Abstract: Difficult access and a lack of in situ data limit monitoring of high-Andean wetlands, which are key components of water regulation in central Chile. This study analyzes the multitemporal dynamics of vegetation in three high Andean wetlands of the headwater (1HW), lateral (2LW), and confluence (3CW) types in the Los Nogales Nature Sanctuary between 2018 and 2025. We integrated Sentinel-2 Level 2A images, annual accumulated precipitation from the ERA5-Land product (lag-1 year), and high-resolution UAV-derived boundaries to characterize six spectral indices (NDVI, EVI, NDRE704, NDRE705, NDWI, and SAVI) and their relationship with water variability. Annual precipitation ranged from ~420 to 780 mm during a regional megadrought. The headwater wetland showed the greatest climate sensitivity, with significant correlations between the previous year's precipitation and NDVI, NDRE705, EVI, SAVI, and NDWI (|R| ≥ 0.70; p < 0.05), while in the lateral and confluence wetlands, the relationships were moderate or weak. Multitemporal mosaics showed maximum vegetative vigor between 2018 and 2021, followed by a decline. Overall, the results confirm that integrating the Sentinel-2 series, climate reanalysis, and UAV delimitation is an effective tool for ecohydrological monitoring and management of high-Andean wetlands.

Abstract: Emerging landslides and severe floods highlight the urgent need to analyse and support predictive models and early warning systems. Soil moisture is a crucial parameter and it can now be determined from space with resolution of few tens of meters, potentially leading to a continuous global monitoring of landslide risk. We address this issue by determining the volumetric water content (VWC) of a testbed in Southern Italy (bare soil with significant flood and landslide hazard) through the comparison of two different satellite observations on the same day. In the first observation (Sentinel-1 mission of the European Space Agency, C-band Synthetic Aperture Radar (SAR)), the back-scattered radar signal is used to determine the VWC from the dielectric constant in the microwave range, also using a time-series approach to calibrate the algorithm. In the second observation (hyperspectral PRISMA mission of the Italian Space Agency), the short-wave infrared (SWIR) reflectance spectra are used to calculate the VWC from the spectral weight of a vibrational absorption line of liquid water (wavelengths 1800−1950nm). As the main result, we obtained a Pearson’s correlation coefficient of 0.4 between the VWC values measured with the two techniques and a separate ground-truth confirmation of absolute VWC values in the range 0.10−0.30 within ±0.05. This overlap validates that both SAR and hyperspectral data can be well calibrated and mapped with 30 meter ground resolution - given the absence of artifacts or anomalies in this particular testbed (e.g. vegetation canopy or cloud presence). If hyperspectral data in the SWIR range become more broadly available in the future, our systematic procedure to synchronise these two technologies in both space and time can be further adapted to cross-validate the global high-resolution soil moisture dataset. Ultimately, multi-mission data integration could lead to quasi-real time hydrogeological risk monitoring from space.

Abstract: Accurate crop yield information is crucial for regional agricultural monitoring; however, many existing approaches rely on complex models or extensive input datasets. This study presents a low-complexity method for estimating county-level maize and sunflower yields in Fejér County, Hungary, using Harmonized Landsat–Sentinel (HLS) Normalized Difference Vegetation Index (NDVI) time series at 30 m spatial resolution. Seasonal NDVI profiles were smoothed using a double-Gaussian fitting approach. Two modelling strategies were investigated: a robust approach using all agricultural pixels and a crop-specific approach restricted to maize or sunflower pixels. The models were tested through leave-one-year-out cross-validation against official yield statistics. For maize, the crop-specific predictive model provided the most accurate estimates (R² = 0.997; mean absolute percentage error (MAPE) = 2.0%). The MAPE remained below 4% even about 30–50 days before the end of harvest. For sunflower, the highest accuracy was obtained using the robust predictive model (R² = 0.928; MAPE = 2.73%). All models showed stable performance across years, including the extreme drought year of 2022. These findings indicate that a simple NDVI-based method can provide reliable county-scale yield estimates and may serve as a practical component in regional monitoring or early-warning systems.

Abstract: Commercial remotely piloted aircraft systems (RPAS) are advancing rapidly, offering improved endurance, expanded sensor payloads, and increasingly sophisticated software capabilities. However, their operational efficiency remains limited by the need for on-site skilled human operators. Semi-autonomous drone-in-a-box (DIAB) systems are emerging as a practical solution, enabling automated, repeatable missions for applications such as construction monitoring, security, and critical infrastructure inspection. Beyond industry, these systems hold significant promise for scientific research, particularly in long-term environmental monitoring where cost, accessibility, and safety are critical factors. In this technology demonstration, we detail the system implementation, discuss flight-planning challenges, and assess the overall feasibility of deploying a DJI Dock 2 DIAB system for remote monitoring of an unstable mountain slope in northwestern Canada (Yukon Territory). The system was deployed approximately 2.5 km from the landslide and operated remotely from across the country in Montreal about 4,000 km away. This study highlights the potential of DIAB systems to support reliable, low-maintenance monitoring of remote natural hazards.

Abstract: Modern Earth observation combines high spatial resolution, wide swath, and dense temporal sampling, producing image grids and sequences far beyond the regime of standard vision benchmarks. Convolutional networks remain strong baselines but struggle to aggregate kilometre-scale context and long temporal dependencies without heavy tiling and downsampling, while Transformers incur quadratic costs in token count and often rely on aggressive patching or windowing. Recently proposed visual state-space models, typified by Mamba, offer linear-time sequence processing with se-lective recurrence and have therefore attracted rapid interest in remote sensing. This survey analyses how far that promise is realised in practice. We first review the theoretical substrates of state-space models and the role of scanning and serialization when mapping two- and three-dimensional EO data onto one-dimensional sequences. A taxonomy of scan paths and architectural hybrids is then developed, covering cen-tre-focused and geometry-aware trajectories, CNN– and Transformer–Mamba back-bones, and multimodal designs for hyperspectral, multisource fusion, segmentation, detection, restoration, and domain-specific scientific applications. Building on this ev-idence, we delineate the task regimes in which Mamba is empirically warranted—very long sequences, large tiles, or complex degradations—and those in which simpler op-erators or conventional attention remain competitive. Finally, we discuss green com-puting, numerical stability, and reproducibility, and outline directions for phys-ics-informed state-space models and remote-sensing-specific foundation architectures. Overall, the survey argues that Mamba should be used as a targeted, scan-aware com-ponent in EO pipelines rather than a drop-in replacement for existing backbones, and aims to provide concrete design principles for future remote sensing research and op-erational practice.

Abstract: The land use and land cover (LULC) of many regional landscapes are changing due to natural effects and anthropogenic activities, impacting biodiversity and ecosystem services. LULC dynamics reflect the altered flow of energy, water, and greenhouse gases, influencing the pillars of sustainability: society, environment, and economy. Thus, assessing LULC changes is vital for understanding the relationship between nature and society. This study used multi-temporal remotely sensed imagery to examine LULC change between 1990 and 2019 in the context of Forest Transition Theory (FTT) across the Greater Shawnee National Forest (GSNF) area of southern Illinois, USA, using a Random Forest algorithm, and projecting change to 2050 with a Land Change Model integrated with IPCC temperature and precipitation scenarios. From 1990 to 2019, LULC analysis showed increases in deciduous forest (1.35%), mixed forest (26.40%), agriculture (2.15%), and built-up areas (6.70%), while hay/grass/pasture declined (16.0%). LULC change intensity was highest from 1990 to 2001 (2.35% annually), slowing to 0.23% (2001–2010) and 0.18% (2010–2019). LULC classification accuracy ranged from 92.9% to 95.9% with kappa coefficients of 0.89–0.94. Projections to 2050 showed consistent increases in built-up areas (17.12%–42.61%), water (28.75%–39.70%), and hay/grass/pasture (6.23%–38.38%), while overall forest cover declined in all scenarios. Deciduous forests decreased by 3.11%–19.87% and were replaced by mixed forests in some scenarios (12.45%-23.63%), while evergreen forests showed mixed responses, ranging from a decline of up to 17.13% to an increase of 2.90%. The results showed that the GSNF broadly follows the FTT framework: forest recovery since 2001 coincided with rural depopulation, slow agricultural expansion, and rising incomes. However, climate change is expected to disrupt this recovery, pushing transitions toward mixed and evergreen forests. Findings demonstrate the importance of integrating remote sensing-based LULC with socio-economic trends and climate adaptation strategies to sustain forests and ecosystem services under future environmental pressures.

Abstract: Over the past decade, drones have moved from experimental tools to practical workhorses in environmental monitoring. Their real value lies in the sensors and actuators they carry, which determine the kind of information that can be collected and how reliably it can be gathered. This review looks at the range of sensor systems now used in ecological, forestry, and agricultural work—from standard RGB cameras to multispectral and hyperspectral units, LiDAR, thermal imagers, and radar. It also considers the often-overlooked role of actuators, such as gimbals and stabilizing mounts, that keep sensors steady and improve data quality, as well as devices for collecting physical samples. Drawing field-based examples, this study explores how these technologies are deployed for tasks like mapping forest biomass, tracking habitat change, assessing fire recovery, and monitoring crops. It concludes by discussing practical constraints, emerging sensor designs, and the likely direction of drone-based monitoring in the years ahead.

Abstract: Zachlod et al. [1] analysed satellite data from Malaysian oil palm plantations and concluded that RSPO certification reduces plantation efficiency by lowering canopy coverage. Our reanalysis of 93,987 ha shows that their methods and interpretation are fundamentally flawed, primarily because they failed to account for routine replanting, which caused temporary canopy loss in 32.7% of certified areas between 2018 and 2023. Using validated remote sensing methods and cross-verified timelines, we found no significant decline in oil palm coverage and no evidence of reduced oil palm coverage in certified plantations. Given the risk of policy misinterpretation, we recommend that Zachlod et al. correct or retract their study and call for more rigorous, transparent, and context-aware evaluations of sustainability outcomes in plantation systems.

Abstract: Foundation models such as AlphaEarth introduce a new paradigm in remote sensing by providing semantically rich, pretrained embeddings that integrate multi-sensor, spatio-temporal, and contextual information. This study evaluates the performance of AlphaEarth embeddings for land-cover classification under both pixel-based and object-based paradigms within the Google Earth Engine (GEE) environment. Sentinel-2 imagery for 2024 was used to map a 1,930-hectare region in Pabbi Tehsil, Khyber Pakhtunkhwa, Pakistan, where rapid urbanization is reshaping traditional land use. Four experimental configurations—Pixel-Based Spectral Indices (PBSI), Pixel-Based AlphaEarth Embeddings (PBAE), Object-Based Spectral Indices (OBSI), and Object-Based AlphaEarth Embeddings (OBAE)—were implemented using a Random Forest classifier.The results show that AlphaEarth embeddings consistently outperformed spectral index–based models, improving overall accuracy by ≈ 5 percentage points and Kappa by ≈ 3. Object-based approaches enhanced spatial coherence and boundary delineation, particularly for built-up and road classes, while maintaining stable area statistics across pipelines. The findings demonstrate that pretrained embeddings can achieve deep-learning-level accuracy through lightweight, cloud-native workflows, offering an efficient pathway for land-cover mapping and urban-cadastral monitoring in data-scarce regions.

Abstract: We present a novel framework for hyperspectral satellite image classification that explicitly balances spatial nearness with spectral similarity. The proposed method is trained on closed-set datasets, and it was tested on zero-shot agricultural scenarios that include both class distribution shifts, and presence of novel and absence of known classes. This scenario is reflective of real-world agricultural conditions, where geographic regions, crop types, and seasonal dynamics vary widely and labeled data are scarce and expensive. The input data are projected onto a lower-dimensional spectral manifold, and a pre-trained pixel-wise classifier generates an initial class probability saliency map. A kernel-based spectral-spatial weighting strategy fuses the spatial-spectral features. The proposed approach improves the classification accuracy by 7.22%–15% over spectral-only models on benchmark datasets after iterative convergence. Incorporating the additional unsupervised learning and weak labeling helped surpass several recent state-of-the-art methods. Requiring only 1%–10% labeled training data and at most two tuneable parameters, the framework operates with minimal computational overhead, qualifying it as a data-efficient and scalable learning solution. Recent deep architectures, although exhibiting high closed-set accuracies, often show limited transferability under low-label, open-set or zero-shot agricultural conditions where class distributions shift and novel classes emerge. The rice crop play a pivotal role in global food security but are also a significant contributor to greenhouse gas emissions, especially methane, and extent mapping is very critical. We demonstrate transferability to new domains-including unseen crop class (e.g. , paddy), seasons, and regions (e.g. , Piedmont, Italy)—without re-training. This work presents a novel perspective on hyperspectral classification and domain transferability, suited for sustainable agriculture with limited labels and low-resource domain generalization.

Abstract: Vegetation plays an important role in the exchange of heat and moisture in the earth-atmosphere system, and the vegetation index can not only better measure the growth and changes of vegetation, but also provides required data for meteorological, hydrological, ecological and other studies. Therefore, in-depth study of the spatial and temporal changes in regional vegetation index is of great significance for guiding eco-logical environment protection and governance. This paper takes the northern foot-hills of the Qinling Mountains and its north area in Shaanxi as the research object, an-alyzes the sensitivity of normalized vegetation index (NDVI) acquired from Sentinel-2, Landsat and MODIS data to the phenological characteristics of vegetation, and uses the linear regression slope method and partial correlation coefficient method to ana-lyze and discuss the spatiotemporal distribution characteristics and influencing factors of NDVI at different time scales. The results show that: 1) Sentinel-2 images and MODIS images have better spatiotemporal consistency in obtaining monthly NDVI data and refining vegetation phenological characteristics in the study area than Landsat images; 2) NDVI at different time scales in the study area showed a signifi-cant upward trend during 2001-2023, with a large NDVI increase mainly concentrated in the eastern part of the study area, especially in the Lishan Mountain area, while the areas with a decrease or no significant change in NDVI are mainly urban areas where human life and production are relatively frequent; 3) The NDVI in most areas of the study area showed an increasing trend from 2019 to 2023, especially in the Lishan Mountain area within the ecological protection and restoration project area, but there was a large area of NDVI decrease in the western part of the Qinling mountain front flood plain; 4) The lagging effects of precipitation and air temperature on monthly NDVI are both positive in a short period of time. Meanwhile, in spring and monthly scales, precipitation and air temperature have a positive correlation with vegetation growth in the study area, and the influence of air temperature is more significant.

Abstract: This study investigates the spatial and temporal dynamics of carbon dioxide (CO₂) and methane (CH₄) in the Hefei region from 2009 to 2020 using satellite-based column ob-servations (GOSAT) and the HYSPLIT model. The results reveal a significant increasing trend for both greenhouse gases, with Hefei's CO₂ growth rate (2.43 ppm/year) exceed-ing the global average. Pronounced seasonal cycles were identified: CO₂ concentrations peak in winter due to enhanced fossil fuel combustion for heating and weakened plant photosynthesis, and reach a minimum in summer owing to strong biospheric carbon uptake. In contrast, CH₄ concentrations are highest in summer and autumn, likely driven by agricultural activities such as rice cultivation, and lowest in winter and spring. Diurnal variations show a synchronous pattern for CO₂ and CH₄, peaking around noon, influenced by the interplay of anthropogenic emission cycles and plane-tary boundary layer dynamics. Backward trajectory analysis and clustering further elu-cidate that the transport pathways and source regions of greenhouse gases are domi-nantly controlled by the East Asian monsoon. The summer is characterized by clean, locally-influenced marine air masses, while the winter is predominantly influenced by long-range transport from the polluted northwestern interior of China, which acts as a critical pollution corridor. This research underscores the combined roles of local an-thropogenic activities, regional transport, and meteorological systems in driving the characteristics of greenhouse gases in an inland subtropical city.

Abstract:

Geostationary atmospheric motion vectors (e.g., FY4A AMVs) are routine mid-upper atmospheric observations used in numerical weather prediction (NWP) models, yet their complex spatiotemporal errors and assimilation limitations, i.e., high-temporal/coarse-spatial data and large-scale-adjustment/direct-assimilation scheme, leave unclear impacts of AMVs assimilation on nowcasting forecasts. To this end, a Nudging-Forced-3DVar (NFV) scheme is designed within a multi-scale (i.e., 12-, 4-, and 1-km) regional NWP framework to exploit AMVs characteristics; ablation experiments for the Zhengzhou “7·20” rainstorm isolate Nudging and 3DVar impacts on assimilation and nowcasting. Results show: 1) large-scale Nudging and high-resolution 3DVar both improve mid-upper analyses, with the former ingesting more observations; 2) Nudging retains large-scale background updates but yields significant misses, whereas 3DVar intensifies rainfall extremes yet blurs fine structures; 3) NFV merges their strengths, modulating deep convection through upper-level systems and markedly improving rainfall spatiotemporal patterns. Therefore, NFV is recommended for the FY4A AMVs’ future numerical nowcasting, which provides useful guidance for regional application of geostationary 3D-winds.

Abstract: Low clouds significantly impact weather, climate, and multiple environmental and economic sectors such as agriculture, fire risk management, aviation, and renewable energy. Accurate knowledge of cloud base height (CBH) is critical for optimizing crop yields, improving fire danger forecasts, enhancing flight safety, and increasing solar energy efficiency. This study evaluates a shadow-based cloud base height retrieval method using MODIS satellite visible imagery and compares the results against collocated lidar measurements from the MPLNET ground stations. The shadow method leverages sun–sensor geometry to estimate cloud base height from the displacement of cloud shadows on the surface, offering a practical and high-resolution passive remote sensing technique, especially useful where active sensors are unavailable. Validation results show strong agreement, with a correlation coefficient of R = 0.96 between shadow-based and lidar-derived CBH estimates, confirming the robustness of the approach for shallow, isolated cumulus clouds. The method’s advantages include direct physical height estimation without reliance on cloud top heights or stereo imaging, applicability across archived datasets, and suitability for diurnal studies. This work highlights the potential of shadow-based retrievals as a reliable, cost-effective tool for global low cloud monitoring, with important implications for atmospheric research and operational forecasting.

Abstract: Earthquake prediction remains one of the central unsolved problems in geophysics, and ionospheric variability offers a promising yet debated window into the earthquake preparation process through lithosphere–atmosphere–ionosphere coupling. Progress has been hindered by methodological limitations in prior studies, including the use of inappropriate performance metrics for highly imbalanced seismic data, the reliance on geographically and temporally narrow data, and inclusion of inherent spatial or temporal features that artificially inflate model performance while preventing the discovery of genuine ionospheric precursors. To address these challenges, we introduce a global, temporally validated machine learning framework grounded in thirty-five years of ionospheric observations from thirty-seven stations. Our framework eliminates lookahead bias through strict temporal partitioning, evaluates precursor sensitivity through systematic relaxation of the Dobrovolsky radius, and applies ensemble feature selection that excludes spatial and temporal identifiers to prevent leakage and coincidence effects. Cross-regional validation shows that gradient boosting models yield the strongest classification skill, with a weighted F1 score of 77%, and that ionospheric parameters account for a substantial portion of earthquake magnitude variability. These findings provide quantitative support for LAIC processes while revealing the multivariate nature of seismic precursors. Our study demonstrates learnable, spatiotemporally invariant ionospheric precursors, though the 15.4% of undetected earthquakes indicates that ionospheric monitoring alone is insufficient for operational deployment and multimodal fusion with complementary precursor channels is required.

Abstract: Hyperspectral Image (HSI) classification is a critical task in remote sensing, yet it faces significant challenges including spectral redundancy, complex spatial-spectral dependencies, and the scarcity of labeled samples. While deep learning models, especially State-Space Models such as Mamba, show promise, current approaches often employ fixed spectral transformations and may not fully capture intricate spatial-spectral relationships. To address these limitations, we propose the Adaptive Spectral-Spatial Fusion Mamba (ASSF-Mamba) framework, designed for superior HSI classification through adaptive spectral processing and enhanced context-aware spatial-spectral fusion. ASSF-Mamba integrates three novel modules: an Adaptive Spectral Projection and Decorrelation (ASPD) module for learnable spectral dimension reduction; a Contextual Mamba Fusion (CMF) module extending Mamba with multi-scale spatial attention and cross-dimensional modulation for long-range dependency capture; and a Hierarchical Feature Enhancement (HFE) module employing multi-level residual connections and adaptive gating for robust feature representation. Comprehensive experiments on benchmark datasets including Indian Pines, Kennedy Space Center, and Houston demonstrate that ASSF-Mamba consistently achieves state-of-the-art classification accuracies, significantly outperforming a wide range of baselines, including advanced Mamba-based models. Furthermore, our framework exhibits superior robustness under limited training data conditions, maintains competitive computational efficiency, and yields visually more coherent classification maps. An ablation study confirms the critical contribution of each proposed module to the overall performance.

Abstract: Forest resources are among the most important ecosystems on the earth. The semantic segmentation and accurate positioning of ground objects in forest remote sensing (RS) imagery are crucial to the emergency treatment of forest natural disasters, especially forest fires. Currently, most existing methods for image semantic segmentation are built upon convolutional neural network (CNN). Nevertheless, these techniques face difficulties in directly accessing global contextual information and accurately detecting geometric transformations within the image’s target regions. This limitation stems from the inherent locality of convolution operations, which are restricted to processing data structured in Euclidean space and confined to square-shaped regions.Inspired by the Graph Convolution Network (GCN) with robust capabilities in processing irregular and complex targets, as well as Swin Transformers renowned for exceptional global context modeling, we present an innovative semantic segmentation framework for forest remote sensing imagery termed GSwin-Unet. This framework embeds GCN model into Swin-Unet architecture, and for the first time apply the method of combining GCN and Transformer in the domain of forest RS imagery analysis. GSwin-Unet features an innovative parallel dual-encoder architecture of GCN and Swin transformer. First, we integrate the Zero-DCE (Zero-Reference Deep Curve Estimation) algorithm into GSwin-Unet to enhance forest RS image feature representation. Second, a feature aggregation module (FAM) is proposed to bridge the dual encoders by fusing GCN-derived local aggregated features with Swin transformer-extracted features. Our study demonstrates that the GSwin-Unet significantly improves performance on the Forest Remote Sensing Dataset and exhibits good adaptability on GID dataset.

Abstract: High-resolution Digital Terrain Models (DTMs) are essential for precise terrain analy-sis, yet their production remains constrained by the high cost and limited coverage of LiDAR surveys. This study introduces a deep learning framework based on a modified Residual Channel Attention Network (RCAN) to super-resolve 10 m DTMs to 1 m res-olution. The model was trained and validated on a 568 km² LiDAR-derived dataset us-ing custom elevation-aware loss functions that integrate elevation accuracy (L1), slope gradients, and multi-scale structural components to preserve terrain realism and ver-tical precision. Performance was evaluated across 257 independent test tiles repre-senting flat, hilly, and mountainous terrains. A balanced loss configuration (α = 0.5, γ = 0.5) achieved the best results, yielding Mean Absolute Error (MAE) as low as 0.83 m and Root Mean Square Error (RMSE) of 1.14–1.15 m, with near-zero bias (–0.04 m). Er-rors increased moderately in mountainous areas (MAE = 1.29–1.41 m, RMSE = 1.84 m), confirming the greater difficulty of rugged terrain. Overall, the approach demonstrates strong potential for operational applications in geomorphology, hydrology, and land-scape monitoring, offering an effective solution for high-resolution DTM generation where LiDAR data are unavailable.

Abstract:

Cocoa production in West Africa—dominated by Côte d’Ivoire, Ghana, Nigeria, Cameroon, and Togo—faces interconnected agronomic, environmental, and socio-economic challenges that limit productivity and threaten smallholder livelihoods. Integrating Regenerative Agriculture (RA), Unmanned Aerial Systems (UAS), and Artificial Intelligence (AI) present a transformative framework for achieving sustainable and climate-resilient cocoa farming. This review synthesizes evidence from 2000 to 2024 and establishes a tri-axial model that unites ecological regeneration, spatial diagnostics, and predictive intelligence. Regenerative practices such as composting, mulching, cover cropping, and agroforestry rebuild soil organic matter, enhance biodiversity, and strengthen ecosystem services. UAS-based multispectral, thermal, and LiDAR sensing provide high-resolution insights into canopy vigor, nutrient stress, and microclimatic variability across heterogeneous cocoa landscapes. When coupled with AI-driven analytics for crop classification, disease detection, yield forecasting, and decision support, these tools collectively enhance soil organic carbon by 15–25%, stabilize yields by 12–28%, and reduce fertilizer and water inputs by 10–20%. The integrated RA–UAS–AI framework also facilitates carbon-credit quantification, ecosystem-service valuation, and inclusive participation through cooperative drone networks. Overall, this convergence defines a precision-regenerative model tailored to West African cocoa systems, aligning productivity gains with ecological restoration, resilience, and regional sustainability.

Abstract: This study estimated peanut (Arachis hypogaea L.) leaf area index (LAI), a critical vegetation parameter, using spectral bands and vegetation indices (VIs) derived from PlanetScope (~3m) imagery by comparing Random Forest (RF), eXtreme Gradient Boosting (XGBoost), and Partial Least Squares Regression (PLSR) algorithms. Most VIs exhibited strong relationships with LAI but showed saturation when LAI reached 3 m²/m². Thirteen VIs were individually evaluated for estimating LAI using the aforementioned machine learning and statistical algorithms, and the results showed that the best single predictors of LAI are: SR and RTVIcore (RF, R2 = 0.84, RMSE = 0.62 m2/m2); RTVIcore (XGBoost, R2 = 0.88, RMSE = 0.52 m2/m2); and RTVIcore and MSAVI (PLSR, R2 = 0.61, RMSE = 0.96 m2/m2). The top six ranked VIs were selected to calibrate the RF, XGBoost, and PLSR algorithms. The validation of the algorithms showed that the RF achieved the highest prediction accuracy (R2 = 0.844, RMSE = 0.858 m²/m², RRMSE = 25.17%), followed by XGBoost (R2 = 0.808, RMSE = 0.92 m²/m², RRMSE = 26.99%), while the PLSR showed relatively lower model accuracy (R2 = 0.76, RMSE = 0.983 m²/m², RRMSE = 28.85%). Further results demonstrate that VIs derived from spectral bands provide superior model accuracy in estimating peanut LAI compared to the use of spectral bands alone. Overall, the presented results are significant for future crop monitoring using RF to reduce overreliance on multiple models for peanut LAI.