Zhongbing Chang

Beach litter poses persistent ecological and socioeconomic risks to coastal zones, yet reliably detecting extremely small objects (ESOs) in unmanned aerial vehicle (UAV) imagery remains challenging: these targets occupy only a very small image area on the order of 10^-5 of a typical RGB frame and are easily confused with textured sand backgrounds. We propose SRGO-YOLO, a super-resolution and granularity-optimized detector for UAV-based beach-litter monitoring. The network integrates a Scale-Adaptive Spatial Recalibration with Channel Attention module, a lightweight C3K2-MGLite neck, and a Spatial-Optimization-Aware Head with an additional P2 tiny-object branch to enhance shallow high-resolution features, perform mixed-granularity multi-scale fusion, and refine predictions for low-contrast ESOs. To evaluate SRGO-YOLO, we construct the Beach-Litter-UAV dataset by screening two public UAV garbage datasets and combining them with a newly collected Zhuhai Coastal Beach Litter Dataset, in which more than 80% of annotated instances have normalized sizes below 0.02. Experiments show that SRGO-YOLO achieves 81.4% mAP₅₀ and 48.2% mAP_50:95, surpassing YOLOv11s by 21.8 and 13.5 percentage points, respectively, while maintaining 127 FPS on an RTX 3090 GPU. On a Jetson AGX Orin 64GB platform with TensorRT FP16 deployment, SRGO-YOLO further sustains 31.21 FPS in a 10-min continuous inference test with an average end-to-end latency of 32.04 ms. Per-class results further show consistent gains across all five categories: compared with YOLOv11s, SRGO-YOLO improves mAP₅₀ by 28.8%, 31.2%, 5.4%, 29.5%, and 14.5% for plastic, flexible pouch, metal, glass, and food container, respectively, with particularly strong benefits for visually ambiguous small items such as flexible pouch and food container. Qualitative comparisons further indicate that SRGO-YOLO recovers more complete instances under partial occlusion, tight adjacency, and overlapping litter patches. These results indicate that SRGO-YOLO not only improves ESO detection accuracy but also shows strong potential for real-time UAV-based beach-litter detection, supporting timely shoreline litter mapping and targeted coastal cleanup planning.

Research on monitoring forest disturbances and analyzing its driving factors is crucial for the sustainable management of forest ecosystems. To quantitatively identify the spatial distribution and dynamic changes of forest disturbance and its driving factors in Guangdong Province from 1990 to 2019, the long-term Landsat time series imagery and the LandTrendr change detection algorithm were utilized. The impact of forest disturbances on four types of landscape fragmentation (attrition, perforation, shrinkage, and subdivision) was analyzed using the Forman index. The Geodetector model was used to analyze the driving factors of forest disturbance from human activity and the natural environment. The results showed that the LandTrendr algorithm achieved a Kappa coefficient of 0.79, with an overall accuracy of approximately 82.59%. The findings indicate a consistent increase in shrinkage patches, both in quantity and area. Spatially, the centroids of forest fragmentation processes exhibited a clear inland migration trend, reflecting the growing ecological pressures faced by inland forest ecosystems. Furthermore, interactions among driving factors, particularly between population density and economic factors, significantly amplified their combined impacts. The correlation between forest disturbances and socio-economic factors revealed distinct regional variations, highlighting significant differences in forest disturbance dynamics across cities with varying levels of economic development. This study provides critical insights into the spatiotemporal dynamics of forest disturbances under rapid urbanization and economic development. It lays the groundwork for sustainable forest management strategies in Guangdong Province and may contribute to global discussions on managing forest ecosystems during periods of rapid socio-economic transformation.

Forests cover around one-third of the global land surface, store about half of the terrestrial carbon, and are the dominant contributors to terrestrial net primary production

Clarifying the relationship between human activities and the provision of ecosystem services has received significant interest in recent years because of a growing need for sustainable socio-ecological system development. Using multi-source remote sensing data, we assessed the spatial and temporal distribution of the human footprint index and five ecosystem services under four human activity gradients from 2010 to 2020 in the Xiangjiang River Basin. The five ecosystem services include water supply, soil conservation, food production, habitat quality, and carbon sequestration. The relationship between human footprint and ecosystem services was analyzed from quantitative and spatial perspectives. The results showed that over the past 10 years, water supply and habitat quality decreased by 4.59% and 16.49%, respectively. The other three services increased, and the upstream area of the basin had a higher level of ecosystem services provision. The human footprint index increased by 28.83% over the 10 years and was characterized by point and patchy clustering in the middle and lower reaches. In terms of quantitative characteristics, the relationship between human footprint and ecosystem services was primarily negative. The ecosystem services were sensitive to the human footprint index within the 0−0.4 range. In terms of spatial characteristics, the relationship was dominated by trade-offs. The risky “high–low” trade-offs were mainly distributed in the middle and lower reaches. As the gradients of human activity increased, the maximum fluctuation in ESs was 43%, and the maximum fluctuation in human footprint was 28%, making their relationship more complex. Our results identified response thresholds of ecosystem services to human activities, providing a guide for ecological management and sustainable development of basins.

Exploring the relationship between topography and forest multifunctionality enhances understanding of the mechanisms maintaining forest multifunctionality and proves beneficial for managing overall forest functions across different landscapes. Leveraging census data from a 20 ha subtropical forest plot, we investigated the topographic variations in individual functions, multifunctionality, and their interrelationships. Our results revealed that relative to lower elevations, higher elevations had higher woody productivity, sapling growth, and recruitment that drove higher average forest multifunctionality (FMA). However, forest multifunctionality at the 50% threshold level (FMt50) had no significant difference between high and low elevations. Compared with the valley and slope, higher woody productivity, higher sapling recruitment, and higher soil organic carbon stock drove higher forest multifunctionality (FMA and FMt50) in the ridge. These results indicate the ridge serves as a forest multifunctionality “hotspot” within the Tiantong 20 hm² plot. Additionally, relative to the low elevation, the degree of synergy among functions at the high elevation was significantly lower, indicating difficulties in attaining high forest multifunctionality at the high elevation. Our work underscores the importance of topography in regulating subtropical forest multifunctionality and relationships between forest functions at a local scale, suggesting that future forest management strategies (such as regulating synergistic or trade-off relationships between functions) should give particular attention to topographic conditions.

1. Aboveground biomass (AGB) serves as a crucial proxy for productivity, carbon sequestration and carbon balance capacity in terrestrial ecosystems (Houghton 2005; Scholze et al. 2017). Accurate es...

Multiple methods have been developed to identify the transition threshold from the reconstructed satellite-derived normalized difference vegetation indices (NDVI) time series and to determine the inflection point corresponding to a certain phenology phase (e.g., the spring green-up date (GUD)). We address an issue that large uncertainties might occur in the inflection point identification of spring GUD using the traditional satellite-based methods since different vegetation types exhibit asynchronous phenological phases over a heterogeneous ecoregion. We tentatively developed a Maximum-derivative-based (MDB) method and provided inter-comparisons with two traditional methods to detect the turning points by the reconstructed time-series data of NDVI for identifying the GUD against long-term observations from the sites covered by a mixture of deciduous forest and herbages in the Pan European Phenology network. Results showed that higher annual mean temperature would advance the spring GUD, but the sensitive magnitudes differed depending on the vegetation type. Therefore, the asynchronization of phenological phases among different vegetation types would be more pronounced in the context of global warming. We found that the MDB method outperforms two other traditional methods (the 0.5-threshold-based method and the maximum-ratio-based method) in predicting the GUD of the subsequent-green-up vegetation type when compared with ground observation, especially at sites with observed GUD of herbages earlier than deciduous forest, while the Maximum-ratio-based method showed better performance for identifying GUDs of the foremost-green-up vegetation type. Although the new method improved in our study is not universally applicable on a global scale, our results, however, highlight the limitation of current inflection point identify algorithms in predicting the GUD derived from satellite-based vegetation indices datasets in an ecoregion with heterogeneous vegetation types and asynchronous phenological phases, which makes it helpful for us to better predict plant phenology on an ecoregion-scale under future ongoing climate warming.

Mapping the spatial variation of forest aboveground biomass (AGB) at the national or regional scale is important for estimating carbon emissions and removals and contributing to global stocktake and balancing the carbon budget. Recently, several gridded forest AGB products have been produced for China by integrating remote sensing data and field measurements, yet significant discrepancies remain among these products in their estimated AGB carbon, varying from 5.04 to 9.81 Pg C. To reduce this uncertainty, here, we first compiled independent, high-quality field measurements of AGB using a systematic and consistent protocol across China from 2011 to 2015. We applied two different approaches, an optimal weighting technique (WT) and a random forest regression method (RF), to develop two observationally constrained hybrid forest AGB products in China by integrating five existing AGB products. The WT method uses a linear combination of the five existing AGB products with weightings that minimize biases with respect to the field measurements, and the RF method uses decision trees to predict a hybrid AGB map by minimizing the bias and variance with respect to the field measurements. The forest AGB stock in China was 7.73 Pg C for the WT estimates and 8.13 Pg C for the RF estimates. Evaluation with the field measurements showed that the two hybrid AGB products had a lower RMSE (29.6 and 24.3 Mg/ha) and bias (−4.6 and −3.8 Mg/ha) than all five participating AGB datasets. Our study demonstrated both the WT and RF methods can be used to harmonize existing AGB maps with field measurements to improve the spatial variability and reduce the uncertainty of carbon stocks. The new spatial AGB maps of China can be used to improve estimates of carbon emissions and removals at the national and subnational scales.