ISPRS Journal of Photogrammetry and Remote Sensing最新文献

Information on planting dates is crucial for modeling crop development, analyzing crop yield, and evaluating the effectiveness of policy-driven planting windows. Despite their high importance, field-level planting date datasets are scarce. Satellite remote sensing provides accurate and cost-effective solutions for detecting crop phenology from moderate to high resolutions, but remote sensing-based crop planting date detection is rare. Here, we aimed to generate field-level crop planting date maps by taking advantage of satellite remote sensing-derived phenological metrics and proposed a two-step framework to predict crop planting dates from these metrics using required growing degree dates (RGDD) as a bridge. Specifically, we modeled RGDD from the planting date to the spring inflection date (derived from phenological metrics) and then predicted the crop planting dates based on phenological metrics, RGDD, and environmental variables. The ∼3-day and 30-m Harmonized Landsat and Sentinel-2 (HLS) products were used to derive crop phenological metrics for corn and soybean fields in the U.S. Midwest from 2016 to 2021, and the ground truth of field-level planting dates from USDA Risk Management Agency (RMA) reports were used for the development and validation of our proposed two-step framework. The results indicated that our framework could accurately predict field-level planting dates from HLS-derived phenological metrics, capturing 77 % field-level variations for corn (mean absolute error, MAE=4.6 days) and 71 % for soybean (MAE=5.4 days). We also evaluated the predicted planting dates with USDA National Agricultural Statistics Service (NASS) state-level crop progress reports, achieving strong consistency with median planting dates for corn (R²=0.90, MAE=2.7 days) and soybeans (R²=0.87, MAE=2.5 days). The model’s performance degraded slightly when predicting planting dates for fields with irrigation (MAE=5.4 days for corn, MAE=6.1 days for soybean) and cover cropping (MAE=5.4 days for corn, MAE=5.6 days for soybean). The USDA RMA Common Crop Insurance Policy (CCIP) provides county- or sub-county-level crop planting windows, which drive producers’ decisions on when to plant. Within the CCIP-driven planting windows, higher prediction accuracies were achieved (MAE for corn: 4.5 days, soybean: 5.2 days). Our proposed two-step framework (phenological metrics-RGDD-planting dates) also outperformed the traditional one-step model (phenological metrics-planting dates). The proposed framework can be beneficial for deriving planting dates from current and future phenological products and contribute to studies related to planting dates such as the analysis of yield gaps, management practices, and government policies.

China leads in the greening of the world, with a nearly doubled increase in its forest area since the 1980 s revealed by the National Forest Inventory (NFI). However, a significant challenge persists in the absence of consistent and reliable remote sensing data that align with the NFI, hindering a comprehensive understanding of the spatiotemporal patterns of terrestrial ecosystem changes driven by afforestation and reforestation efforts over recent decades in China. Moreover, conventional binary thematic maps and land use and land cover (LULC) maps encounter difficulties in providing a thorough assessment of canopy cover at the subpixel level and trees extending beyond officially designated forest boundaries. This limitation creates substantial gaps in our comprehension of their invaluable contributions to ecosystem services. To confront these challenges, this study presents a systematic framework integrating time-series Landsat satellite imagery and random forest-based ensemble learning techniques. This framework aims to generate China’s inaugural annual tree cover dataset (CATCD) spanning from 1985 to 2023 at a 30 m spatial resolution. Evaluation against multisource reference data shown high correlations ranging from 0.70 to 0.96 and reasonable RMSE values ranging from 5.6 % to 25.2 %, highlighting the reliability and precision of our approach across different years and data collection methodologies. Our analysis reveals that China’s forested area has doubled, expanding from 1.04 million km2 in 1985 to 2.10 million km2 in 2023. Notably, 33 % of this growth can be attributed to a shift from non-forest to forest land categories, primarily observed in the three-north and southwest regions. However, the majority, contributing 67 %, results primarily from crown closure in central and southern China. This realization underscores the limitations of conventional binary thematic maps and LULC maps in accurately quantifying forest gain in China. Furthermore, China’s tree population structure has undergone a transformative shift from 83 % forest trees and 17 % non-forest trees in 1985 to 92 % forest trees and 8 % non-forest trees in 2023, signifying a transition from afforestation to established forests. Our study not only enhances the understanding of tree cover variations in China but also provides valuable data for ecological investigations, land management strategies, and assessments related to climate change.

Investigating the relationship between cooling efficiency (CE) and tree cover percentage (TCP) is critical for planning of green space within cities. However, the spatiotemporal complexities of the intra-city CE-TCP relationship worldwide with distinct climates, as well as the differing impacts of consistently increasing tree cover within urban regions on cooling potential, remain unclear. Here we used satellite-derived MODIS observations to investigate the CE-TCP relationship across 440 global cities during summertime from 2018 to 2020. We further investigated the impacts of enhancing tree cover by a consistent amount in different urban locales on the reduction of population heat exposure among specific age groups. Our results demonstrate a nonlinear CE-TCP relationship globally – CE exhibits an initial sharp decline followed by a gradual reduction as TCP rises, and this nonlinearity is more pronounced in tropical and arid climates than in other climate zones. We observe that 91.4% of cities experience a greater reduction in population heat exposure when introducing the same amount of TCP in areas with fewer trees than in those with denser canopies; and heat exposure mitigation is more prominent for laborers than for vulnerable groups. These insights are critical for developing strategies to minimize urban heat-related health risks.

Forest stress monitoring and in-time identification of forest disturbances are important to improve forest resilience to climate change. Fast-developing drone techniques and hyperspectral imagery provide tools for understanding the forest decline process under stress and contribute to focused monitoring. This study explored and developed hyperspectral drone imagery for early detection of forest stress caused by European spruce bark beetle Ips typographus (L.), before offspring emergence, which is crucial in controlling the spread but has been shown to be challenging.

This study challenges the highest possible detectability of infested trees using a hyperspectral drone system that provided images with very high spectral, spatial, and temporal resolutions in Southern Finland. Images were acquired bi-weekly, four times (T1, T2, T3, T4), covering 8 weeks from trees being attacked by the first filial generation (F1) to the beginning of second filial generation (F2) brood emergence. Very low separability was observed for the reflectance from healthy and attacked trees, but the first and second derivative reflectance captured vitality changes, with the green shoulder region (wavelengths 490–550 nm) exhibiting the highest separability of all wavelengths (400–1700 nm). We discovered that the peak and valley values of the first and second derivative curves in the green shoulder region consistently shifted with longer infestation time.

Based on this finding, we developed green shoulder indices. The detection rates were 0.24–0.31 and 0.76–0.83 for T3 and T4, higher than commonly used VIs such as the Photochemical Reflectance Index and the Red Edge Inflection Position, with detection rates of 0.69 and 0.34 for T4, respectively. We also proposed simplified green shoulder indices using the reflectance from three bands that can be used with multispectral cameras and satellite images for large area monitoring of forest health. We concluded that the detectability of infestations was very low for the first month after attack, and then rapidly increased before brood emergence. We highlighted the great potential of green shoulder indices in quantifying the photochemical functioning of the vegetation under stress. The methodology can potentially be applied for early identification of forests with declining vitality caused by various sources of forest stress and disturbances, such as infestations, diseases and drought.

LiDAR-based semantic scene understanding holds a pivotal role in various applications, including remote sensing and autonomous driving. However, the majority of LiDAR segmentation models rely on extensive and densely annotated training datasets, which is extremely laborious to annotate and hinder the widespread adoption of LiDAR systems. Semi-supervised learning (SSL) offers a promising solution by leveraging only a small amount of labeled data and a larger set of unlabeled data, aiming to train robust models with desired accuracy comparable to fully supervised learning. A typical pipeline of SSL involves the initial use of labeled data to train segmentation models, followed by the utilization of predictions generated from unlabeled data, which are used as pseudo-ground truths for model retraining. However, the scarcity of labeled data limits the capture of comprehensive representations, leading to the constraints of these pseudo-ground truths in reliability. We observed that objects captured by LiDAR sensors from varying perspectives showcase diverse data characteristics due to occlusions and distance variation, and LiDAR segmentation models trained with limited labels prove susceptible to these viewpoint disparities, resulting in inaccurately predicted pseudo-ground truths across viewpoints and the accumulation of retraining errors. To address this problem, we introduce the Temporal-Selective Guided Learning (TSG-Seg) framework. TSG-Seg explores temporal cues inherent in LiDAR frames to bridge the cross-viewpoint representations, fostering consistent and robust segmentation predictions across differing viewpoints. Specifically, we first establish point-wise correspondences across LiDAR frames with different time stamps through point registration. Subsequently, reliable point predictions are selected and propagated to points from adjacent views to the current view, serving as strong and refined supervision signals for subsequent model re-training to achieve better segmentation. We conducted extensive experiments on various SSL labeling setups across multiple public datasets, including SemanticKITTI and SemanticPOSS, to evaluate the effectiveness of TSG-Seg. Our results demonstrate its competitive performance and robustness in diverse scenarios, from data-limited to data-abundant settings. Notably, TSG-Seg achieves a mIoU of 48.6% using only 5% of and 62.3% with 40% of labeled data in the sequential split on SemanticKITTI. This consistently outperforms state-of-the-art segmentation methods, including GPC and LaserMix. These findings underscore TSG-Seg’s superior capability and potential for real-world applications. The project can be found at https://tsgseg.github.io.

The segmentation of individual trees based on deep learning is more accurate than conventional meth- ods. However, a sufficient amount of training data is mandatory to leverage the accuracy potential of deep learning-based approaches. Semi-supervised learning techniques, by contrast, can help simplify the time-consuming labelling process. In this study, we introduce a new semi-supervised tree segmen- tation approach for the precise delineation and classification of individual trees that takes advantage of pre-clustered tree training labels. Specifically, the instance segmentation Mask R-CNN is combined with the normalized cut clustering method, which is applied to lidar point clouds. The study areas were located in the Bavarian Forest National Park, southeast Germany, where the tree composition includes coniferous, deciduous and mixed forest. Important tree species are European beech (Fagus sylvatica), Norway spruce (Picea abies) and silver fir (Abies alba). Multispectral image data with a ground sample distance of 10 cm and laser scanning data with a point density of approximately 55 points/m² were acquired in June 2017. From the laser scanning data, three-channel images with a resolution of 10 cm were generated. The models were tested in seven reference plots in the national park, with a total of 516 trees measured on the ground. When the color infrared images were used, the experiments demonstrated that the Mask R-CNN models, trained with the tree labels generated through lidar-based clustering, yielded mean F1 scores of 79 % that were up to 18 % higher than those of the normalized cut baseline method and thus significantly improved. Similarly, the mean over- all accuracy of the classification results for the coniferous, deciduous, and standing deadwood tree groups was 96 % and enhanced by up to 6 % compared with the baseline classification approach. The experiments with lidar-based images yielded slightly worse (1–2 %) results both for segmentation and for classification. Our study demonstrates the utility of this simplified training data preparation pro- cedure, which leads to models trained with significantly larger amounts of data than is feasible with with manual labelling. The accuracy improvement of up to 18 % in terms of the F1 score is further evidence of its advantages.

Plant species diversity is key to ecosystem functioning, but in recent decades anthropogenic activities have prompted an alarming decline in this community trait. Thus, developing strategies to understand diversity dynamics based on affordable and efficient remote sensing monitoring is essential, as well as examining the relevance of spatial scale and vegetation structural complexity to these dynamics. Here, we used two mathematical approaches to assess the relationship between tropical woody species diversity and spectral diversity in a human-modified landscape in two vegetation types differing in their degree of complexity. Vegetation complexity was measured through the fraction of species that concentrate different proportions of the cumulative importance value index. Species diversity was assessed using Hill numbers at three spatial scales, and metrics of spectral heterogeneity, vegetation indices, as well as raw data from Landsat 9 and Sentinel-2 sensors were calculated and analysed through general linear models (GLM) and Random Forest. Vegetation complexity emerged as an important variable in modelling species from remote sensing metrics, indicating the need to model species diversity by vegetation type rather than region. Hill numbers showed different relationships with remotely sensed metrics, in consistency with the scale-dependency of ecological processes on species diversity. Contrary to multiple previous reports, in our study, GLMs produced the best fits between Hill numbers of all orders and remotely sensed metrics. If we are to meet the need of conducting efficient and speedy woody species diversity monitoring globally, we propose modelling this diversity from remotely-sensed variables as an attractive strategy, so long as the intrinsic properties of each vegetation type are acknowledged to avoid under- or overestimation biases.

Remote sensing imagery has an ever-increasing impact on important downstream applications, such as vegetation monitoring and climate change modelling. Clouds obscuring parts of the images create a substantial bottleneck in most machine learning tasks that use remote sensing data, and being robust to this issue is an important technical challenge. In many cases, cloudy images cannot be used in a machine learning pipeline, leading to either the removal of the images altogether, or to using suboptimal solutions reliant on recent cloud-free imagery or the availability of pre-trained models for the exact use case. In this work, we propose VPint2, a cloud removal method built upon the VPint algorithm, an easy-to-apply data-driven spatial interpolation method requiring no prior training, to address the problem of cloud removal. This method leverages previously sensed cloud-free images to represent the spatial structure of a region, which is then used to propagate up-to-date information from non-cloudy pixels to cloudy ones. We also created a benchmark dataset called SEN2-MSI-T, composed of 20 scenes with 5 full-sized images each, belonging to five common land cover classes. We used this dataset to evaluate our method against three alternatives: mosaicking, an AutoML-based regression method, and the nearest similar pixel interpolator. Additionally, we compared against two previously published neural network-based methods on SEN2-MSI-T, and evaluate our method on a subset of the popular SEN12MS-CR-TS benchmark dataset. The methods are compared using several performance metrics, including the structural similarity index, mean absolute error, and error rates on a downstream NDVI derivation task. Our experimental results show that VPint2 performed significantly better than competing methods over 20 experimental conditions, improving performance by 2.4% to 34.3% depending on the condition. We also found that the performance of VPint2 only decreases marginally as the temporal distance of its reference image increases, and that, unlike typical interpolation methods, the performance of VPint2 remains strong for larger percentages of cloud cover. Our findings furthermore support a cloud removal evaluation approach founded on the transfer of cloud masks over the use of cloud-free previous acquisitions as ground truth.

Floods are among the most devastating natural disasters, posing significant risks to life, property, and infrastructure globally. Earth observation satellites provide data for continuous and extensive flood monitoring, yet limitations exist in the spatial completeness of monitoring using optical images due to cloud cover. Recent studies have developed gap-filling methods for reconstructing cloud-covered areas in water maps. However, these methods are not tailored for and validated in cloudy and rainy flooding scenarios with rapid water extent changes and limited clear-sky observations, leaving room for further improvements. This study investigated and developed a novel reconstruction method for time series flood extent mapping, supporting spatially seamless monitoring of flood extents. The proposed method first identified surface water from time series images using a fine-tuned large foundation model. Then, the cloud-covered areas in the water maps were reconstructed, adhering to the introduced submaximal stability assumption, on the basis of the prior water occurrence data in the Global Surface Water dataset. The reconstructed time series water maps were refined through spatiotemporal Markov random field modeling for the final delineation of flooding areas. The effectiveness of the proposed method was evaluated with Harmonized Landsat and Sentinel-2 datasets under varying cloud cover conditions, enabling seamless flood mapping at 2–3-day frequency and 30 m resolution. Experiments at four global sites confirmed the superiority of the proposed method. It achieved higher reconstruction accuracy with average F1-scores of 0.931 during floods and 0.903 before/after floods, outperforming the typical gap-filling method with average F1-scores of 0.871 and 0.772, respectively. Additionally, the maximum flood extent maps and flood duration maps, which were composed on the basis of the reconstructed water maps, were more accurate than those using the original cloud-contaminated water maps. The benefits of synthetic aperture radar images (e.g., Sentinel-1) for enhancing flood mapping under cloud cover conditions were also discussed. The method proposed in this paper provided an effective way for flood monitoring in cloudy and rainy scenarios, supporting emergency response and disaster management. The code and datasets used in this study have been made available online (https://github.com/dr-lizhiwei/SeamlessFloodMapper).

Coastal tidal wetlands are crucial for environmental and economic health, but facing threats from various environmental changes. Detecting changes of tidal wetlands is essential for promoting sustainable development in coastal areas. Despite extensive researches on tidal wetland changes, persistent challenges still exist. Firstly, the high similarity among tidal wetland types hinders the effectiveness of existing common indices. Secondly, many current methods, relying on hand-crafted features, are time-consuming and subject to personal biases. Thirdly, few studies effectively integrate multi-temporal and semantic information, leading to misinterpretations from environmental noise and tidal variations. In view of the abovementioned issues, we proposed a novel temporal-spectral-semantic-aware convolutional transformer network (TSSA-CTNet) for multi-class tidal wetland change detection. Firstly, to address spectral similarity among different tidal wetlands, we proposed a sparse second order feature construction (SSFC) module to construct more separable spectral representations. Secondly, to get more separable features automatically, we constructed temporal-spatial feature extractor (TSFE) and siamese semantic sharing (SiamSS) blocks to extract temporal-spatial-semantic features. Thirdly, to fully utilize semantic information, we proposed a center comparative label smoothing (CCLS) module to generate semantic-aware labels. Experiments in the Greater Bay Area, using Landsat data from 2000 to 2019, demonstrated that TSSA-CTNet achieved 89.20% overall accuracy, outperforming other methods by 3.75%–16.39%. The study revealed significant area losses in tidal flats, mangroves, and tidal marshes, decreased by 3148 hectares, 35 hectares, and 240 hectares, respectively. Among the cities in GBA, Zhuhai shows the most significant area loss with a total of 1626 hectares. TSSA-CTNet proves effective for multi-class tidal wetland change detection, offering valuable insights for tidal wetland protection.