https://eoge.ut.ac.ir/ Earth Observation and Geomatics Engineering en daily 1 Tue, 01 Dec 2026 00:00:00 +0330 Tue, 01 Dec 2026 00:00:00 +0330 https://eoge.ut.ac.ir/article_106432.html Gaseous pollutants present growing challenges in major urban centers, posing serious threats to public health and the environment. In cities such as Tehran, where pollution levels are critically high, nitrogen dioxide (NO₂) and carbon monoxide (CO) play significant in environmental harm and health concerns. Conventional prediction models frequently depend on single-source datasets, overlooking the advantages that multi-source Remote Sensing (RS) data can offer in improving the accuracy of forecasts. To address this issue, We propose a novel approach using Long Short-Term Memory (LSTM) neural networks to estimate gaseous pollutant concentrations by incorporating multi-source RS data. This approach utilizes input from MODIS sensors (surface temperatures, aerosol optical depth, thermal emissivity) along with ground-based particulate matter data (PM₁₀, PM₂.₅). Using a five-year dataset (January 2018–January 2024) from Tehran's air quality stations, we allocated 70% of the data for training and 30% for testing and validation. Our results reveal that the LSTM-based model outperformed existing methods like Gated Recurrent Unit (GRU), Random Forest (RF), One-Dimensional Convolutional Neural Network (1D-CNN) in predicting pollutant concentrations, with greater accuracy and reliability for forecasting. For CO, the R² values were 0.531 (GRU), 0.390 (1D-CNN), 0.498 (RF), 0.522 (Transformer), and 0.566 (LSTM); for NO₂, they were 0.953, 0.952, 0.907, 0.864 and 0.957, respectively. These results validate the superior performance of LSTM in capturing complex non-linear relationships in multi-source RS data for our case study. The improved model performance is beneficial for urban planning, pollution mitigation strategies, and public health protection. https://eoge.ut.ac.ir/article_106433.html Mud volcanoes are dynamic surface expressions of deep over pressured fluids and gases, requiring precise monitoring to understand their behavior and related hazards. This study investigates ground deformation at the Tang Mud Volcano in southeastern Iran using the Small Baseline Subset (SBAS) InSAR technique applied to 16 Sentinel-1 IW SLC images acquired from December 31, 2023, to December 13, 2024.Data were processed using the GMTSAR toolbox under a Linux environment, supported by SNAP and GMT. Both Ascending and Descending tracks were analysed to validate deformation signals across viewing geometries. Interferometric analysis included co-registration, baseline estimation, interferogram generation, Goldstein filtering, phase unwrapping, and time-series inversion using Singular Value Decomposition (SVD). Coherence and standard deviation maps ensured data quality.Results show minimal displacement at the main vent (P1), with cumulative deformation limited to approximately +5 mm, indicating a stable structure. In contrast, a nearby flank area (P2) experienced significant uplift (~+47 mm), suggesting active subsurface processes such as fluid migration or sediment extrusion. Despite the relatively stable behavior observed at the Tang mud volcano itself, the adjacent areas demonstrated significant uplift of up to 6 cm. This level of deformation, occurring over a one-year observation period, highlights the geodynamic activity in parts of the Kahir region. Velocity maps revealed localized deformation rates in mm/year, correlating with fault structures and internal pressure variations.Time-series analysis highlighted contrasting deformation trends at P1 and P2. The integrated interpretation confirms the effectiveness of SBAS-InSAR in detecting subtle, spatially variable deformation and enhances understanding of the volcano’s internal dynamics. These findings provide valuable insight for geohazard assessment and long-term monitoring of mud volcanoes in tectonically active regions. https://eoge.ut.ac.ir/article_106437.html Synthetic Aperture Radar (SAR) imagery, due to its capability for imaging under diverse weather and illumination conditions, has become one of the key tools for environmental monitoring, change detection, and both military and civilian applications. However, the presence of speckle noise and statistical heterogeneity of the background poses significant challenges to the anomaly detection process. In this study, a simple and computationally efficient approach based on noise reduction using a median filter, adaptive thresholding, and anomaly map generation is proposed. The method calculates statistical indices (mean and standard deviation) and defines a sensitivity threshold to effectively distinguish anomalous regions from complex backgrounds. For evaluation, the TenGeoP-SARwv dataset, containing more than 37,000 Sentinel-1 SAR images, was utilized. Experiments conducted on five sample images showed that the mean intensity (μ) ranged from 13,763 to 25,487, the standard deviation (σ) from 2,329 to 3,621, and the adaptive threshold (τ) from 18,749 to 32,730. These values demonstrate the adaptability of the proposed method to diverse statistical conditions. The final results indicate that the percentage of detected anomalous regions across the images varied between 2.44% and 3.25%. Notably, Image 4, which exhibited the lowest mean intensity, produced the highest anomaly percentage (3.25%), reflecting the algorithm’s capability to reveal unusual patterns even in low-backscatter data. Conversely, higher-intensity images, such as Image 3, showed lower anomaly ratios, highlighting the robustness of the method varying conditions. Visual analysis of the anomaly maps and histogram plots confirmed the quantitative findings, showing that the proposed approach can accurately highlight specific regions without generating widespread false detections. These characteristics make it an efficient tool for applications such as environmental monitoring, oil spill detection, natural resource management, and surveillance systems. Furthermore, the proposed approach can serve as a foundation for developing more advanced machine learning and deep learning-based methods in the future. https://eoge.ut.ac.ir/article_106438.html Soil moisture is an important variable for many studies, such as crop yield estimation, drought monitoring, evapotranspiration, agricultural and water resources management. Today, remote sensing data are widely used to estimate soil moisture. The main purpose of this study is to estimate soil moisture using Sentinel 1 radar data. In the first step, 8 features were extracted from Sentinel 1 data (dual polarized radar backscatter). The ground probe was used for field measurement of soil moisture. In the next step, soil moisture retrieval was done using machine learning methods. These data were used to train and validate machine learning methods. In this research, support vector regression (SVR), decision tree, random forest (RF) and neural network (ANN) methods were performed to estimate soil moisture. Evaluation of the accuracy of different methods was done using field measurement and comparing it with the estimated moisture. The random forest method has the highest accuracy with, Root Mean Square (RMSE=0.03kg/kg) and correlation coefficient (R2=98.6%). The results of this research showed that the use of different features extracted from Sentinel 1 data along with a suitable machine learning method can significantly increase the accuracy of soil moisture retrieval. https://eoge.ut.ac.ir/article_106441.html Rooftop type classification plays a crucial role in numerous urban applications such as 3D city modelling, solar potential estimation, disaster management, and smart city development. Leveraging high-resolution UAV imagery and advances in machine learning and deep learning, this study conducts a comparative evaluation of traditional machine learning and deep learning algorithms for classifying rooftop types, gable, half-hip, and complex, using orthophotos from Rasht City, Iran. In order to achieve that, firstly a structured and annotated dataset of rooftops was constructed through manual digitization and image cropping, followed by extensive training data augmentation. In the second step, two handcrafted feature sets (HOG and HOG+LBP) were used to train five machine learning classifiers (SVM, Random Forest, KNN, XGBoost, and Logistic Regression). The best results among these was achieved by SVM (F1-score = 0.74), while KNN performed the weakest. Also, ensemble learning methods were utilized through the aggregation of diverse model predictions. Using ensemble learning techniques, particularly stacking, significantly enhanced classification accuracy, reaching an F1-score of 0.79. In the third step, deep learning models including a custom CNN and a MobileNetV2-based transfer learning approach were utilized. The latter achieved the highest overall classification accuracy (83.3%) and macro F1-score (83%), outperforming all traditional methods by learning spatial hierarchies directly from data. The results demonstrate that while classical methods benefit from interpretable feature design and faster training, deep learning showed better within-domain generalization on our Rasht dataset. https://eoge.ut.ac.ir/article_106443.html Objective: Medical imaging, particularly chest X-ray analysis, plays a vital role in diagnosing and treating lung diseases such as pneumonia, pulmonary fibrosis, and lung cancer. Despite its importance, accurate interpretation of these images faces several challenges, including low quality, noise, illumination variations, and the high cost and effort of manual annotation. Moreover, the complex anatomy of the lungs requires advanced algorithms to achieve precise delineation. Method: This study presents an automatic lung segmentation approach based on the U-Net convolutional neural network. With its encoder–decoder architecture, U-Net effectively integrates compressed and expanded feature representations to produce accurate segmentation. The model was trained on 563 annotated chest X-ray images and evaluated on 141 independent cases.Results: Experimental results demonstrate 91% accuracy on the training set and over 84% accuracy on the test set, confirming its strong performance in extracting lung regions. The findings reveal that U-Net can reliably detect lung structures and lesions, even with limited training data or varying image quality. This reduces dependence on manual interpretation, lowers associated costs, and minimizes human error while accelerating the diagnostic process. The model’s high generalizability further supports its potential for use across diverse clinical settings.Conclusions: In summary, this research emphasizes the value of deep learning architectures such as U-Net for precise medical image segmentation. The proposed method enhances diagnostic efficiency and accuracy, providing a reliable tool for supporting clinicians. Future extensions, including integration with complementary deep learning techniques, may further advance intelligent healthcare applications. https://eoge.ut.ac.ir/article_106444.html Objective: In recent decades, Iran has encountered significant challenges in land ownership, registration, and spatial management due to population growth and the expansion of land use. The coexistence of formal and informal registration systems, the diversity of cadastral documents, and the fragmentation of responsible institutions have emphasized the need for a unified and standardized data model. Method: This study qualitatively examines the Land Administration Domain Model (LADM) and its main packages and classes, analyzing their structure and relationships in relation to Iranian cadastral and legal frameworks. A preliminary feasibility analysis was conducted to evaluate the implementation of a national LADM profile for Iran, followed by outlining key implementation stages including profile scope, stakeholders’ involvement, and conceptual modeling.Results: For each class and its fields, the relevant laws were extracted. The relationships, their cardinality, and several code lists were reviewed and customized based on national regulations to ensure semantic alignment with Iran’s legal and administrative context.Conclusions: The findings highlight that the LADM framework can effectively support the modernization of Iran’s land administration system by enabling the integration of legal and spatial data within a harmonized structure, thereby enhancing transparency and promoting efficient land governance and combating corruption. https://eoge.ut.ac.ir/article_106445.html The employment of digital maps has become a common tool and a widespread resource and an integral aspect of today's digital lifestyle. These maps are primarily created not through official sources but rather with the involvement of the users themselves, with OpenStreetMap (OSM) being a prominent example. However, the trustworthiness of these maps has always been a significant issue. This study focuses on how accurate and complete OpenStreetMap (OSM) data is in the city of Zanjan by comparing it to high-resolution 1:2000 reference maps. Using methods like spatial similarity and weighted averages, this research discovered that while OSM data was generally reliable, there were noticeable differences depending on the part of the city. The results revealed that in historical and deteriorated regions, the accuracy of the data was approximately 10% lower, and their completeness was about 35% less than in other areas. This highlights a key issue: Volunteered Geographic Information (VGI) often depends on where and who contributes to mapping.Therefore, relying solely on volunteered data can lead to uneven quality across different urban zones. This study highlights the importance of taking local context into account when utilizing VGI and considers targeted efforts to offer users indicators of information quality as crucial. https://eoge.ut.ac.ir/article_106447.html Monitoring and analyzing land use and land cover (LULC) is crucial for understanding environmental transformation, urban planning, deforestation, and water resource management. The Urmia Lake basin-a critical water body in northwestern Iran- has vital LULC changes over the past two decades.The study aims to predict future LULC in the Urmia Lake basin and generate LULC maps using deep learning models applied to satellite imagery. by comparing the performance of diverse deep learning models, the study seeks to identify most accurate approach for modeling spatio-temporal LULC changes. To achieve this, four models are implemented: a Multilayer Perceptron (MLP) which captures complex nonlinear relationships in data, two-dimensional and three-dimensional Convolutional Neural Network (2D CNN and 3D CNN) for effective extraction of spatial features from satellite imagery, and a Recurrent Neural Network (RNN) to model temporal changes over time. The input data consist of 64 features derived from six optical Moderate Resolution Imaging Spectroradiometer (MODIS) bands, Land Surface Temperature (LST), vegetation and surface indices (Normalized Difference Vegetation Index (NDVI), Normalized Difference Snow Index (NDSI), and Normalized Difference Water Index (NDWI)), topographical layers (Digital Elevation Model (DEM), slope, and aspect), 50 ERA5-Land monthly climate variables, and precipitation data from the PERSIANN Climate Data Record (PERSIANN-CDR). The dataset spans 22 years and includes eight land cover classes as output labels. Among the evaluated models, the 3D CNN reached the highest performance with a test accuracy of 92.82%, mean Intersection over Union (IoU) of 0.6351, and the lowest number of mismatched pixels. Test accuracy of 3D CNN indicates that 92.82% of the pixels in the independent test dataset for the year 2022 were correctly classified into their respective land use/land cover categories. These results confirm its superior ability to capture complex spatio-temporal patterns for accurate LULC prediction in the Urmia Lake basin. https://eoge.ut.ac.ir/article_106448.html Objective: Tree decline is a serious multifaceted problem in Zagros forests of Iran. The prevalence of this complex phenomenon in the recent decades highlights the need for high-resolution geospatial monitoring approaches. While unmanned aerial vehicle (UAV)-based photogrammetry offers a flexible and cost-effective means of capturing forest health, conventional top-of-canopy imaging fails to sufficiently represent critical under canopy features, including stem morphology and lower crown structure that is commonly prone to early symptoms of tree decline. Method: We presented a dual-layer UAV photogrammetric framework that combines above- and below-canopy imagery to detect phenotypic decline in Quercus brantii using a novel 3d tree reconstruction method followed by a MobileNetV2 deep learning model to detect decline symptoms on stems. Using this detection, we computed the Phenotypic Decline Index (PDI) and Decline Acuteness Index (DAI) to describe decline severity and trends in continuous form.Results: The MobileNetV2 achieved overall classification accuracy of 96.3% (F1-score = 0.94) in distinguishing healthy and declined stems (n = 299). This performance, derived from a confusion matrix with 166 true positives, 9 false negatives, 11 false positives, and 133 true negatives, highlights the model's high reliability. Furthermore, the UAV-derived DAI correlated strongly with multi-year field-based decline trajectories (Pearson r = 0.718, Spearman ρ = 0.928) , confirming the method’s reliability.Conclusions: Decline severities during three years of field data collection followed by suggested consistent shift amongst levels of tree decline, while UAV-based phenotypic analysis was shown to enable capturing nuanced changes in tree vitality. By yielding high correlations, we provide a cost-effective and high-resolution workflow for phenotyping oak decline, which enables multi-scale analysis of structural and symptomatic indicators using RGB-only data. https://eoge.ut.ac.ir/article_106451.html Soil salinity is a major driver of land degradation, and its reliable mapping remains challenging under frequent surface conditions. To address these limitations, this study presents a polarimetric based soil salinity classification framework that integrates dual polarimetric signature with texture information. In this context, an ascending Sentinel-1 dual-polarization (VV/VH) SLC scene acquired on 8 March 2024 was analyzed over the coastal districts of Khulna, Satkhira, and Jessore in southwest Bangladesh, which was deliberately selected to be temporally consistent with the field campaign conducted between 1–9 March 2024. First, texture measures were extracted from VV and VH intensity images using gray level co-occurrence matrix statistics in order to capture salinity related spatial pattern variations associated with surface roughness and crust formation. Next, dual polarimetric signatures were generated by synthesizing polarization states through linear combinations of the available channels, and a set of signature shape descriptors (e.g., pedestal height, extrema-related slopes, skewness, kurtosis, peak intensity, and signature width) was derived to enrich the feature space and improving class separability. Subsequently, to mitigate redundancy among heterogeneous predictors and enhance generalization, a feature selection step was applied to retain the most informative features for salinity discrimination. Finally, soil salinity was classified into three levels (low, moderate, and high) using two non-linear classifiers, support vector machine and multilayer perceptron, and performance was evaluated on an independent test set using confusion matrix metrics. Overall, the best-performing configuration achieved an overall accuracy of 91.6% and a Kappa coefficient of 0.861, thus demonstrating strong agreement with reference labels and underscoring the complementary value of texture and synthesized polarimetric response for dual-pol Sentinel-1 salinity mapping in coastal regions.