Forest cover has expanded across tropical and subtropical Asia in recent decades, but area alone is an inadequate metric because it does not capture forest structure, which is critical for supporting key ecosystem functions and services, including biodiversity. Here, we apply sub-meter resolution satellite imagery to characterize the complexity of forests from a horizontal perspective by quantifying the diversity of tree crown sizes and their spatial arrangement across India, Southeast Asia, and southern China. We reveal large mismatches between reported forest area and the complexity of the forests. While overall, the majority of the forests still have a relatively high complexity, a recent change towards low complexity forests is observed. India, despite its large forest extent, shows a particularly low forest complexity, with more than half of its forests classified as low-complexity. Analysing forests affected by gain or loss in area between 2000 and 2024, we find that approximately three quarters of newly established forests are of low structural complexity, while high-complexity forests, concentrated in Myanmar, Laos, Cambodia, and Indonesia, continue to disappear. Our results highlight the need to move beyond forest cover statistics toward quality-based indicators for monitoring in the support of conservation policy.

ABSTRACT Afforestation connects isolated forests into larger contiguous forests, reducing forest fragmentation. This process decreases edge areas by transforming edge forests into new interior forests (termed transformed forests). However, the extra climate benefits from edge reductions in transformed forests, beyond those provided by the planted forests themselves, remain unclear. Here, CO 2 sequestration from increased biomass (biogeochemical effect) and emissions from decreased albedo (biophysical effect) of transformed forests in China are estimated, using multiple high‐resolution remote‐sensing datasets. The planted forest area (89.6 M ha) accounted for 35.5% of China's forest area in 2015, transforming 51.8 M ha of edge forests into interior forests. A cumulative increase of 1.4±0.2 Pg CO 2 e in the transformed forests is found, compared with a biomass increase of 10.3±0.4 Pg CO 2 e in the planted forests over ~1980–2015. These transformed forests also induce a biophysical warming effect of −0.9 Pg CO 2 e, partially offsetting the cooling effect from increased biomass. Combining both effects, transformed forests provide a net CO 2 e gain of 0.5±0.2 Pg CO 2 e, representing an extra 6.6±2.7% of the direct climate benefits from planted forests. This study reveals previously ignored extra climate benefits from reduced forest fragmentation alongside forest expansion, offering new perspectives on mitigating climate warming through afforestation.

Topographic shadows in high-resolution optical remote sensing cause coupled degradation of texture, contrast, and spectral fidelity, conventional illumination normalization or histogram transfer methods offer limited effectiveness on deep shadow regions, as the missing information cannot be recovered through brightness correction alone. This study proposes a boundary-guided generative adversarial network (GAN) for restoring vegetation shadows in high-resolution multispectral imagery over karst terrain: boundary cues explicitly control shadow transitions, and a selective fusion rule preserves nonshadow pixels by construction, training employs a realistic shadow synthesis pipeline that jointly simulates illumination attenuation, chromatic shift, contrast suppression, and texture degradation, combined with complementary losses enforcing texture fidelity, edge consistency, and brightness recovery. Experiments on Gaofen-2 multispectral data show that the proposed method achieves a peak signal-to-noise ratio (PSNR) of 35.78 dB and an structural similarity index (SSIM) of 0.96 at 200 epochs, outperforming the U-Net baseline by 13.8% and 4.3% and the ResNet baseline by 28.5% and 12.9%, respectively, notably, it reaches 20.34 dB PSNR at epoch 1 (versus 8.42 dB for U-Net and 9.81 dB for ResNet), demonstrating significantly faster convergence, while both baselines plateau or degrade after epoch 100. These results indicate that generative shadow restoration can effectively extend the usable coverage of high-resolution optical imagery in topographically complex regions.

Counting individual trees is a fundamental task for environmental monitoring, yet remains largely unexplored with satellite imagery. At these resolutions, isolated trees may still be identifiable, but crown boundaries become ambiguous in dense forests, making the notion of an individual tree inherently ill-defined. Moreover, large-scale manual annotations of individual trees are prohibitively expensive. While scalable supervision can be derived from airborne LiDAR, the resulting annotations are noisy and difficult to exploit effectively. We address these challenges by formulating tree counting as a spatial density matching problem supervised through Unbalanced Optimal Transport. This formulation naturally accommodates both precise localization of isolate trees and robust density estimation in dense forests. We further introduce a self-correction mechanism that leverages transport residuals to progressively refine noisy supervision during training. We evaluate our approach on TinyTrees, a new benchmark spanning three continents and three satellite sensors, comprising over 215 million tree annotations (including 773K manually verified instances) across 23,000 sq.km. Our method consistently outperforms detection-based, regression-based, and transport-based distribution-matching baselines, demonstrating the effectiveness of unbalanced transport and reliability-aware supervision for large-scale tree counting from satellite imagery. Code, data and models are available at https://github.com/dgominski/treematch.

Accurately quantifying canopy height and above-ground carbon across diverse land-cover types is crucial for understanding carbon storage dynamics and guiding climate-mitigation strategies. Yet existing maps often overlook non-forest ecosystems. Here we present a deep learning framework based on a U-Net architecture that combines radar, optical, elevation and slope data to produce a 10 m canopy height map across China. The model is trained with laser measurements from NASA's GEDI mission validated using unmanned aerial vehicle lidar data (MAE = 2.39 m). We then estimate the above-ground biomass and carbon from these heights using a Random Forest model (MAE = 37.71 Mg ha-1). By deriving carbon from canopy height, we take advantage of U-Net's ability to capture trees in non-forest ecosystems such as croplands, grasslands and urban areas. Our nationwide 30 m carbon map reveals that trees outside forests contribute 20.8-32.9% of China's above-ground carbon in 2019 (3.62-5.72 Pg C), underscoring their importance.

Human activities substantially reduce net ecosystem productivity (NEP) globally, yet debates remain over the contributions of land-use and land-cover change (LUCC), such as afforestation (afforestation and reforestation), versus non-LUCC ecosystem management (EM; e.g., forest tending, mountain forest restoration, and fire control). Here, we developed an analytical framework by harmonizing structurally consistent remote sensing-driven and climate-driven ecological process models to quantify the dynamic effects of LUCC and eight EM types on NEP from 2001 to 2021 in China by isolating anthropogenic effects from global change factors. We found that the NEP, which averaged 327 Tg C yr–1 across the 9.6×106 km2 country, increased at a rate of 16.1 Tg C yr–2. Forest management, including forest tending (5.53 Tg C yr–2) and mountain forest restoration (2.7 Tg C yr–2), primarily drove carbon sink increases. Although afforestation induced greater NEP growth per unit area, the total effect of forest management—due to its much greater coverage—was 4.14 times greater than that of afforestation (1.6 Tg C yr–2). Notably, the acceleration of China’s NEP after 2010 was closely associated with intensified forest management efforts. Moreover, the rate of NEP gains caused by forest tending investment (8.54 kg C yr–2 $–1) was much greater than that caused by afforestation investment (0.25 kg C yr–2 $–1). Our findings highlight the critical role of forest management in cost-effectively enhancing the carbon sink. This has important implications for global forest management strategies and achieving net-zero emissions. Forest management primarily drives the carbon sink enhancement in China from 2001 to 2021, achieving carbon gains at a rate over 4 times higher than afforestation, according to an integrated framework combining remote sensing-driven and climate-driven process models.

Abstract. Southwest China has emerged as a key global carbon stock due to widespread forest expansion and aboveground biomass (AGB) increases driven by major ecological restoration since 2000, making accurate AGB estimations vital for assessing restoration efficacy. However, existing global and national-scale AGB products exhibit substantial limitations in this region, with little correlation with National Forest Inventory (NFI) plot and UAV LiDAR data, which is likely related to the pronounced spatial heterogeneity induced by Karst landscapes and large-scale restoration efforts that exacerbate mixed-pixel effects. To address these challenges, this study proposes a Canopy Structure-driven Multi-feature Fusion Network (CSMF-Net) designed for high-precision AGB estimation in complex regions. The method takes NFI plots data as ground truth and integrates GaoFen imagery, horizontal structure derived from tree crown segmentation and vertical structure represented by canopy height data. Based on this approach, we generated a fine-grained 30 m AGB dataset (FineKarstAGB) covering four provinces in Southwest China (Yunnan, Guizhou, Guangxi, and Hunan). Accuracy assessment against independent NFI plot data demonstrated the model's robust performance (r = 0.83, RMSE = 28.51 Mg/ha), showing no evidence of saturation in high-biomass regions. Furthermore, a structural consistency assessment using an independent UAV LiDAR-derived Canopy Height Model (CHM) confirmed that FineKarstAGB maintains high ecological consistency with the true forest vertical structure (R2 = 0.54). Other public datasets show a weak correlation with both NFI (r < 0.4) and LiDAR data (R2 < 0.1). Due to the tree-level segmentation, our dataset also quantifies AGB contributions from sparse trees outside forests, thus enabling more comprehensive and spatially explicit carbon accounting. This dataset provides critical support for regional carbon cycle assessments, fine-scale evaluations of ecological restoration outcomes, and progress toward national carbon neutrality targets. The dataset is available at https://doi.org/10.57760/sciencedb.33452 (Li et al., 2026).

Forest structure is an essential variable in forest management and conservation, as it has a direct impact on ecosystem processes and functions. Previous remote sensing studies have primarily focused on the vertical structure of forests, which requires laser point data and may not always be suited to distinguish plantations from old forests. Sub-meter resolution remote sensing data and tree crown segmentation techniques hold promise in offering detailed information that can support the characterization of forest structure from a horizontal perspective, offering new insights in the tree crown structure at scale. In this study, we generated a dataset with over 5 billion tree crowns and developed a Horizontal Structure Index (HSI) by analyzing spatial relationships among neighboring trees from remote sensing optical images. We first extracted the location and crown size of overstory trees from optical satellite and aerial imagery at sub-meter resolution. We subsequently calculated the distance between tree crown centers, their angles, the crown size and crown spacing, and linked this information with individual trees. We then used principal component analysis (PCA) to condense the structural information into the HSI and tested it in China, Rwanda and Denmark. Our result showed that the HSI has the potential to distinguish monoculture plantations from other forest types, which provides insights that extend beyond metrics derived from vertical forest structure. The proposed HSI is derived directly from tree-level attributes and supports a deeper understanding of forest structure from a horizontal perspective, complementing existing remote sensing-based metrics.

Forest cover dynamics are studied on a routine basis, but how changes in forest cover impact forest fragmentation has rarely been studied over a long time period resolution. This is, however, important because forest fragmentation critically impacts ecosystem services, such as biodiversity and cooling effects. Here, we apply a long time series of Landsat images from 1986–2018 and study how forest fragmentation has changed along with forest cover dynamics in southern China. Furthermore, we attribute drivers and study the impact on local air temperature changes. The region is particularly relevant as it was largely deforested three decades ago, and most of the current forests are the result of protection and forestation measures. We found a reduction in the forest fragmentation index FFI (−34.4%) from 1986 to 2018. In 81.2% of the area, forest cover increased and fragmentation decreased, while 18.5% of the area showed increases in both forest cover and fragmentation. The contribution of human activities to forest fragmentation increased by 9%, with a distinct spatial correlation between areas of increasing forest fragmentation and high levels of human disturbance. Furthermore, we found that the average level of cooling effects in areas with increased forest cover of less than 40% is heavily dominated by forest fragmentation, whereas the cooling effects are primarily controlled by changes in forest cover. These findings underscore the role of human disturbance in driving forest fragmentation, which in turn affects the functioning of forest ecosystems. The results emphasize the need for integrated land management strategies that balance forest restoration with the mitigation of human-induced fragmentation to sustain ecosystem services in the face of ongoing environmental change.

Due to the provisioning of essential ecosystem goods and services by forests, the monitoring of forests has attracted considerable attention within the academic community. However, the majority of remote sensing studies covering large areas primarily focus on tree cover due to resolution limitations. It is necessary to integrate innovative spatial methods and tools in the monitoring of forest ecosystems. Forest Structure Complexity, representing the spatial heterogeneity within forest structures, plays a pivotal role in influencing ecosystem processes and functions. In this study, we use multi-spectral remote sensing image data to extract the crown information of the single tree through deep learning technology; Subsequently, we analyze the relationship between each single tree and its neighboring trees, and explore the structural characteristics at tree level. Finally, we developed the canopy structural complexity index and applied it to Nordic forests, urban areas, savanna, rainforest, and the most complex tree plantations and natural forests in China Karst. This study aims to gain a deeper understanding of the forest structure complexity in diverse ecosystems and provide valuable information for sustainable forestry management and ecosystem conservation. The method developed in this study eliminates the need for additional field measurement and radar data, offering robust tool support for extensive and efficient the monitoring of forest structure complexity, which has a wide application prospect.

Forest canopy height reflects the vertical structure of forests and gives indications on the growth capacity of trees and the level of forest biomass. Despite an increasing availability of global canopy height maps, there is a lack of maps reflecting temporal dynamics, which is required to understand forest carbon sink capacity. This study uses publicly available global maps and Landsat data to construct a long time series of canopy height maps in southern China (1986 to 2019). Our predictions are in the same range as National Forest Inventory data, both spatially and temporally. The dataset shows clear signs of tree growth, with small trees dominating the first period, which grow taller over the 3 decades. In numbers, the mean canopy height of the forests in southern China increased from 6.4 ± 3.99 m in 1986 to 10.3 ± 5.54 m in 2019, which is an increase of 61.25%. Secondary forests have an overall higher canopy height (17.8 ± 2.12 m) as compared to plantations (14.6 ± 2.05 m), but plantations grow faster (0.20 ± 0.08 m/year) as compared to secondary forests (0.13 ± 0.08 m/year). A driver analysis shows that management is the dominating factor for the canopy height increase, while natural factors play a minor role. Our study reaffirms the human made surge in forest areas in China, quantifies their canopy height dynamics, and demonstrates that spectral data can indeed be used to track canopy height changes at scale.

Canopy height is an important aspect of forest structure and functioning. Although water availability is important for canopy height growth, the climatic niche for tall trees remains poorly understood. Here we use global spaceborne lidar-derived canopy height to study its dependence on climate variables. We find that vapour pressure deficit (VPD) strongly controls geographical patterns of canopy height, observing a negative association also in tropical regions where water limitations are modest. Taller trees are prevalent in humid tropical regions, but canopy height decreases sharply as mean annual VPD surpasses 0.68 kPa. By 2100, projected increases in VPD under a warming climate could enhance limitations to canopy height growth, resulting in height losses in 87% of the humid tropical regions. Conversely, we project a widespread increase in canopy height across drylands, linked primarily to changing precipitation regimes. These results suggest that limitations on height growth driven by shifts in atmospheric dryness could lead to reduced future forest carbon sequestration.

Trees play a crucial role in urban environments, offering various ecosystem services that contribute to public health and human well-being. China has initiated a range of urban greening policies to increase the number of urban trees, but monitoring urban tree dynamics at a national scale has proven challenging. Here, we used high-resolution nanosatellite images to quantify urban tree cover in all major Chinese cities in 2019 and study changes in tree cover between 2010 and 2019. We show that 11.47% of urban areas were covered by trees in 2019, and 76% of the cities experienced an increase in tree cover compared with 2010. Notably, the increase in tree cover in the mega-cities of Shanghai, Beijing, Shenzhen and Guangzhou (6.64%) was higher than that in other cities analyzed. Tree cover increases also vary between urban land use types, with public service (3.09%) and residential areas (1.79%) having the highest values. The study employed a data-driven approach toward assessing urban tree cover changes, showing clear signs of overall increases that nonetheless do not benefit all cities equally.

Forests can be hotspots for ecosystem services, such as carbon stocks, biodiversity and cultural values, but economic drivers have replaced most old forests with monoculture plantations, which have very limited ecosystem services. Remnants of ancient old forests exist, in particular in rural mountain landscapes such as China Karst, but conservation typically focuses on large contiguous forest areas, often overlooking smaller patches of old forests. Here we use sub-meter resolution satellite data from recent years to locate 25 billion trees in Southern China. We find that 728 million (2%) of those trees have the potential of being part of old-growth forests. Out of these, only 15% are located in nature reserves, but the remaining ones are scattered in small clusters, possibly being remnants of old forests and should be considered as designated protection areas. Our work shows how modern satellite technology can be used for advancing biological conservation of ecologically unique forest habitats, by locating millions of forest patches in and around the karst region of China, which have the potential to be hotspots of biodiversity and species preservation.

Understanding the ecological effects of strong earthquakes and post-seismic vegetation dynamics in mountainous areas is essential for mitigating post-earthquake disasters. Here, a detailed palynological and radiocarbon-dating study was carried out on the tufa sediments from Huohuahai Lake, Jiuzhaigou National Nature Reserve, China. Rapid transitions in palynological assemblages, from high contents of shrubs and herbs to Quercus, Betula and Pinus tree pollen, have repeatedly occurred over the past similar to 1650 years. These short-term variations, combined with historical seismic records, probably record post-seismic (open forest) and inter-seismic (mixed forest) periods. Strong earthquakes caused damage to forests and induced hydrological changes, resulting in low pollen concentrations and increases in shrubs and herbs after such events. Subsequently, mixed coniferous broad-leaved forest gradually developed during inter-seismic periods. A particularly large earthquake occurred in similar to 520 CE near Jiuzhaigou, and the vegetation recovery time from this was up to similar to 214 years; this recovery time could also have been influenced by the sudden cooling and gradual drying climate and/or a previous strong earthquake. Our findings confirm the high seismic risk in the Min Shan uplift zone and the important driving force of tectonic activity on forest succession in mountainous areas.

China has experienced a rapid urbanization during recent decades, strongly affecting vegetation dynamics in areas undergoing a transformation from rural to urban areas. At the same time, national greening policies have been implemented to promote urban sustainability and urban greening in China in recent years. However, it is unclear how urban greening compensates vegetation losses from urban expansion at national scale. Here, we use Moderate Resolution Imaging Spectroradiometer and Landsat satellite normalized difference vegetation index time series to study 974 major cities (urban area > 20 km 2 ) in China during 2000 to 2020 and develop an urban vegetation change typology including 5 types of vegetation dynamics (greening, browning, stable, reversal, and recovery). We document a rapid urban expansion associated with a browning in urban areas before 2011, followed by widespread regreening of the urban areas after 2011. This recovery in greenness was found in 63.45% of the cities, while 14.68% showed a continuous browning, and 8.13% a continuous greening. Our findings reveal to what extent, where, and when vegetation browning from urban expansion is balanced by urban greening in urban core areas, which may indicate that initial vegetation losses are offset by urban greening initiatives.

Abstract Objectives Analyze quantitative changes of iris and retinal vessels in diabetic macular edema (DME) after intravitreal anti-vascular endothelial growth factor (anti-VEGF) and evaluate their correlations. Methods This was a case-cohort study. A total of 26 eyes of DME patients received anti-VEGF treatments and were reviewed three times of follow-up. Images of iris and retinal vessels were obtained before and after treatment and the area density of the vessel (VAD) and the density of the vessel skeleton (VSD) were quantitatively analyzed. Results There was no significant change in the iris VAD after the third injection (p > 0.05), but the VSD of iris decreased (p ＜0.05). Further linear regression analysis showed that the difference between postoperative and pretreatment iris VSD was negatively correlated with baseline(R = 0.793, B = -1.242, p = 0.000), but not with age, sex, and baseline visual acuity (all p > 0.05). Conclusions Iris vessels are more sensitive to anti-VEGF than retinal vessels.

Non-linear trend detection in Earth observation time series has become a standard method to characterize changes in terrestrial ecosystems. However, results are largely dependent on the quality and consistency of the input data, and only few studies have addressed the impact of data artifacts on the interpretation of detected abrupt changes. Here we study non-linear dynamics and turning points (TPs) of temperate grasslands in East Eurasia using two independent state-of-the-art satellite NDVI datasets (CGLS v3 and MODIS C6) and explore the impact of water availability on observed vegetation changes during 2001-2019. By applying the Break For Additive Season and Trend (BFAST01) method, we conducted a classification typology based on vegetation dynamics which was spatially consistent between the datasets for 40.86 % (459,669 km2) of the study area. When considering also the timing of the TPs, 27.09 % of the pixels showed consistent results between datasets, suggesting that careful interpretation was needed for most of the areas of detected vegetation dynamics when applying BFAST to a single dataset. Notably, for these areas showing identical typology we found that interrupted decreases in vegetation productivity were dominant in the transition zone between desert and steppes. Here, a strong link with changes in water availability was found for >80 % of the area, indicating that increasing drought stress had regulated vegetation productivity in recent years. This study shows the necessity of a cautious interpretation of the results when conducting advanced characterization of vegetation response to climate variability, but at the same time also the opportunities of going beyond the use of single dataset in advanced time-series approaches to better understanding dryland vegetation dynamics for improved anthropogenic interventions to combat vegetation productivity decrease.

Sacred forests are increasingly disappearing due to increasing land pressure and a decline in cultural values. Protecting the remaining sacred forests plays a crucial role in preserving biodiversity. The existence of remaining old forests often related to local people and their culture, but this relationship has rarely been quantified at large regional scales. This study analyzes the relationship between old forest and ethnic minorities based on the location of the old forest at a high spatial resolution (2 m). We found a significantly positive correlation (p < 0.01) between the proportion of ethnic minority population and old forest patch number, area, aggregation, and maximum patch area. However, there was no correlation with the connectivity of the old forest (p = 0.14). We further show that both environmental and anthropogenic factors are important for the distribution of old forests. Hydrothermal conditions contribute to the growth of forests, and local ethnic customs and the corresponding ecological wisdom contributes to the preservation of old forest remnants (r = 0.12, p < 0.05). Our findings highlight the significance of social dimensions for the conservation of old forests. We encourage that forest management should consider the role of indigenous people and their cultural wisdom for a better conservation and restoration of degraded ecosystems.

Forest structure complexity is an essential variable in forest management and conservation, as it has a direct impact on ecosystem processes and functions. Previous studies have primarily focused on tree cover as a proxy, which often falls short in providing comprehensive information on the structural complexity of forests. Sub-meter resolution remote sensing data and tree crown segmentation techniques hold promise in offering detailed information that can support the characterization of forest structure and complexity. In this study, we generated a dataset with over 5 billion tree crowns, and developed an Overstory Complexity Index (OCI) to characterize forest structure complexity from a horizontal perspective, by analyzing spatial relationships among neighboring trees from remote sensing optical images. We first extracted the location and crown size of overstory trees from optical satellite and aerial imagery at sub-meter resolution. We subsequently calculated the distance between tree crown centers, their angles, the crown size and crown spacing and linked this information with individual trees. We then used Principal Component Analysis (PCA) to condense the structural information into the OCI and tested it in China’s Guangxi province, Rwanda, and Denmark. In addition, we conducted a comparative analysis of OCI between protected and unprotected areas and among different forest types across these regions. Finally, we explored the relationships of terrain slope, distance to settlement and aboveground biomass with the OCI. Our result showed that the distribution of OCI values varies across the different bioclimatic regions, closely related to their respective forest characteristics. Higher OCI values were observed in protected areas as compared to unprotected areas, and OCI showed a positive correlation with terrain slope, distance to settlement and aboveground biomass. The proposed OCI is derived directly from standard tree-level attributes and supports a deeper understanding on forest structure and complexity in diverse ecosystems as compared to existing proxies.