Wei Yang

With the advancement of remote sensing technology, image acquisition has become more convenient and the amount of information captured has significantly increased, playing a vital role in numerous fields. However, cloud cover often results in missing image data, severely affecting data usability. In recent years, although deep learning methods have made progress in cloud removal tasks, the complexity of modeling multispectral band relationships and the scarcity of paired data remain major challenges. To address this, this paper proposes a two-stage codebook fusion vector-quantized generative adversarial network (TCF-VQ GAN) and a training framework. The first stage employs synthetic aperture radar (SAR), MODIS, and cloud-free data for unsupervised training; the second stage performs fusion fine-tuning using SAR and MODIS on paired cloudy/cloud-free data. The model incorporates a space-channel jointed gated convolution (SCGC) module to model irregular cloud cover and combines channel attention for band selection, while a dynamically weighted wavelet alignment loss function (DW²A) is designed to enhance multiscale feature representation. Experiments on the SEN12MS-CR and SMILE-CR datasets demonstrate that the proposed method outperforms existing methods across all metrics: on SEN12MS-CR, PSNR is 31.0397 and SAM is 4.7243; they are 33.5191 and 2.1663, respectively, on SMILE-CR. Furthermore, under fixed paired data conditions, simply adding auxiliary and cloud-free data further improves performance, validating the method’s effectiveness in data-scarce scenarios.

Alpine grasslands are vital for biodiversity and ecosystem service delivery, yet their responses to climate change remain a focus of intense scientific interest. While studies have examined separate ecological processes such as biodiversity changes or treeline shifts, broader ecosystem changes are not as well understood. Here, leveraging time series of high‐resolution images from the Landsat satellite, we developed an automated method to track the shifts in the upper limits of alpine grasslands on the high and expansive Tibetan Plateau, a region of high ecological and climatic significance. Our analysis revealed modest upward boundary shifts of −0.55 to 0.99 m year ⁻¹ (2nd–98th percentile; mean = 0.12 m year ⁻¹ ) across the plateau over nearly four decades, from 1986 to 2023, with faster rates in wetter areas, yet drastically slower than the rapid climate‐driven isotherm shifts of 3.35 to 12.04 m year ⁻¹ . This substantial lag, further confirmed by ≤ 3 m resolution satellite images, is potentially attributed to water scarcity, poor soil quality, and a lack of stable substrates beyond the current boundaries and geomorphological features. Consequently, the upward expansion of alpine grasslands on the Tibetan Plateau—attributed to the shift of the upper grassland boundary—was limited to approximately 6100 km ² . Notably, two high‐spatial‐resolution CMIP6 models simulated more rapid upward expansion of alpine grasslands and greater carbon sequestration than observed. These findings underscore the need to integrate local environmental nuances into future predictions. Our study elucidates the resilient yet vulnerable nature of alpine ecosystems, sparking new conversations regarding effective strategies to safeguard these extraordinary landscapes under changing climate.

In recent years, Earth observation using geostationary satellites has experienced remarkable development. In particular, third-generation geostationary satellites, beginning with Japan’s Himawari-8, have enabled sub-hourly measurements of the Earth. Following Himawari-8, the U.S. GOES-R series, China’s FY-4A, Korea’s GK-2A, and Europe’s MTG-1 have successively begun observations, making it possible to observe nearly the entire globe at sub-hourly temporal resolution. Along with these advances, research targeting terrestrial ecosystems such as forests, grasslands, and croplands has rapidly expanded beyond traditional meteorological applications. However, current studies on land-surface ecosystem observation with geostationary satellites still overlook an important but critical issue: topographic effects arising from the relatively large viewing angles of geostationary satellites compared with polar-orbiting satellites. As the distance from the satellite sub-satellite point increases, discrepancies grow between the latitude and longitude coordinates on the reference ellipsoid to which geostationary satellite data are projected and the actual geographic coordinates of the land surface. In addition, topographic features such as high mountains create invisible areas that are partially or entirely invisible to geostationary satellites. Despite these effects, few studies have considered topographic effects when comparing with in situ observations or polar-orbiting satellite data. In this study, we address this issue by using high-resolution digital surface model (DSM) data that are sufficiently detailed to represent sub-pixel topographic variability within individual geostationary satellite pixels. Specifically, we simulated topographic effects on geostationary satellite observations using a 30 m resolution DSM provided by the Japan Aerospace Exploration Agency (JAXA). Our results show that, due to topographic effects, the correspondence between ellipsoidal latitude–longitude coordinates and actual surface coordinates can be shifted by more than one pixel in some regions. We further confirmed that this spatial mismatch leads to differences in the seasonal variation patterns of vegetation indices. In addition, when attempting ray-matching with polar-orbiting satellite observations, we found significant differences between geostationary and polar-orbiting satellite data when topographic effects were not considered. These findings demonstrate that accounting for topographic effects is essential for accurate land-surface observation using geostationary satellites. Our results provide valuable guidance for future studies that aim to compare geostationary satellite data with in-situ observations or to perform data fusion with polar-orbiting satellites, and they will contribute to achieving more precise and reliable land-surface monitoring.

Presents corrections to the paper, (“Global Upscaling of Gross Primary Productivity Using a Simple and Robust Modeling Scheme”).

Dense and widespread tropical rainforests across Southeast Asia are crucial for the global carbon cycle. A new generation geostationary satellite, Himawari-8/9, and onboard sensor, the Advanced Himawari Imager (AHI), enables hyper-temporal vegetation monitoring in this cloud-prone region. However, AHI’s fixed viewing geometry varies spatially, making consistent vegetation monitoring challenging across a wide area at a seasonal scale. This study evaluated four different sun-target-sensor geometry conditions using two-band enhanced vegetation index (EVI2) to achieve near-uniform geometry and enhance the ability to monitor vegetation activities: (i) nadir condition, which used nadir viewing and solar geometry at local solar noon; (ii) local solar noon (LSN) condition, which used pixel-by-pixel fixed viewing and solar geometry at local solar noon; (iii) temporally-constant scattering angle (T-CSA) condition, which used pixel-by-pixel fixed viewing and solar geometry corresponding to the scattering angle which meets the criterion (Gao et al 2024 Remote Sens. Environ. 315 114407); and (iv) spatially-constant scattering angle (S-CSA) condition, which used pixel-by-pixel fixed viewing and solar geometry corresponding to a uniform scattering angle of 140°. Among these, the S-CSA condition most effectively mitigated angular artifacts. The derived EVI2 showed higher correlations with tower-based gross primary productivity (GPP), indicating it better captured seasonal variations in vegetation activity. It also demonstrated spatially consistent temporal variations, least affected by AHI’s unique observation geometry. Applying a near-uniform sun-target-sensor geometry based on the S-CSA condition improves monitoring capability in Southeast Asia and enhances our understanding of vegetation dynamics.

Crop phenological information is essential for understanding the processes of crop growth and development. There is a growing interest in detecting crop phenology using Synthetic Aperture Radar (SAR) data, as the utility of optical data is significantly constrained in cloudy regions owing to insufficient data availability. However, most existing SAR-based phenology detection methods rely on data-driven models that require a substantial number of ground-based phenology samples, which may be challenging to obtain over large spatial extents. The development of corresponding analytical models remains inadequate. In this study, we propose an analytical method incorporating the concept of shape model, referred to as Two-Step Shape Model Fitting (TS-SMF). TS-SMF firstly employs spatiotemporal data for global geometrical matching to simulate crop growth trajectories in SAR time series. Subsequently, TSSMF applies local geometrical matching to more precisely capture crop phenological characteristics. We compared TS-SMF with three existing methods: Shape Model Fitting (SMF), SMF by Separate Phenological Stage (SMF-S), and Dynamic Time Warping (DTW). These methods were applied to Sentinel-1 VH (vertical transmit and horizontal receive) time-series data. The experimental results demonstrate that TS-SMF outperformed other methods, achieving the smallest mean absolute error (MAE) between phenology estimates and phenology observations across multiple datasets, including corn and soybean data from PhenoCam sites during 2019–2022, state-level Crop Progress Reports (CPR) for 2023 in the U.S. Corn Belt, and canola phenology observations from cloudy sites in Sichuan Province, China, in 2024. For instance, the MAE averaged over all six phenological stages for corn in the U.S. Corn Belt was 7.9 days for TS-SMF, whereas the MAE values for other methods ranged from 10.0 to 20.0 days. The ablation experiment further emphasized the necessity of two-step geometrical matching in TSSMF. The TS-SMF extends the application of shape model fitting technology to SAR time series. Its ability to provide reliable phenology estimates with minimal requirements for groundbased phenology samples makes it highly promising for broadscale applications in cloudy regions.

Gross primary productivity (GPP) through photosynthesis is a crucial ecosystem function that significantly influences food security, carbon cycle, and climate change. Current remote sensing estimates of GPP rely on look-up tables containing biome-specific parameters to model light-use efficiency (ε) using coarse resolution interpolated meteorology data, resulting in significant uncertainties in global GPP estimates. To address this challenge, we propose a simple yet effective ecosystem light-use efficiency (eLUE) model to GPP from FLUXNET tower sites to a global scale. Defined as GPP/PAR, eLUE differs from the traditional LUE (GPP/APAR, or ε) in that eLUE essentially integrates canopy light absorption (fAPAR) and the physiological efficiency of photosynthesis (ε), thus eliminating the need for a separate estimate of ε. eLUE was calibrated as a function of MODIS Enhanced Vegetation Index (EVI), and then GPP can be modelled directly as eLUE × PAR. To quantify the carbon cycle error budget, we analytically derived GPP uncertainty based on the law of error propagation. Cross-validation against 120 global FLUXNET sites, encompassing 11 plant functional types (PFTs), demonstrated satisfactory performance of the eLUE model (R² = 0.74, RMSE = 2.05 g C m^-2 d^-1, NSE = 0.74), outperforming or performing comparably to more sophisticated models. Our estimate of global total terrestrial GPP, averaged between 2001 and 2024, is 135.12±11.02 Pg C yr^-1. Meanwhile, we found a significant increasing trend in global total GPP at a rate of 0.26±0.06 Pg C yr^-1 (p < 0.001) from 2001 to 2024, primarily driven by the enhanced CO₂ sequestration in terrestrial ecosystems across the Northern Hemisphere. We suggest that our eLUE model, with its robust performance and clear error representation, will help constrain the global carbon budget and improve the diagnostic analysis of carbon cycle dynamics and climate change feedback. The eLUE-GPP product, available at both global scale and FLUXNET sites, can be accessed for free at: https://doi.org/10.5061/dryad.v9s4mw74h.