ABSTRACT Feature extraction is a core aspect in hyperspectral image classification, which can extract key information closely related to ground cover from complex scene, thus improving classification accuracy. Therefore, designing an effective feature extraction network is a hotspot and a challenge in the current research. In this paper, a feature extraction framework based on 3-D homogeneous attribute decomposition (3D-HAD) is proposed for HSI classification, which consists of the following key technologies. First, the principal component analysis algorithm is applied to the raw HSI to extract the principal components (PCs), and the raw HSI is clustered into many 3D superpixel blocks according to the first three PCs-based over-segmentation strategy. Then, a superpixel intrinsic attribute decomposition (SIAD) is designed to capture reflectance feature and suppress shading feature. Meanwhile, a metric entropy is introduced into the decomposition process to overcome the spectral-spatial weak assumption among pixels. Next, superpixel-guided recursive filtering is employed to preserve global details of HSI to enhance accuracy in HSI classification. Finally, the support vector machine classifier is used to obtain classification results of HSI. Experiments performed on several real hyperspectral data sets with limited training samples indicate that the proposed 3D-HAD method outperforms the classic, advance, and deep learning classification methods.

ABSTRACT The Taihu Lake basin is one of the fastest-growing regions in China, where the natural environment has been seriously affected by humans. The plant area index (PAI) is an important parameter reflecting the change in vegetation growth, which plays a crucial role in studying vegetation growth and protecting the ecological environment. Advancements in remote sensing technology, complemented by machine learning techniques, have facilitated the accurate and efficient acquisition of PAI over large areas. In this study, the Taihu Lake Basin vegetation area was taken as the research object. Global Ecosystem Dynamics Investigation (GEDI) point cloud data and Landsat-8 remote sensing images were the primary information sources. MODIS land cover types were utilized to classify the vegetation into six categories. Three classical machine learning models, namely, Random Forest (RF), Support Vector Regression (SVR), and Back Propagation Neural Network (BPNN), were used to estimate the PAI in the Taihu Lake Basin. It was found that the RF model showed the best performance. The determination coefficients (R2) for grassland, evergreen forest, mixed forest, deciduous forest, farmland, and wetland were 0.71, 0.67, 0.69, 0.66, 0.65, and 0.69, respectively. Over 2000-2022, the PAI exhibited an absolute change rate of 0.035, with an overall increasing trend. The area of improved and degraded vegetation accounted for 58.33% and 41.67% of the total area, respectively. The study also revealed that PAI was positively correlated with precipitation (R = 0.64, P < 0.05) and negatively correlated with temperature (R = -0.58, P < 0.05). Different land types’ effects on PAI were also analyzed, with wetland PAI having the smallest mean value and evergreen forest PAI having the most considerable mean value. This research underscores the effectiveness of integrating GEDI data and Landsat-8 imagery in PAI assessment, providing valuable insights for environmental monitoring and analysis in the Taihu Lake Basin.

ABSTRACT Vegetation height, and its spatial and temporal changes, is an important environmental parameter required for understanding natural habitats, estimating carbon stores and monitoring forestry activities. Recent satellite LiDAR altimetry sensors have discontinuous spatial coverage but can be combined with spatially complete remote sensing data to extrapolate to large regions. Earlier studies have focused on producing a single spatially continuous vegetation height product. This research builds on past studies, using Landsat (annual surface reflectance and fractional cover products) and the Global Ecosystem Dynamics Investigation (GEDI) data to generate annual vegetation height layers from 1988 to 2022. GEDI data for 2019 were used to train and validate the model, resulting in a root mean square error (RMSE) of 5.45 m, mean absolute error (MAE) of 3.82 m, and coefficient of determination (R2) of 0.63. This accuracy reduces when the modelled height for 2020, 2021, and 2022 is compared to GEDI data for the same years (RMSE = 6.08–6.29 m, MAE = 4.36–4.73 m, and R2 = 0.48–0.54). Validation with independent field measurements across Australia from 2011 to 2021 shows an RMSE, MAE, and R2 of 8.2 m, 5.2 m, and 0.48, respectively. One source of error is the saturation of the Landsat signal in tall, closed canopy vegetation. While model accuracy is correlated with plot-based vegetation height measurements, results indicate that accuracy reduces for the years outside of the model calibration year (i.e. 2019). When compared to other vegetation height products (also produced using GEDI and spatial remote sensing data) from three independent published studies (one for 2009, one for 2019, and one for 2020), the model developed here tends to estimate 2–4 m taller than the first two studies and around 5 m shorter when compared to the third study. This investigation demonstrates the potential to produce multiyear vegetation height at a continental scale but also highlights the large uncertainty in modelled estimates especially when extrapolating to years other than the model calibration year.

ABSTRACT The fall armyworm (FAW), Spodoptera frugiperda (J.E. Smith), invasion endangered the maize production worldwide, including India. The objective of this study was to quantify the FAW damage severity and its impact on leaf area index (LAI), biomass and grain yield of maize and to detect the field damage using high-resolution multispectral spaceborne remote sensing data. Maize growing fields in the Kurnool District of Andhra Pradesh and the Gadwal District of Telangana, India, were randomly surveyed to collect detailed ground-truth information. Foliar damage due to FAW was recorded, and the fields were categorized into various severity grades (healthy, low, medium and severe). FAW infestation caused significant change in LAI between the severity grades, which formed the basis for its damage detection using multispectral spaceborne remote sensing. Severe FAW infestation caused significant reduction in LAI, biomass and grain yield ranging between 36.9 and 39.9% compared to healthy grade. The infestation at the leaf collar (LC) stage caused significant yield loss of up to 26.5% compared to the tassel initiation (TI) and tasselling and silking (TS) stages. Canopy spectral reflectance from healthy and FAW-infested plants showed significant differences in the visible and near infrared (NIR) regions. A reflectance peak was observed in the NIR region of healthy plants compared to infested plants. Among various spaceborne vegetation indices, the Soil Adjusted Vegetation Index (SAVI) performed better in identifying the FAW infestation (R2 = 0.61**), biomass (R2 = 0.70**) and yield loss (R2 = 0.82**). These findings indicate the feasibility of utilizing multispectral remote sensing data for monitoring FAW infestation on a spatial scale, thus enabling the site-specific management.

ABSTRACT Estuaries and coasts runoff plays a crucial role in the dynamics of the ocean system. However, the current emphasis on synthetic aperture radar (SAR) retrieved current velocities has been on studying strong boundary currents and large-scale ocean circulation, resulting in relatively limited observational research on estuarine current velocities. This study aims to observe the current velocities in the Yangtze River estuary area using Sentinel-1A SAR data. The findings indicate a high agreement between the current velocities retrieved by Sentinel-1A and those observed at hydrographic stations, showcasing the characteristics of current velocity distribution associated with seafloor topography. This highlights the significant potential of SAR in observing estuarine current velocities, contributing to the analysis of benthic topography and the statistics of river runoff.

"Those who cannot remember the past are condemned to repeat it." This maxim underscores the importance of historical awareness in medicine, particularly for obstetricians and gynecologists (ObGyns). ObGyns significantly impact societal health through their care for pregnant women, fetuses, and newborns, uniquely positioning themto advocate for health initiatives with lasting societal benefits. Despite its importance, the history of medicine is underrepresented in medical curricula, missing opportunities to foster critical thinking and ethical decision-making. In today's climate of threatened reproductive rights, vaccine misinformation, and harmful ideologies, it is imperative for ObGyns to champion comprehensive historical education. The history of medicine, particularly in relation to societal issues- such as racism, discrimination, genocides, pandemics, and wars- provides valuable context for addressing challenges like maternal mortality, reproductive rights, vaccine hesitancy, and ethical issues. Understanding historical milestones and notable ethical breaches, such as the Tuskegee Study and the thalidomide tragedy, informs better practices and safeguards patient rights. Technological advancements in hygiene, antibiotics, vaccines, and prenatal care have revolutionized the field, yet contemporary ObGyns must remain vigilant about lessons learned from past challenges and successes. Integrating historical knowledge into medical training enhances clinical proficiency and ethical responsibility, fostering innovation and improving health outcomes. By reflecting on historical achievements and their impacts, current and future ObGyns can advance the field, ensuring comprehensive and ethically sound approaches to patient care. This paper highlights the crucial role of historical knowledge in shaping modern ObGyn practices, advocating for its integration into medical education to address contemporary health challenges and ethical considerations.

ABSTRACT As a renewable energy source, ocean wind energy plays an important role in addressing challenges such as global energy shortages and climate warming. In the past decade, the offshore wind power industry has developed rapidly. However, its development has also inevitably affected local social, economic and environmental aspects. Therefore, a timely understanding of offshore wind power dynamics development is crucial for its healthy and sustainable development. Based on this, this study designs and develops a more economical, reliable and real-time offshore wind turbine (OWT) extraction method by combining deep learning and the Google Earth Engine (GEE) cloud computing platform. The method consists of two main steps. The first part utilizes multiple semantic segmentation models to construct a multi-model detection method to initially detect OWTs. The second part utilizes the GEE cloud computing platform for installation time detection and secondary purification processing of the preliminary results. The results show that the number of global OWTs reached 13,609 by 2023, and the accuracy of the detection results reached 99.93%. China has been the fastest-growing country in offshore wind power in the last decade, from installing only 4 units in 2015 to installing 6,775 units in 2023 and surpassing the UK in 2020 and becoming the country building the most OWTs worldwide. Currently, 85% of the world’s OWTs are located in China and European North Sea waters. Additionally, other regions have great potential for offshore wind development. Finally, this study provides the world’s most up-to-date and complete OWT dataset, which can provide data support for research on marine ecological and environmental protection, marine spatial planning, and socioeconomic benefits.