Soil moisture depletion precedes declines in vegetation productivity by days to weeks, creating a natural prediction window for agricultural drought impacts. We test whether satellite root-zone soil moisture (RZSM) from the Soil Moisture Active Passive (SMAP) Level-4 product can provide skillful sub-seasonal forecasts of crop productivity anomalies in Near-Infrared Reflectance of Vegetation (NIRv) across drought-prone croplands of Eastern and Southern Africa. We test three input configurations: NIRv-only, NIRv with concurrent RZSM, and NIRv with explicitly lag-shifted RZSM, to forecast NIRv anomalies at lead times of 8 to 40 days using a Convolutional Long Short-Term Memory (ConvLSTM) model chosen for its ability to jointly capture local spatial dependence and temporal lag structure. All configurations outperform the climatological NIRv and reduce the RMSE by 22–24%. The concurrent RZSM configuration achieves the highest domain-averaged anomaly correlation coefficient (ACC = 0.560), with the soil moisture contribution growing monotonically from negligible at 8 days to +0.056 ACC at 40 days. When we stratify by aridity, the RZSM advantage concentrates in water-limited croplands (mean ΔACC =+0.022 and +0.024 across the 48-day and 80-day input window, respectively, versus ≈0 in energy-limited croplands), where soil moisture–vegetation coupling is stronger. In energy-limited regions, soil moisture adds minimal predictive value. Our case study of the 2024 southern African El Niño drought shows that the RZSM-informed model resolves the spatial pattern of crop-productivity anomalies approximately 32 days in advance, reducing spatial mean absolute error by 30% and improving spatial correlation by 29% over the autoregressive NIRv-only forecast. Our results indicate that SMAP root-zone soil moisture is a viable predictor of sub-seasonal crop-productivity anomalies in water-limited African croplands. Our forecast target is a satellite proxy for vegetation productivity (NIRv anomaly), not crop yield, production, or documented crop failure; translating these cropland-masked anomalies into agricultural-impact outcomes requires crop-specific calibration and integration with food-security indicators.
Coastal ecosystems worldwide are increasingly threatened by the impacts of climate and land use change, which alter habitat structure, disrupt ecological processes, and reduce ecosystem resilience. Satellite imagery provides an important tool for monitoring these changes and informing scientists and policymakers. However, coastal systems are extremely dynamic in space and time, and many fine-scale processes and short-lived events are missed by satellites that provide resolutions coarser than 10 m with multiday revisit times, e.g., Landsat and Sentinel-2. Commercial satellites have historically offered higher spatial resolution, and the emergence of large constellations has enabled a shift toward more systematic global coverage. The Planet Dove constellation combines 3 m spatial resolution with near-daily acquisition, creating a large archive of repeated meter-scale observations of coastal environments. Here, we provided practical insights into the opportunities and limitations of using Planet Dove (PlanetScope) imagery in coastal environments by analyzing global coastal image availability. Within 20 km of the world's coastlines, the Planet Dove archive contains over 100 million images between 2014 and 2024. From 2016 to 2024, considering mostly clear imagery (≥75% clear), each global coastal region was imaged on average 516 times. Of these areas, 6.2% of grid cells had a median revisit of 1 day, 63.6% had a revisit interval of 7 days or less, and 87.5% had a revisit interval of 30 days or less. We examined the implications of these revisit times for tidal observations and found large amounts of spatial variability in tidal coverage provided by Planet Dove. In some areas, the archive imaged nearly the full tidal range and maintained ~weekly revisit times for a given tidal stage. By contrast, certain regions, such as portions of South America and Australia, captured only one tidal mode, and others, such as Alaska, experienced much longer revisit times (e.g., > 3 months) when tidal conditions were controlled for. These results show that a high volume of imagery does not necessarily translate into improved monitoring capability, as the timing and environmental context of observations can determine their suitability for a given application. The spatial and temporal coverage provided by Planet Dove is particularly useful for monitoring short lived coastal processes (e.g., nearshore sediment plumes), mapping small and fragmented coastal habitats, and monitoring applications where controlling for tidal variability is important. However, issues with sensor calibration, geometric quality, and data cost have limited the development of standardized high resolution data products on continental to global scales.
The paper is aimed at the retrieval of total column water vapor (TCWV) over a melting glacier using high spatial resolution hyperspectral measurements performed by the German EnMAP (Environmental Mapping and Analysis Program) satellite. Such measurements are important for understanding the intra-pixel variability of water vapor satellite products with coarser spatial resolution especially in regions with drastic changes in the water vapor abundances. The retrieved TCWV is compared with that derived using other spaceborne optical measurements such as performed by Sentinel-5P. The close correspondence of both retrievals has been found with a smaller variance over derived EnMAP TCWV values.
Land surface temperature (LST) serves as a critical state variable governing Earth system interaction. Although thermal infrared remote sensing allows for global LST monitoring, most studies are based on the assumption of a homogenous surface, and most data fusion studies overlook the effect of complex surface structural characteristics, which aim for a trade-off between spatial and temporal resolutions. To address this challenge, this study introduces a radiative transfer and energy balance-coupled model (STREAM) for global simulations, which explicitly accounts for the effect of vegetation structure in surface physical processes. This enables the simulation of land surface component temperatures (LSCT) and brightness temperatures in a unified manner. The proposed model is validated against satellite directional observations from Sentinel-3 SLSTR and in-situ measurements from FLUXNET, with its performance compared to the widely used SCOPE model. Results demonstrate that: (1) Based on SLSTR observations, STREAM effectively mitigates the underestimation of LST found in the SCOPE model, reducing the overall bias against SLSTR observations from −3.03 K to −1.77 K, and the root mean squared error (RMSE) from 4.52 K to 4.06 K; (2) Based on FLUXNET measurements, STREAM reduces RMSE from 3.36 K to 2.52 K compared to the SCOPE model. This work establishes STREAM as a robust and efficient physical framework for generating global temperature, which highlights a gap in current modeling capabilities: high-fidelity three-dimensional models are too computationally intensive for global-scale applications, whereas prevalent one-dimensional models have limitations in modeling complex vegetation types.
Satellite-based solar-induced chlorophyll fluorescence (SIF) has opened a new era in monitoring gross primary productivity (GPP), and near-infrared reflectance of vegetation (NIRv), an effective proxy for SIF, has enabled observations at high spatial resolution. However, the inherent trade-off between spatial and temporal resolution in remote sensing often leads to the loss of critical information on carbon cycling, particularly in heterogeneous and cloud-prone regions. To address this challenge, this study introduces the unified, high-resolution intelligent carbon quantification and explanation (UNIQUE) framework for high-spatiotemporal-resolution satellite-based GPP monitoring. UNIQUE comprises two components. Part 1 generates high-accuracy GPPL and GPPM from Landsat and the Moderate Resolution Imaging Spectroradiometer, respectively. Part 2 fuses these products to create GPPUNIQUE, which provides daily GPP estimates on a 30 m grid, with fine-scale spatial variability primarily derived from Landsat-based vegetation information. In Part 1, a light gradient boosting machine model using NIRvP, the product of NIRv and photosynthetically active radiation, as the key predictor achieved a root mean square error (RMSE) of 2.17–2.29 gC m-2 d-1, representing a 13–28% reduction relative to existing global GPP products. In Part 2, the proposed stochastic–deterministic combined architecture for fine-level regression (U-SCALER) achieved performance comparable to that of a deterministic UNET baseline. Its performance improved when available Landsat observations were incorporated, achieving an RMSE of 2.60 gC m-2 d-1 at test sites. U-SCALER also showed lower sensitivity to Part 1 errors and image-pair selection. Although meteorological drivers were incorporated at coarser spatial resolutions, GPPUNIQUE represents fine-scale spatial variability on a 30 m grid by leveraging Landsat-derived vegetation information. Overall, UNIQUE provides a practical approach for integrating satellite observations with different spatial and temporal characteristics. U-SCALER in Part 2 also contributes to the evaluation of generative modeling for image fusion and super-resolution. With potential extensions to multiple satellite sensors and to vegetation and carbon-related variables beyond GPP, UNIQUE could support more comprehensive carbon monitoring across diverse domains and spatiotemporal scales.
Statistical and machine learning models, trained with in situ species observations and continuous covariate data, are commonly used to produce marine habitat maps by predicting ecological features over large areas. Here we systematically compare such approaches based on species distribution models (SDM) and more recent remote sensing (RS) models for mapping coastal habitats spanning seven European marine regions. We implemented each model using conventional approaches and data on species occurrences as well as covariates from 1) environmental maps (SDM), 2) Sentinel-2 satellite images and bathymetry (RS), and 3) commercial very high resolution multispectral images and bathymetry (RS – only two areas). We measured each model's predictive performance using cross-validation and analysed how it changed with data quality, depth and water opacity. The SDM approach generally outperformed the Sentinel-2-based RS approaches, being superior in five out of seven areas. However, SDMs performed poorly in the Mediterranean Sea likely due to small variability in the available environmental covariates. RS showed promise for producing benthic habitat maps even though it required larger in-situ datasets and reusing historical survey data proved more challenging than with SDMs. The performance of the SDMs depended on the availability of high-quality, ecologically relevant covariate maps and the taxonomic resolution of the in-situ data. The present study offers an in-depth comparison of the advantages and disadvantages between the approaches, especially on how they perform across different areas for coastal habitat mapping, providing practical guidance on their use, as well as suggestions for further development and co-use.
Observations of water surface elevation (WSE) from the Surface Water and Ocean Topography (SWOT) Mission have been available for almost three years and have shown promising results in preliminary studies over rivers. SWOT's vector node and reach products over rivers will significantly contribute to the computation of river discharge, refine hydraulic and hydrological models, and deliver reliable flood forecasts. Several validation studies to date have confirmed SWOT's superior performance relative to existing altimetry missions and good agreement with in-situ gauge measurements. However, SWOT accuracy varies from site to site. This study presents a systematic evaluation of SWOT WSE measurements across rivers in India, the USA, and Australia, examining how key hydraulic, topographic, and land cover parameters affect WSE performance across regions. We evaluate the effects of different filtering strategies on the retention and relative accuracy of SWOT observations, thereby generating a robust and reliable WSE dataset. Our research establishes degraded geolocation as a major contributor to errors in SWOT observations. River width is identified as a key parameter influencing the relative accuracy of SWOT products. Furthermore, the overall relative accuracy of WSE is higher in the USA and Australia than in India, with the reach product yielding lower RMSE values compared to the node product. Conversely, in India, the reach product shows greater discrepancy relative to gauge measurements, potentially associated with reach-delineation limitations and other site-specific factors.
Multi-source satellite data fusion, such as that between MODIS and Landsat, is widely employed to retrieve land surface temperature (LST) at high spatiotemporal resolutions. However, the directional viewing effects arising from different sensor scanning geometries are generally neglected. To investigate the influence of observation angles on LST, this paper utilizes MODIS and Landsat-8 satellite data and applies a kernel-driven model to simulate and analyze the angular anisotropy of LST over a large portion of the Mediterranean region from 2015 to 2025. The results reveal that LST differences can arise from discrepancies in emissivity used in retrieval algorithms, overpass time mismatches, and systematic cross-sensor biases, with potential deviations of up to 5 K, 3 K, and 3 K, respectively. The viewing-angle difference can lead to a maximum LST discrepancy of approximately −9 K between MODIS and Landsat; roughly 50% of the differences fall between 0.20 K and − 1.9 K, and about 75% are below 0 K. Simulations based on the kernel model show that the values of the emissivity kernel (kernel-A) primarily range from −5 K−1 to 5 K−1, whereas those of the solar kernel (kernel-D) are concentrated between −10 K−1 and 10 K−1. The magnitude of angular anisotropy exhibits a strong spatial correlation with terrain variations and fractional vegetation cover (FVC). Validation against in-situ measurements demonstrates consistent temporal patterns in angular anisotropy, despite minor absolute deviations. Angular correction using the kernel model significantly reduces the LST discrepancy between MODIS and Landsat: the median temperature differences during the dormancy and vegetation periods shift from −0.87 K and − 0.92 K to 0.09 K and − 0.05 K, respectively. This study provides a critical foundation for improving cross-sensor LST consistency between MODIS and Landsat.
Field-based monitoring of tree species in forests is often sparse due to logistical constraints. Remote sensing enables repeated, spatially contiguous collection of reflectance data across large areas. Tree species classification accuracy using such data is variable, likely because most studies use observational datasets where species occurrence correlates with environmental variation. We used two sites of a tree biodiversity experiment in Germany (BIOTREE: Kaltenborn and Bechstedt), where different species have been planted with high replication under controlled diversity levels, to assess how well tree species could be classified using reflectance data from airborne imaging spectroscopy and different classification methods (linear discriminant analysis, LDA, and a non-linear support vector machine, SVM). Reflectance data for 589 wavelengths between 400–2400 nm were acquired at 1 m spatial resolution during peak growing season. Reflectance spectra showed large and significant variation between taxonomic classes, orders, and species, and weak, but still significant, interactions between classes or orders and diversity levels. Classification accuracy reached 100% in training datasets, 77%–83% for the four species in Kaltenborn prediction datasets, and 31%–49% for the 16 species in Bechstedt prediction datasets. LDA provided more accurate predictions than SVM, and using similarly-spaced original wavelengths with LDA was as efficient as using principal components derived from the original data. While airborne imaging spectroscopy effectively distinguished up to four tree species in our datasets, classification accuracy was lower when 16 species had to be distinguished. In these cases, the methodology may be more useful for functional diversity monitoring than for tree species classification.
Orbiting at Sun-Earth Lagrange-1 point, the Earth Polychromatic Imaging Camera (EPIC) provides sunlit Earth images at 10 channels from ultraviolet (UV) to near-infrared (NIR) every 1–2 h. Here, we presented the version 2 of the algorithm to retrieve aerosol optical depth (AOD) and aerosol optical centroid height (AOCH) from visible (443 and 680 nm) and O2 absorption (688 and 764 nm) bands of EPIC (hereafter AOCH-v2 algorithm). The EPIC AOCH-v2 algorithm differs from the version 1 (AOCH-v1) algorithm in several aspects. First, an independent calculation of ultraviolet aerosol index (UVAI) was added by using two UV bands (340 and 388 nm) to identify absorbing aerosols for AOCH retrieval. Second, we updated the real part of aerosol refractive index to represent a new smoke aerosol model for large smoke AOD scenarios over North America, while retaining the original smoke model generated from the Aerosol Robotic Network (AERONET) climatology for background aerosol cases. Third, a new AOD retrieval scheme was developed to constrain the surface reflectance using the climatological surface reflectance ratios. Finally, the cloud mask scheme was improved for cloud screening over bright surfaces. The EPIC AOD from the AOCH-v2 algorithm displays a better agreement with AERONET than AOCH-v1 algorithm. The evaluations against Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) measurements indicate that the AOCH overestimation in AOCH-v1 algorithm is significantly mitigated by AOCH-v2 algorithm, suggesting uncertainties in the EPIC AOCH retrievals are notably reduced through a better characterization of surface reflectance, enhanced AOD retrievals, and an updated smoke aerosol model.
Irrigated agriculture is essential for regional and global food security and poverty alleviation — but growing water scarcity and climate change challenge the balance between agricultural production, communities’ livelihoods and ecosystem sustainability. The Murray-Darling Basin (MDB, ∼1.06 million km2), Australia’s largest river system and primary agricultural region, exemplifies this challenge. This research presents the first multi-decadal (1987–2024), high-resolution (∼30 m) Landsat-based monthly irrigation water-use (IWU) dynamics across the MDB. The assessment is underpinned by a framework using advanced blending and spatiotemporal interpolation that produces ∼30 m monthly, gap-free vegetation indices and actual evapotranspiration (ETa) for the MDB. Spatiotemporal patterns of summer irrigation are identified and predicted by Artificial Neural Networks (ANNs) from the Normalised Difference Vegetation Index (NDVI) and ETa timeseries. Together with soil moisture accounting, they provide timeseries quantifying irrigated areas and estimates of both rainfall-derived and irrigation-derived IWU components. Validation against: (i) independent field ETa; (ii) multi-scale agricultural survey statistics; and (iii) water-use records demonstrated encouraging performance. The annual timeseries of irrigated area and IWU reveal nearly four decades of evolution in irrigated extent, broad irrigation composition, and water-use under shifting climatic and management conditions; offering unprecedented insights into how long-term irrigation dynamics respond to hydroclimate variability and policy reform. Underpinned by operational Earth Observation inputs and an established ETa product, the framework demonstrates how Landsat observations can support systematic irrigation accounting spanning ∼40 years over large regions. While several components remain under development, the framework establishes a scalable and transferable pathway towards a sustained irrigation information product, consistent with emerging principles for operational satellite-based environmental monitoring systems.
Accurate monitoring of maize diseases is crucial for global food security, yet current remote sensing approaches lack mechanistic quantification of disease development. To address this issue, we proposed a disease transmission dynamics module of the cmSCOPE-D (coupled maize - mSCOPE) model by integrating epidemic dynamics with canopy radiative transfer theory, enabling full-process simulation from pathogen infection to canopy optical responses. The proposed module offers advancements in three key areas: 1. Extending the Susceptible-Exposed-Infectious-Removed (SEIR) model into a maize disease transmission dynamics module with environmental response functions, systematically quantifying the effects of temperature, humidity, and growth stage; 2. Establishing dynamic equations for leaf biochemical components as functions of the Disease Index (DI); 3. Formulating rules for 3D lesion spatial distribution that follow natural infection patterns.Our module was validated using field data, with relative errors in DI simulation below 3% and high retrieval accuracy for key biochemical components (chlorophyll and carotenoids; NRMSE <0.2). Our results revealed two key findings: First, the three-dimensional spatial distribution of lesions and disease severity were critical factors driving the canopy spectral signature, with the strongest modulation on reflectance originating from middle-canopy lesions due to their larger projected proportion in the viewing direction. Second, under disease stress, solar-induced chlorophyll fluorescence (SIF) exhibited decreased intensity and a blue-shifted peak, while the reflectance spectrum showed a blue shift in the red-edge position. These findings provide a new theoretical basis for early disease detection. We also identified that current leaf-level radiative transfer models do not adequately represent disease-specific physicochemical changes, highlighting a priority for future model refinement. Overall, the proposed module offers theoretical and methodological advancement for crop disease remote sensing, as it enhances understanding of optical response mechanisms under disease stress and establishes a foundation for intelligent monitoring and early-warning systems.
NASA’s Earth Surface Mineral Dust Source Investigation (EMIT) imaging spectrometer, operating on the International Space Station, has produced the first hectare-scale mineral maps of Earth’s arid regions. This study assesses the temporal consistency of these maps and tests for changes in surface mineralogy over the first two years of data collection. We fit a discrete-state Hidden Markov Model (HMM) to time series data from 39 globally distributed study sites to quantify measurement noise and distinguish sensor artifacts from real mineralogical changes. Our findings indicate that mineral classifications are highly repeatable, showing 80%–98% consistency depending on the specific mineral. We then use the HMM framework to test for transitions in surface mineralogy. This research provides a principled statistical basis for interpreting EMIT mineral products and combining noisy remote time series observations into unified mineral maps.
