International Journal of Remote Sensing

Dash, Saroj Kumar; Sembhi, Harjinder; Sinha, Rajiv

doi: 10.1080/01431161.2025.2593684pmid: N/A

Airborne remote sensing has facilitated high-resolution canopy and soil water assessment, especially in agriculture-intensive regions. This study presents a field-scale assessment of crop water stress index (CWSI) and soil moisture (SM) using unmanned aerial vehicle (UAV)-mounted thermal imagery combined with in-situ hydrometeorological data over a key agricultural site in India’s Ganga basin. The UAV-based LST (LST-Aerial) is modelled using the spectral emissivity from ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) spectral library and atmospheric radiances from a radiative transfer model. The LST-Aerial is subsequently utilized to estimate field-scale CWSI (CWSI-Aerial) and SM (SM-Aerial) across two principal crop seasons (paddy and wheat) using an empirical and multiple linear regression model, respectively. While the radiometer data reveals a significant correlation (R2 = 0.58 to 0.76, p < 0.001) between canopy – air temperature difference and vapour pressure deficit, a temperature difference of ~ 5°C was noticed between non-transpiring baselines of paddy and wheat for the airborne window. The radiometer-derived CWSI (CWSI-Rad) showed a relatively higher value (0.57) during the wheat season compared to paddy (0.19), reflecting the influence of monsoon-fed cropping in north India. Partial least squares regression reveals solar radiation and relative humidity as major meteorological drivers of CWSI-Rad during the paddy and wheat growing period, respectively. While CWSI-Aerial demonstrated superior accuracy (R2 = 0.85, p < 0.05) to CWSI-Rad, it exhibited a high negative correlation (R =  −0.75 to −0.97) to concurrent SM and SM-Aerial. Additionally, the predicted SM-Aerial agrees well with ground-based SM with errors ranging from 0.01–0.14 m3.m−3, showcasing the robust UAV-based SM prediction. Findings of this study offer valuable insights into smart and precision agriculture, enabling well-informed regulation of crop water resources in tropical water-limited regions.

Wu, Yu-Qing; Guo, Li-Biao; Tan, Wei-Xian; Huang, Ping-Ping; Ma, Ming-Ze; Zhao, Yang

doi: 10.1080/01431161.2025.2593683pmid: N/A

Grassland aboveground biomass (gAGB) is an important indicator for assessing ecosystem services. Traditional ground plot harvesting is accurate but costly and spatially sparse, limiting its ability to represent accuracy and heterogeneity. Existing models seldom consider scale dependence or the effects of management and disturbance. This study examined enclosed and grazed grasslands in Xilingol and Hulunbuir, Inner Mongolia. Centimetric Unmanned Aerial Vehicle (UAV) Light Detection and Ranging (LiDAR) data were used to derive the canopy height model (CHM), canopy gap fraction (CGF), and leaf area index (LAI) at spatial resolutions from 0.05 m to 0.25 m. Spatial heterogeneity was quantified using geostatistical semi-variograms (as nugget, sill, range) based on canopy structure parameter and linked to the accuracy of gAGB models analysis. Three modelling approaches were evaluated: linear regression (LR), random forest (RF), and backpropagation neural network (BPNN) in gAGB estimation. The compared results show that grazing disturbance produced a more homogenized canopy structure across all spatial resolutions. Compared with CGF and LAI, CHM shown lower heterogeneity and higher stability, and its inclusion contributes more to enhancing inversion accuracy. RF model achieved higher and more stable accuracy than LR and BPNN. The results also showed that finer resolutions (0.05 m) captured micro-scale variation more effectively and were suitable for high-precision monitoring. The proposed workflow provides spatially continuous, centimetric structural information and applies heterogeneity diagnostics to guide the selection of suitable gAGB measuring predictors and spatial scales. Based on the grazing sample plots and dataset analysed, this approach significantly improved gAGB modelling accuracy and offers a practical basis for fine-scale biomass monitoring and grassland management.

Karjalainen, Väinö; Koivumäki, Niko; Hakala, Teemu; Muhojoki, Jesse; Hyyppä, Eric; George, Anand; Suomalainen, Juha; Honkavaara, Eija

doi: 10.1080/01431161.2025.2579803pmid: N/A

Drones are increasingly used in forestry to capture high-resolution remote sensing data, supporting enhanced monitoring, assessment, and decision-making processes. While operations above the forest canopy are already highly automated, flying inside forests remains challenging, primarily relying on manual piloting. In dense forests, relying on the Global Navigation Satellite System (GNSS) for localization is not feasible. In addition, the drone must autonomously adjust its flight path to avoid collisions. Recently, advancements in robotics have enabled autonomous drone flights in GNSS-denied obstacle-rich areas. In this article, a step towards autonomous forest data collection is taken by building a prototype of a robotic under-canopy drone utilizing state-of-the-art open source methods and validating its performance for data collection inside forests. Specifically, the study focused on camera-based autonomous flight under the forest canopy and photogrammetric post-processing of the data collected with the low-cost onboard stereo camera. The autonomous flight capability of the prototype was evaluated through multiple test flights in boreal forests. The tree parameter estimation capability was studied by performing diameter at breast height (DBH) estimation. The prototype successfully carried out flights in selected challenging forest environments, and the experiments showed promising performance in forest 3D modelling with a miniaturized stereoscopic photogrammetric system. The DBH estimation achieved a root mean square error (RMSE) of 3.33 - 3.97 cm (10.69 - 12.98 %) across all trees. For trees with a DBH less than 30 cm, the RMSE was 1.16 - 2.56 cm (5.74 - 12.47 %). The results provide valuable insights into autonomous under-canopy forest mapping and highlight the critical next steps for advancing lightweight robotic drone systems for mapping complex forest environments.

Gabrič, Adam; Grigillo, Dejan; Kokalj, Žiga

doi: 10.1080/01431161.2025.2593570pmid: N/A

Despite the importance of woody vegetation landscape features (individual trees, rows of trees, hedgerows, riparian vegetation etc.) in halting the loss of biodiversity in agricultural areas, data on their extent remains lacking. We classified woody vegetation (i.e. the woody vegetation landscape features and forest) from the Slovenian national orthophoto. We used convolutional neural networks with the goal of providing the most accurate classification in the agricultural land. We tested whether adding patches, covered with forest, to the training dataset would improve the classification of woody vegetation in the agricultural land. The results show significant differences in the classification performance of the models when comparing patch sizes, model architectures, band combinations and maximum forest cover of the training patches. The top-performing architecture is HRNet (High-Resolution Network); the patches that provide the best results include red, green and near-infrared spectral bands, have sides of length 256 pixels and are covered with forest to a maximum extent of 60%. With these parameters, a model achieved Jaccard index of 80.13% in agricultural areas. We used this model to classify woody vegetation in Slovenian outstanding landscapes, which proved the model is applicable in the majority of the country. Most misclassifications are in specific environments (e.g. surfaces covered with reed and saltworks), which were not represented in the training data. The comprehensive model assessment in diverse Slovenian landscapes also informs similar efforts worldwide.

de Matos Valerio, Aline; Montanari, João L.; Kampel, Milton; Ward, Nicholas D.; Richey, Jeffrey E.; Cunha, Alan C.

doi: 10.1080/01431161.2025.2596811pmid: N/A

Chlorophyll-a concentration (Chla) is a key indicator of phytoplankton biomass and aquatic trophic status. However, satellite-derived Chla in sediment-rich waters, such as those found in the Lower Amazon River, remains challenging. The present study characterizes in situ Chla levels and their relationships with geographic, physical, and biogeochemical parameters in the Lower Amazon. Data collected between 2014 and 2017 across four hydrological seasons included measurements of Chla, remote sensing reflectance, and water quality parameters such as total suspended sediment, conductivity, water surface temperature, dissolved oxygen, pH, dissolved organic carbon and coloured dissolved organic matter. An empirical model was developed to estimate Chla using simulated hyperspectral bands from NASA’s PACE mission, achieving high performance (R2 = 0.76; RMSE = 0.11 μg·L−1). Red bands proved particularly effective for Chla retrieval, while the addition of ultraviolet bands further enhanced model accuracy. The application of the developed model to satellite imagery yielded results consistent with in situ observations for the same hydrologic season. Seasonal variation and geographic location were major factors influencing Chla dynamics. This study provides a novel contribution to Chla estimation in optically complex, highly turbid waters and highlights the potential of the PACE mission to enhance global aquatic ecosystem monitoring. By offering freely available hyperspectral data with high radiometric resolution, PACE represents a significant advancement in the realm of remote sensing of aquatic environments.

Xiao, Songyi; Zhong, Shengyiliu; Zhu, Liangliang; Xie, Jialei; Geng, Xiurui

doi: 10.1080/01431161.2025.2593685pmid: N/A

Hyperspectral band selection (BS) methods relying solely on volume-based criteria often fail to exclude noisy bands and may yield redundant subsets with limited representativeness. To address these limitations, we propose two novel BS algorithms that integrate canonical correlation analysis (CCA) and projection energy measures with determinant-based volume criterion. By modelling the correlation structure between selected and unselected bands, we effectively suppress the inclusion of noisy or outlier bands and enhance both the representativeness and independence of the chosen subsets. Furthermore, we introduce complexity reduction techniques, including the simplified computation of canonical correlation coefficients and recursive updates of inverse matrices, to facilitate efficient implementation for large-scale datasets. Experimental results on five real-world hyperspectral datasets demonstrate that the proposed methods achieve higher classification accuracies, greater robustness against noise, and significantly lower computational costs compared to traditional volume-based and linear algebra-based BS algorithms. These findings underscore the effectiveness and practical value of integrating correlation structure into volume-based BS frameworks.

Coffer, Megan M.; Schaeffer, Blake A.; Salls, Wilson B.; Minucci, Jeffrey M.; Cronin-Golomb, Olivia

doi: 10.1080/01431161.2025.2575515pmid: 41736721

Satellite constellations often launch platforms over several years, increasing observational frequency and capturing additional, potentially more extreme, events. Consequently, reported changes in satellite-derived data may inadvertently capture variations in observational frequency rather than true environmental trends. This study used the Sentinel-3 Cyanobacteria Index (CI-cyano) to assess impacts of varying observational frequency on data distributions and trends. Daily CI-cyano was temporally aggregated into weekly composites using maximum, mean, and median values as both continuous and ordinal observations. Sentinel-3A, Sentinel-3B, and combined Sentinel-3A & -3B were compared using the Wilcoxon signed-rank test. For continuous observations, temporally aggregating via the maximum value showed a large 9% increase for combined Sentinel-3A & -3B versus Sentinel-3A or Sentinel-3B individually, compared to a small 1% decrease for temporal aggregation via the mean and negligible differences via the median. For ordinal observations, temporal aggregation via the maximum and mean showed large increases of up to 25% for combined Sentinel-3A & -3B, while the median showed small decreases of up to 5%. The seasonal Mann-Kendall trend test was then applied to Sentinel-3 imagery from 2016 to 2023, with and without observations from Sentinel-3B. Temporal aggregation via the maximum showed a moderate 20% increase with Sentinel-3B compared to a small 8% increase without Sentinel-3B; mean and median showed negligible trends. An abbreviated assessment using Sentinel-2 had similar results, with large increases for combined Sentinel-2A & -2B via the maximum, but small and moderate decreases via mean and median. Results suggest that temporal aggregation impacts multi-platform datasets. For more consistent summaries, continuous datasets should be temporally aggregated using the mean or median, and ordinal datasets using the median. Results are applicable to any satellite-derived water quality datasets with varied observational frequency. This study addresses a critical gap in the remote sensing community, ensuring relevant statistical concepts are appropriately applied in multi-platform analyses.

Feng, Lili; Zhao, Anzhou; Liu, Weihua; Niu, Zhongen; Lv, Yan

doi: 10.1080/01431161.2025.2596810pmid: N/A

Water is an important limiting factor for sparse vegetation growth in northern China. The trade-off between sparse vegetation growth and water consumption highly related to the carbon and water cycle process is important for studying the water demand and optimal utilization of vegetation restoration in northern China. In this study, the sensitive band to sparse vegetation was selected to establish a new vegetation index by remote sensing and the time series similarity comparison method was used to obtain the Mean Absolute Distance (MAD) distribution to indicate the sparse vegetation growth. The Priestley-Taylor Jet Propulsion Laboratory (PT-JPL) model was driven by multiple sources data such as ground and remote sensing to simulate sparse vegetation transpiration. The results show that vegetation has increased from 2001 to 2015 in northern China. The upward trend in transpiration of sparse vegetation was aligned primarily with changes in sparse vegetation. However, there are localized areas in eastern Inner Mongolia and northwest Xinjiang where a decreasing trend was observed. The transpiration and water use are highly related to vegetation cover during the afforestation process. Optimal vegetation cover considering water use was obtained by analysing the relationship between sparse vegetation and transpiration, and the optimal vegetation cover value (MAD = 0.4) for water use was proposed. The results of the study can fill the gaps in sparse vegetation distribution and transpiration data in desertification areas in China, and provide theoretical support for the selection of optimal site conditions in the process of vegetation restoration to strengthen forest management.

Ritz, Alison L.; Wynne, Randolph H.; Thomas, Valerie A.; Wagner, Fabien H.; Green, P. Corey; Schroeder, Todd A.; Saatchi, Sassan

doi: 10.1080/01431161.2025.2598075pmid: N/A

Convolutional neural networks (CNNs), specifically U-Nets, successfully detect and segment individual tree crowns when applied to high spatial resolution remotely sensed data. However, much of the prior work for individual tree detection and segmentation has been in arid climates such as the western United States and the African Sahel. In this study, we used the U-Net-id CNN architecture and National Agriculture Imagery Program (NAIP) imagery to investigate forest-cover-specific training on individual tree crown detection in Virginia, U.S. We trained three models using hand-delineated crowns: one model with mixed/deciduous forests (3,457 trees), one with pine plantations (10,680 trees), and one that combined all the training data (14,137 trees) from both taxonomic groups. The training data was selected using 128 × 128 pixel patches across the study area where one interpreter hand-delineated all the tree crowns in the patch. Accuracy for the models was assessed over the entire model application area, by forest cover, and plantation age groups. Accuracy assessment included comparing the hand delineation and model results for tree counts and crown area as well as precision, recall, and F1 scores calculated. The Combined model performed poorly, achieving an F1 score of 0.4. However, further assessment of the forest cover and plantation age level showed increased F1 scores. The Plantation model’s F1 scores indicated the best-performing age class was the 16–19-year-olds, with a score of 0.74. However, the younger plantations had F1scores of 0.42, 0.28, and 0.07 for the 8–11, 4–7, and 0–3-year-olds, respectively. The best count and crown area relationships came from the Mixed/Deciduous model, which achieved an F1 score of 0.64. In this study, we found that the representative quality of the training data versus a larger quantity of data more heavily impacts this U-Net-id CNN architecture and demonstrates the impact of training data quality on a CNN model performance.

Slapikas, Ryan; Donnelly, Ryan C.; Nippert, Jesse B.; Pau, Stephanie

doi: 10.1080/01431161.2025.2593712pmid: N/A

Grasslands exhibit high taxonomic and functional diversity, particularly at fine spatial scales, posing challenges for remote sensing due to patchiness and species turnover. The spatial resolution of most remote sensing platforms often exceeds the size of homogeneous grassland patches, resulting in mixed pixels that hinder vegetation mapping. To address this, we applied Multiple Endmember Spectral Mixture Analysis (MESMA) to high-resolution (1 m2) hyperspectral imagery from the NEON Airborne Observatory Platform (AOP) to assess the predictive accuracies of fractional cover and dominance of four major grass evolutionary lineages, Andropogoneae, Panicoideae, Chloridoideae, and Pooideae, across four U.S. Great Plains grasslands. MESMA performance was evaluated using different endmember selection strategies, including leaf- vs. plot-level spectral endmembers and site-specific vs. multiple-site endmembers. Overall classification accuracy reached ~90% (Matthews Correlation Coefficient ~0.84) using optimal endmember combinations. While no single approach was universally superior, in general, leaf-level endmembers from focal sites and plot-level endmembers aggregated across all sites yielded higher overall accuracies. These results demonstrate that plot-level endmembers are more transferable across sites compared to leaf-level endmembers. Our results furthermore demonstrate that incorporating information about evolutionary relatedness can improve spectral unmixing results. This study advances sub-pixel mapping of grassland composition, offering insights for ecological modelling, land change prediction, and assessing grassland responses to environmental change and community composition.