relation. The GR relation states that for every large earthquake of magnitude greater than M_T, there are on average N_GR small earthquakes of magnitude M_S. In this paper we use this basic relation, combined with the Receiver Operating Characteristic (ROC) formalism from machine learning, to compute the probability of a large earthquake. The probability is conditioned on the number of small earthquakes n(t) that have occurred since the last large earthquake. We work in natural time, which is defined as the count of small earthquakes between large earthquakes. We do not need to assume a probability model, which is a major advantage. Instead, the probability is computed as the Positive Predictive Value (PPV) associated with the ROC curve. We find that the PPV following the last large earthquake initially decreases as more small earthquakes occur, indicating the property of temporal clustering of large earthquakes as is observed. As the number of small earthquakes continues to accumulate, the PPV subsequently begins to increase. Eventually a point is reached beyond which the rate of increase becomes much larger and more dramatic. Here we describe and illustrate the method by applying it to a local region around Los Angeles, California, following the 17 January 1994 magnitude M6.7 Northridge earthquake.

The use of CCD-array spectrometers has substantially increased in recent years in many different fields. Although they have numerous advantages over conventional scanning spectrometers, they need to be thoroughly characterized to correct for various sources of error. This study focuses on the experimental characterization of the dark signal of Avantes AvaSpec-2048 CCD-array spectrometers, used to measure solar UV and VIS-NIR radiation. In order to have a large number of measurements of the dark signal at different integration times and temperatures, a ramp methodology has been followed and validated against stabilized-temperature experiments. These data have allowed the analysis of the individual dependencies of the dark signal with integration time and temperature, as well as the proposal of a final multivariate model including both variables. This is one of the first multivariate models proposed for the dark signal of a detector used by a CCD-array spectrometer to measure solar UV radiation. The dependence of the dark signal with integration time and temperature is found to be linear and nonlinear, respectively. The model performs remarkably well, with R² values above 0.99 and relative root mean squared error around 0.1 and 0.05 for the UV and VIS-NIR spectrometers, respectively. The improvement achieved by using an individual model for each pixel is discussed, obtaining notably better results with this model than when using an average model, as suggested by other authors. This study presents a positive contribution to the characterization of the dark signal from CCD-array spectrometers, and the proposed methodology can be extended to other instruments.

The Gravity Recovery and Climate Experiment (GRACE) and its successor, GRACE-Follow On, play an important role in monitoring mass transport across the Earth. Compared to spherical harmonic solutions, mass concentration (mascon) solutions offer less signal leakage and a “higher” spatial resolution. How the shapes, sizes, and positions of mascon are parameterized influences the accuracy of the solutions. In this study, we derive a variable-sized mascon solution that enhances spatial resolution in polar regions by considering orbital coverage of satellites. To this end, we present a numerical simulation aimed at evaluating the performance of different parameterizations in the mascon solutions. We demonstrate that using variable-sized mascons reduce parameterization error by up to 17% and improve goodness of fit by up to 34%. The accuracy of signal recovery improves by about 23%, 34%, and 42% for basin scales, respectively, in low-latitude, mid-latitude, and high-latitude zones. When applied to the GRACE (-FO) data, we see the optimized parameterization scheme reduces noise by up to 1.84 cm in the surface mass change time series. Additionally, the optimally parameterized mascon solution help to enhance signal recovery in mid-to-high latitude regions. We discuss and quantify benefits of variable-sized mason parameterizations for surface mass change recovery and suggest the optimal scheme based on the simulation and real data processing. Overall, the optimized parameterization scheme will benefit finer-scale mass change signal recovery for mascon solution.

The Mohorovičić discontinuity (Moho) marks the boundary between Earth's crust and the underlying mantle, serving as a critical interface for understanding Earth's structure, composition, and geodynamic processes. This study introduces a novel iterative and stable algorithm for global Moho depth inversion. We first derive the gravity disturbance of the Moho interface in the spherical harmonic domain, expressed as a series of spherical harmonic coefficients. These forward expressions are then reformulated into an iterative scheme for Moho depth estimation. To ensure convergence, a damping factor is applied to suppress high-frequency noise, and the process is constrained by observed gravity data to minimize residuals. The algorithm is validated using a synthetic Airy–Heiskanen interface in a closed-loop test. Results show stable convergence within approximately three iterations, yielding minimal gravity residuals (∼0.05 mGal) and small depth errors (standard deviation: 0.07 km), demonstrating the method's high accuracy. A sensitivity analysis of constant and variable Moho density contrasts further shows that when density varies from 450 to 600, the mean difference is less than 1.0 km and the standard deviation is only 1.1 km, indicating that the solution is largely insensitive to density changes. Importantly, incorporating a variable density contrast significantly improves Moho depth recovery along mid-ocean ridges. Finally, the method is applied to refined gravity disturbances that are maximally correlated with Moho depth, successfully recovering global Moho topography. Comparison with the CRUST1.0 seismic Moho model shows strong consistency in both spatial distribution and statistical measures, with depth residuals (standard deviation: 4.23 km) and gravity residuals (∼1.89 mGal), further confirming the robustness of the method. Notably, the use of variable Moho density contrast again provides substantial improvements along mid-ocean ridges.