Sensors onboard satellite platforms with short revisiting periods acquire frequent earth observation data. One limitation to the utility of satellite-based data is missing information in the time series of images due to cloud contamination and sensor malfunction. Most studies on gap-filling and cloud removal process individual images, and existing multi-temporal image restoration methods still have problems in dealing with images that have large areas with frequent cloud contamination. Considering these issues, we proposed a deep learning-based method named content-sequence-texture generation (CSTG) network to generate gap-filled time series of images. The method uses deep neural networks to restore remote sensing images with missing information by accounting for image contents, textures and temporal sequences. We designed a content generation network to preliminarily fill in the missing parts and a sequence-texture generation network to optimize the gap-filling outputs. We used time series of Moderate-resolution Imaging Spectroradiometer (MODIS) data in different regions, which include various surface characteristics in North America, Europe and Asia to train and test the proposed model. Compared to the reference images, the CSTG achieved structural similarity (SSIM) of 0.953 and mean absolute errors (MAE) of 0.016 on average for the restored time series of images in artificial experiments. The developed method could restore time series of images with detailed texture and generally performed better than the other comparative methods, especially with large or overlapped missing areas in time series. Our study provides an available method to gap-fill time series of remote sensing images and highlights the power of the deep learning methods in reconstructing remote sensing images.

Ensuring food security is crucial in the context of climate change and increased extreme weather events. Crop production depends not only on yield but also on the harvestable fraction (HF), the ratio of harvested areas to planted areas. While the impacts of climate fluctuations on crop yields are well‐document, the role of HF‐a critical yet underexplored factor‐remains poorly understood. This study introduces HF as a key metric for assessing how temperature, drought, and precipitation affect the production of major staple crops (winter wheat, spring wheat, and maize) in the United States at the county level. Our findings indicate that Killing Degree Days (KDD) and Drought Days (DD) are key drivers of production declines, reducing both yield and HF. From 1982 to 2020, changes in KDD and DD led to more significant reductions in crop production in the Midwest compared to other regions. Projections for 2021–2100 under different Shared Socioeconomic Pathways (SSPs) indicate even steeper declines in yield, HF and production, especially under the high‐emission scenario (SSP5‐8.5), with anticipated net production decreases of 6.08%, 8.2%, and 7.57% for the three crops. Although HF may increase in colder regions, this does not offset losses in warmer areas, leading to net HF decreases of 0.36%, 2.09%, and 5.82% for the three crops. This study finds that drought and extreme heat considerably reduce food production by simultaneously lowering yields and HF and underscores the need for adaptive strategies that address both yield and HF to enhance food security in a changing climate.

Urban forest canopy height is an import indicator for urban carbon storage, vegetation ecosystems services, and devising effective forest management strategies to combat global climate change. Although both spaceborne or airborne LiDAR could offer forest canopy height information, there presents a trade-off between the spatial resolution and coverage. Addressing this, we aim to integrate the multimodal remote sensing data and digital elevation model (DEM) data to map ultra-high spatial resolution urban forest canopy height mapping over large urban area. We proposed ARFCNet, an innovative deep learning model that synergizes convolutional, self-attention, and upsampling mechanisms for precise forest canopy height mapping. ARFCNet processes data from unmanned aerial vehicle (UAV) imagery, Sentinel-1, and DEMs. We compare the potential of forest canopy height mapping under two data binning strategies: (1) involving aerial RGB imagery, Sentinel-1 HV, and 1m resolution DEM, and (2) the same setup but with a 30m resolution DEM. Our findings indicate that ARFCNet, under the first strategy, significantly outperforms than other models, achieving the highest R² (0.98) and the lowest RMSE (1.33m). The second strategy, utilizing 30m resolution DEM, showed reduced accuracy, with an R² of 0.78 and RMSE of 4.38. Furthermore, applying ARFCNet, we mapped 1-m resolution forest canopy heights in Guangzhou, revealing a citywide mean vegetation height of 12.45m with a standard deviation of 4.26m. Comparative validation against four existing canopy height products and field measurements in Guangzhou demonstrated our model's robustness, with R² values ranging from 0.45 to 0.72 and RMSE from 3.93 to 6.04. Our ultra-high resolution (1m) forest canopy height mapping offers detailed insights, particularly in urban areas, promising significant advancements in national or global vegetation monitoring, biomass mapping accuracy, and contributions toward carbon neutrality objectives.

Accurate detection of individual trees is essential for urban forest management and ecological assessment, yet remains challenging due to the heterogeneous backgrounds, variable sizes of tree crowns, and significant variations across urban scenarios. To address these issues, we propose Tree-SAM, a city-scale individual tree detection architecture built upon the visual foundation model Segment Anything Model (SAM) and equipped with three task-specific modules, i.e., Cross-Correlation Feature Backbone (CCFB), Hierarchical Instance Aggregation Neck (HIAN), and Context-Aware Adaptation Head (CAAH). These modules synergistically fuse general semantics with fine-grained structural cues, enable multi-scale feature aggregation, and adaptively refine predictions based on specific scene contexts. On the GZ-Tree Crown dataset, Tree-SAM achieves F1-scores of 0.762, 0.732, and 0.830, with corresponding AP@50 values of 0.478, 0.454, and 0.526 in the forest, mixed, and urban scenarios, respectively, consistently ranking first across all scenes and demonstrating strong adaptability to diverse intra-city landscapes. Additional evaluations on the BAMFORESTS dataset and the SZ-Dataset further confirm its robustness across varied geographic contexts. Tree-SAM provides a reliable, automated framework for large-scale urban tree mapping, providing reliable data support for urban forest management, carbon stock estimation, and ecological assessment.

In the context of global climate change, extreme climate events are becoming increasingly frequent. Extreme climate events constitute major risks to global food security. The simultaneous occurrence of multiple extreme climate events may have a much greater impact than individual extreme events in isolation. Here we quantitatively analyzed the impact of individual and combined extreme climate indices, including cold days (CD), warm degree days (WDD), precipitation, and compound hot – windy - dry (HWD), on the yields of three major crops (winter wheat, soybeans, and maize) globally by establishing a linear mixed-effects model. CD, HWD, and WDD are identified as the most significant driving factors causing yield losses in winter wheat, soybeans, and maize, respectively. During the planting to the jointing stage, per 10 days of CD account for a 3.2% reduction in winter wheat yield. During the jointing to heading stage, per 10 h of HWD and per 10 °C day-1 WDD result in a 7.5% reduction in soybean yield and a 2.7% reduction in maize yield, respectively. We quantified "yield shocks" and found that the regions experiencing yield shocks exhibit a similar spatial distribution to extreme climate indices. These extreme climate indices are likely to be the driving factors behind yield shocks for the three crops. Our findings indicate that multiple individual extreme climate factors, as well as compound heat-drought-wind (HDW) indices that have been overlooked in traditional risk assessments, impact the yield of the three major crops globally.

Near real-time fine-resolution land surface phenology (LSP) prediction is essential for understanding surface attributes and ecosystem functions, and solving important ecological processes related to phenology at the landscape scale. In this paper, we applied the Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model (ESTARFM) to fuse image pairs of Landsat 8 and Moderate-resolution Imaging Spectroradiometer (MODIS) as train data, and then applied the first derivative method to retrieve phenophase transition dates from fused time series of satellite data as label data. The convolutional neural network (CNN) model was trained using fusion images as inputs and the label data as targets. The trained model was further used to predict LSP dates from individual Landsat images. As evaluated using the reference data, the predict land surface phenological dates and could match the reference well with the coefficient of determination of 0.77 and root mean squared errors of 3.535, and our study provides an alternative method to predict land surface phenological dates using individual Landsat images.