Retrieving accurate 3D deformation fields from Interferometric Synthetic Aperture Radar)InSAR (line-of-sight (LOS) measurements is challenging because LOS data provide only one-dimensional motion, and the absence of ground control points (GCPs) in remote regions makes reliable 3D reconstruction even more difficult. This study introduces a deep learning-assisted approach to map 3D deformation fields without GCPs, integrating Pleiades stereo imagery with InSAR techniques over a 20 km x 20 km segment of the Denali Fault, Alaska. Initial 3D displacements from Pleiades, derived through DEM differencing and COSI-Corr, eliminate GCPs and reveal vertical displacements of +10 mm (uplift/subsidence), with horizontal displacements of +6.4 mm (eastwest) and +5.9 mm (north-south), consistent with the fault's right-lateral strike-slip kinematics and local GNSS IGS14/NNR velocities (2019-2024). These Pleiades displacements, combined with PS-InSAR rates and geological features like slope and Terrain Ruggedness Index (TRI), serve as inputs to a U-Net model that transforms LOS data into 3D fields, expanding displacement ranges to +20 mm across all components. We enhance U-Net estimates with geodetic optimization and use Monte Carlo Dropout (10 samples) to quantify uncertainties of 0.1-0.3 mm for eastwest and 0.5-0.8 mm for north-south and vertical displacements. Validation of the model against Pleiades test data yields RMSEs of 1.17 mm (east-west), 1.46 mm (north-south), and 1.52 mm (vertical), with an RMSE of 1.98 mm against the vertical component of three local GNSS stations (6-50 km distance, IGS14/NNR frame, 2019-2024). This InSAR-Pleiades-deep learning method offers a scalable solution for 3D deformation monitoring, advancing seismic hazard assessment in GCP-free environments.

Building extraction from high-resolution remote sensing imagery remains challenging despite advances in deep learning, particularly when labeled data are limited and models need to generalize across unseen scenes and geographic regions. Therefore, this research proposes a novel prior-conditioned segmentation framework that leverages external, independently generated probabilistic building maps as a reliability-aware guidance signal within an adaptive training strategy. The method consists of three main stages: (1) an external probabilistic building prior is learned independently and used as an additional source of spatial confidence; (2) a U-Net-style encoder–decoder is trained with a prior-conditioned adaptive gating module that embeds the prior into feature space and modulates shallow and mid-level encoder representations using a soft range-based gating function with learnable parameters, followed by channel–spatial attention to selectively enhance high-confidence building evidence while suppressing confounding background responses; (3) training optimization is improved by introducing a new confidence-aware loss function that implements adaptive supervision through dynamic pixel-wise reweighting based on the joint reliability of the external prior and the model’s predictive uncertainty, further complemented by region-consistency and boundary-sensitive terms. Experiments on three benchmarks (WHU, Massachusetts, and Inria) show that with only 30% of the labeled WHU training data, the proposed approach achieves 96.83% F1-score and 94.57% IoU. In cross-domain evaluations, it improves precision by reducing false positives by over 12% compared with a standard U-Net, while also delivering consistent gains over both CNN– and Transformer-based baselines. These results suggest that the proposed framework is well suited for practical building mapping scenarios where annotations are scarce and domain shifts are unavoidable.

The rapid expansion of mussel raft aquaculture has increased the demand for accurate and scalable monitoring systems to support sustainable coastal management. Deep learning (DL) object detectors, particularly the YOLO family, enable automated extraction of aquaculture infrastructures from Very High-Resolution (VHR) satellite imagery. However, conventional detectors rely on horizontal bounding boxes that often include excessive background when objects are elongated or rotated, reducing localization accuracy. Oriented bounding box (OBB) detectors address this limitation by modeling object orientation and geometry. In this study, an open-source Python-based graphical user interface (GUI) was developed for automated mussel raft detection, and a comprehensive comparison of YOLOv8-OBB and YOLO11-OBB architectures was conducted using VHR imagery from Spain’s Ría de Arousa. A dedicated mussel raft dataset was compiled and annotated, and both model families were evaluated across five architectural scales (n, s, m, l, and x). Experimental results showed that the YOLO11 family generally outperformed YOLOv8 in terms of detection accuracy, localization quality, and performance under challenging maritime conditions. Among the evaluated models, YOLO11s achieved the highest F1-score (0.866) and mAP50–90 (85.21%), while YOLO11n achieved the highest Mean IoU (0.872) and mAP50 (96.87%). The results further demonstrated that compact architectures can provide a favorable balance between accuracy and computational efficiency, as larger models did not consistently yield superior performance. Large-scale experiments on a 60,309 × 61,569-pixel (50 cm resolution) VHR scene demonstrated the operational scalability of the proposed framework through a patch-based processing strategy. The comparative evaluation conducted in this study provides valuable insights into recent YOLO-OBB architectures and highlights their potential for scalable AI-based aquaculture monitoring.

Accurate land cover classification maps are essential for effective crisis management and metropolitan planning. Remote sensing data, particularly optical and radar imagery, offer a costeffective and time-efficient means of producing such maps. While each data type has its own strengths and limitations, their integration can significantly improve classification accuracy. This study introduces a novel dual-level fusion approach that combines optical and radar imagery at both the feature and knowledge levels. First, feature-level fusion is applied to generate integrated inputs for the classification model. Next, feature-knowledge fusion is employed during trainingsample selection, reducing reliance on manual labeling and improving classification robustness. The proposed method was evaluated in two geographically distinct urban areas, Tehran and Tabriz. The results show that the approach achieves an overall accuracy of 94.7% and a Kappa coefficient of 0.93. These findings indicate that dual-level data fusion can substantially enhance classification performance, providing a practical and reliable solution for land cover mapping in complex urban environments.

The use of Unmanned Aerial Vehicles (UAVs) is rapidly increasing, making accurate localization essential for safe and reliable navigation. However, GPS signals can become unreliable or entirely unavailable in urban canyons, mountainous terrain, or environments with intentional interference. While many vision-based navigation methods rely on optical imagery, the potential of UAV-borne Synthetic Aperture Radar (UAVSAR) which performs reliably in all-weather and day–night conditions has received relatively little attention.This study presents a new three-stage geo-localization framework specifically designed for UAVSAR imagery: (1) coarse region retrieval by matching each UAVSAR image against low-resolution satellite tiles, (2) fine image alignment within the selected region using high-resolution reference imagery, and(3) final localization based on affine transformation estimation with compensation for the side-looking geometry of SAR. The feature extraction and matching process is driven by a Convolutional Multi-Scale Network with a ResNet-50 backbone, trained to learn robust, invariant features from image pairs captured under varying illumination, viewpoint, and sensor modality. Keypoints are extracted from intermediate feature maps, and mismatches are removed using adaptive distance filtering and RANSAC-based geometric verification. Experiments on a UAVSAR dataset covering more than 46,000 km² and 12 diverse flight paths demonstrate that the proposed method achieves RMSE values between 1.83 and 2.86 m (average 2.23 m) and yields a high Recall@1 in the coarse-retrieval stage.Compared with existing cross-modality matching approaches, the proposed framework consistently achieves higher retrieval accuracy and demonstrates more reliable operational performance.Training and testing across geographically distinct regions further confirm the strong generalization capability of the method.Overall, the results show that SAR-based visual navigation is a practical and robust alternative for GPS-limited or GPS-denied environments, offering a structured multi-stage pipeline and a cross-modality matching strategy that clearly distinguish it from prior work.

Flooding is a major global concern due to its severe impacts on lives, agriculture, and infrastructure, necessitating effective flood management. This research aims to develop a framework for automated, continuous detection and monitoring of floods using Sentinel-1 and Sentinel-2 imagery. Despite recent advances in Remote Sensing, existing methods struggle to distinguish floodplains from features like steeply inclined areas, shadows, moisture, noise, and errors in large areas. To address this problem, a Score-Based Flood Detection (SBFD) index was developed for time-series flood monitoring. This index uses VV and VH polarizations from Ascending and Descending orbits of Sentinel-1 Synthetic Aperture Radar (SAR) images, showing high potential for accurate and continuous flood monitoring. Additionally, the thresholding process was automated using Otsu and unimodal thresholding methods. This study utilizes the average of the Modified Normalized Difference Water Index (MNDWI) and the Flood Water Extraction Index, derived from Sentinel-2 images, to estimate pre-flood water bodies. A decision tree method was used to identify and classify flood extent. Unlike other supervised methods, the proposed framework automatically detects flood areas without requiring training samples. By using images from both Ascending and Descending orbits, the method mitigates errors caused by shadows, steep slopes, and moisture. The advantages of the proposed method include eliminating the need for training samples, automatic thresholding, full utilization of Sentinel-1 SAR image potential, and reduce processing time through the use of Google Earth Engine (GEE). Validation results demonstrated the robustness of the SAR images and the proposed framework in detecting and monitoring floods across diverse regions, and was confirmed by Copernicus Emergency Management Service data. Comparisons with other methods in Ascending and Descending orbits showed that the proposed framework significantly improved flood detection accuracy, particularly in complex terrains. Based on the results, the SBFD index achieved an average Overall Accuracy of 92.32

This research introduces an innovative approach to flood vulnerability reduction by integrating the Particle Swarm Optimization (PSO) algorithm with the Random Forest (RF) model to optimize hyperparameters through a parallel and simultaneous search. This methodology aims to accurately identify flood-prone areas. The study also evaluates the performance of the proposed model against other machine learning models, such as the Alternative Decision Tree (ADTree) and Multilayer Perceptron (MLP). Two datasets were utilized for model analysis: ground-based data, including rainfall, proximity to rivers, and roads, and remote sensing data, including elevation, slope, and land use. The Ottawa-Gatineau region in Canada was chosen for modeling. When both ground and remote sensing data were combined, the RF-PSO model achieved a Kappa coefficient of 0.74, outperforming the ADTree (0.70) and MLP (0.69) models. The study further explored the use of remote sensing data alone, with the RF-PSO model yielding a Kappa coefficient of 0.68, suggesting that even without ground-based data, remote sensing alone can produce reliable results. Notably, when high-resolution remote sensing data was applied, the Kappa coefficient increased to 0.80, demonstrating that improved spatial resolution reduces the dependence on ground-based data, thus enhancing model accuracy. This research highlights the potential of using high-resolution satellite data for flood risk assessment, offering significant insights into crisis management and flood vulnerability reduction.

Accurate surface water detection and mapping using Remote Sensing (RS) imagery is crucial for effective water and flood management and for supporting natural ecosystems and human development. In recent years, RS technology and satellite image processing have significantly advanced in flood and permanent water extraction, particularly in water index, clustering, classification, and sub-pixel analysis. Water-index-based techniques, distinguished by their quickness and convenience, offer notable advantages. The dynamic and extensive nature of surface water and flooded areas make the water index particularly effective for monitoring large areas. However, challenges arise due to the complexity of ground surfaces in aquatic environments, including shadows in builtup, vegetated, and mountainous regions, narrow water bodies, and muddy water. This research presents a new Flood Mapping Index using Sentinel-2 imagery (SFMI) designed to address these challenges and identify water and flooded areas more accurately. The SFMI utilizes visible and near-infrared bands derived from Sentinel-2 data, employing 10-m bands to compensate for errors arising from spectral and spatial changes more effectively. The SFMI index is designed based on the spectral signatures of various land cover classes, utilizing the potential of 10-m resolution bands to identify water bodies and flood areas. Unlike the most conventional methods, the SFMI identifies and extracts water and flood regions without complex thresholding, and thus mitigates the impact of irrelevant features, such as dense vegetation and rugged topography on the flood and water body maps. The proposed index was tested in two large areas with high spectral diversity, yielding promising results. The SFMI index demonstrates an average overall accuracy of 97.1% for pre-flood water extraction, 97.95% for post-flood water extraction, and 98% for flooded area extraction. Moreover, the results showed an average kappa coefficient of 0.958 for pre-flood water extraction, 0.965 for post-flood water extraction, and 0.978 for flooded area extraction. The performance of the SFMI index for extracting flooded areas (Delta SFMI) is superior to its performance for water extraction both before and after the flood. However, it is essential to note that the accuracy of the flooded area map is contingent on the accuracy of the water area map both before and after the flood. Thus, the SFMI index based on 10-m Sentinel-2 imagery accurately detects floods and water bodies over time, without relying on thresholding, making it suitable for flood management and monitoring various water bodies like dams, lakes, wetlands, and rivers. The findings underscore the applicability of the proposed SFMI index in diverse and spectrally rich areas, demonstrating its effectiveness in monitoring various surface water bodies, detecting floods, and managing flood crises.

Urbanization and changes in land use are crucial for national development. Managing and monitoring these changes have become easier and more accurate due to advancements in imaging systems and change detection methods. In recent years, convolutional neural networks have gained significant attention and advancement in remote sensing change detection research. However, designing a model that reduces computational complexity while improving detection accuracy is still under discussion. In this regard, an efficient model based on pseudo-siamese convolutional networks had been developed in this research, which by combining multi-scale features at different levels, increased the accuracy of the final change map, especially in images with both different spatial resolutions and scales. Our model is inspired by Scale Invariant Feature Transform (SIFT), which uses multiple scale-spaces to identify scale-independent keypoints in images. We designed our model based on this concept, in which After passing the bi-temporal images through the encoder network and making the initial feature maps, the feature learning process takes place at several levels of different scales and finally, the change map is calculated. In addition, the proposed architecture leverages point-wise, depth-wise, and neighboring features across various scale levels to harness the optimal features present in high-resolution remote sensing images. Simplicity in design, effectiveness in calculations, and accuracy make our model surpass popular methods such as BIT, DSIFN, etc. The evaluation results on 4 different datasets and comparison with 12 state-of-the-art models demonstrate that the proposed method outperforms other popular models in change detection. Available at: https://github.com/farhadinima75/OctaveNet .

This paper presents a comprehensive study on the recognition of crustal deformation patterns surrounding the North Tabriz Fault in Northwestern Iran, utilizing Multi-Temporal InSAR analysis. The fault, despite its seismic inactivity for over two centuries, has a long history of ancient seismicity, with earthquake recurrence intervals exceeding two centuries. This makes it highly susceptible to future activity and the generation of significant and devastating earthquakes. However, limited research has been conducted on extracting and modeling deformation patterns of the North Tabriz Fault to identify its active segments. The primary objective of this study is to derive a general trend for fault displacement and investigate regions under pressure in terms of abnormal crustal movements. The results indicate that the Earth's crust in the surrounding regions of the central and northwest segments of the fault exhibits an upward movement ranging from approximately 2 to 10 millimeters per year from 2015 to 2022. In contrast, neighboring areas of the northwestern fault, as well as the northwestern, western, and southwestern parts of Tabriz County, experience ground subsidence with rates ranging from approximately 5 to 40 millimeters per year. These findings are consistent with GNSS-derived line-of-sight measurements obtained from some IPGN stations around the fault with an RMSE of 1.72 mm/yr. Furthermore, the study identifies critical points near the fault that exhibit varying and diverse displacement patterns over time, suggesting significant strain and notable stress within the subsurface environment. According to the analysis of time series data on crustal movements at the identified critical points, it has been found that the prevailing motion pattern of the Earth's crust within the fault zone largely conforms to a sinusoidal descending pattern. Additionally, recent earthquakes in the northwest vicinity of the fault have been observed to occur close to these critical points. Using line-of-sight (LOS) data acquired at these critical points, the study estimates a slip rate of 7.71 +/- 0.01 mm/year and a locking depth of 11.27 +/- 0.01 km, contributing to a better understanding of the fault's seismogenic behavior. These findings provide valuable insights into the crustal deformation patterns around the North Tabriz Fault, highlighting active segments and regions under pressure.

Flooding is one of the most severe natural hazards, causing widespread environmental, economic, and social disruption. If not managed properly, it can lead to human losses, property damage, and the destruction of livelihoods. The ability to rapidly assess such damages is crucial for emergency management. Near Real-Time (NRT) spatial information on flood-affected areas, obtained via remote sensing, is essential for disaster response, relief, urban and industrial reconstruction, insurance services, and damage assessment. Numerous flood mapping methods have been proposed, each with distinct strengths and limitations. Among the most widely used are machine learning algorithms and spectral indices, though these methods often face challenges, particularly in threshold selection for spectral indices and the sampling process for supervised classification. This study aims to develop an NRT flood mapping approach using supervised classification based on spectral features. The method automatically generates training samples through masks derived from spectral indices. More specifically, this study uses FWEI, NDVI, NDBI, and BSI indices to extract training samples for water/flood, vegetation, built-up areas, and soil, respectively. The Otsu thresholding technique is applied to create the spectral masks. Land cover classification is then performed using the Random Forest algorithm with the automatically generated training samples. The final flood map is obtained by subtracting the pre-flood water class from the post-flood image. The proposed method is implemented using optical satellite images from Sentinel-2, Landsat-8, and Landsat-9. The proposed method’s accuracy is rigorously evaluated and compared with those obtained from spectral indices and machine learning techniques. The suggested approach achieves the highest overall accuracy (OA) of 90.57% and a Kappa Coefficient (KC) of 0.89, surpassing SVM (OA: 90.04%, KC: 0.88), Decision Trees (OA: 88.64%, KC: 0.87), and spectral indices like AWEI (OA: 84.12%, KC: 0.82), FWEI (OA: 88.23%, KC: 0.86), NDWI (OA: 85.78%, KC: 0.84), and MNDWI (OA: 87.67%, KC: 0.85). These results underscore the superior accuracy and effectiveness of the proposed approach for NRT flood detection and monitoring using multi-sensor optical imagery.

Accurate assessment of surface water from satellite and remote sensing data plays an important role in water and flood management and supporting natural ecosystems and human development. Remote sensing imagery has significantly advanced in water extraction methods, particularly in water index, classification, and sub-pixel analysis. Water-index-based approaches offer notable advantages such as speed and convenience among these methods. The unique characteristics of surface water and flooded areas, including their extensive coverage and dynamic nature, make the water index particularly effective for monitoring large regions. However, the complexity of land surfaces in aquatic environments presents challenges that hinder accurate water extraction. These challenges differ across various factors, such as shadows in urban and mountainous areas, small water bodies, muddy water, and water leakage in unshaded regions. The current study introduces a novel Flood/Water Extraction Index (FWEI) for identifying water and flooded areas to address these challenges. The FWEI utilizes the average ratio of visible and near-infrared bands derived from Sentinel-2 images. The proposed index utilizes images with 10-m and average visible bands and more effectively compensates for errors arising from spectral and spatial changes. Therefore, it demonstrates strong performance by more accurately mapping muddy and clear water within small water bodies and narrow rivers. The performance of the offered FWEI index is compared with other indices, including the Normalized Difference Water Indices (NDWI-G, NDWI-F), Modified NDWI (MNDWI-1, MNDWI-2), and the Automatic Water Extraction Index (AWEI nsh ) without shadow. While other indices excel in specific scenarios, such as built-up or non-built-up areas, and bare lands versus vegetated areas, the FWEI index demonstrates consistently high accuracy and stability in extracting surface water across diverse backgrounds. The FWEI index achieves an average Overall Accuracy (OA) of 94.26% for water extraction and 93.11% for flood extraction. In comparison, the AWEInsh attains an OA of 90.48% and 90.39%, NDWI-F performs at 86.69% and 86.55%, MNDWI-1 at 77.21% and 75.82%, MNDWI-2 at 76.12% and 75.42%, and NDWI-G at 75.26% and 74.78%, respectively. The integration of visible spectral bands with the near-infrared band proves instrumental in enhancing the accuracy of water derivation in complex and expansive environments.

The main factors contributing to the occurrence of an earthquake are under the crust. Also; due to the lack of access to direct measurements of these factors and the parameters involved in the occurrence of an earthquake, the main goal of researchers is to study the earthquake occurrence through its precursors. Currently, monitoring and identifying some of these precursors are made possible by geomatics technologies. It is an undeniable fact that the behavioral variations of the precursors don’t follow a common pattern in all earthquakes. Also, the variations of the precursors show peculiar behaviors in each region. So, it seems infeasible to provide an accurate prediction based on the analysis of the behavioral variations of a single precursor. Unlike previous studies, this study doesn’t have a single-parametrical orientation toward an earthquake prediction process. Accordingly, this study aims to extract the trend of variations in crustal deformation anomalies and thermal anomalies before the earthquake to analyze them through an integrated process based on data mining methods. As a result, the tests of earthquake predictions for 17 cases have shown that the proposed method can make a reliable prediction of the probable time and magnitude range of oblique-thrust earthquakes with a magnitude greater than 5.5. Moreover, the proposed method has been able to accurately estimate the occurrence of the 26th November 2019 Albania earthquake (Mw = 6.4) as well as 21th September 2019 Albania earthquake (Mw = 5.6) before they happen.

3D flight planning for UAV-based photogrammetry in urban areas with an emphasis on solving problems caused by extreme scale differences and occlusion points

Topographic mapping in mountainous areas encounters many challenges due to the potential impasse and lack of access to all locations. Unmanned Aerial Vehicles (UAVs) are an effective alternative to traditional field mapping in different environmental conditions. However, problems, such as large-scale differences, gaps, and errors due to extreme elevation differences in these areas, hinder the use of UAV-based photogrammetry, thus reducing the quality and accuracy of the photogrammetric products and the final extracted map in mountainous areas. By designing an optimal flight network before UAV acquisition, the effect of these problems can be reduced. This paper proposes a method for planning the dynamic three-dimensional imaging network UAV in mountainous terrain based on digital elevation model (DEM) to ensure the uniformity of the scale in the photogrammetric blocks, and avoid collision with obstacles, also gaps or data redundancy. The proposed method was implemented in a semi-mountainous area and was compared to two approaches of 3D network with static overlap and normal 2D flight plan. The results showed that the large-scale changes among the images were reduced and the ground sample distance (GSD) was maintained as constant as possible. Also, planning UAV flight program based on the proposed algorithm increases the accuracy of the photogrammetric products.

In the last two decades, among various Synthetic Aperture RADAR (SAR) imaging modes, Compact Polarimetry (CP) mode has drawn a lot of attention due to less complex imaging system, mass and data rate reduction, and also greater swath width.Having such advantages makes this data very useful for large-scale target mapping, such as forest classification.Different methods have been proposed for forest classification using CP mode, all of which are based on feature extraction.The accuracy of these methods depends on the discrimination of the extracted features.Among these methods, deep learning networks have almost automated the feature extraction phase and obtained impressive results, especially in the classification task.In this paper, the ability of deep learning networks is investigated by using CP mode data in forest classification.The study area of this paper is Petawawa forest located in Ontario, Canada, and the data being used are simulated CP data, Full Polarimetric (FP) data, and also reconstructed Pseudo Quad (PQ) data acquired from RADARSAT-2 in C-band.The proper deep learning network for automatic feature extraction is designed and the classification is performed on CP, FP, and PQ data.The results from all mode classifications are evaluated and compared with each other and also with the results from Wishart classifier and Support Vector Machine (SVM).The results of this paper show that using deep learning networks improves the classification accuracy of CP and PQ modes.

Monitoring the spatiotemporal dynamics of building footprints (BF) is necessary for understanding urbanization growth. It is a difficult task to extract residential sites, mainly BF, because of the complexity of their makeup and spectral variety. Additionally, conventional methods for building mapping typically rely on abundant training data and expertise from human operates. This study presents a new unsupervised Feature-Based Building Footprint Extraction (F2BFE) strategy using Sentinel-1 & 2 satellite images and the SRTM Digital Elevation Model (DEM). The newly developed radar index (NRI) from Sentinel-1 images was utilized to extract the Primary Building Footprints (PBF) through histogram analysis and thresholding techniques, based on the mean of annual Sentinel-1 VV and VH Backscatter channels in the Ascending orbit. In this research, the integration of the Otsu and Unimodal thresholding technique was developed as an optimal thresholding method for feature extraction. Furthermore, Sentinel-2 images were applied to extract spectral indices related to vegetation (NDVI, GNDVI, RDVI indices), water (NDWI index), and residential/built-up (NDBI, BuEI). The qualitative and quantitative validation results indicate that the NRI-based BF map achieved higher Overall Accuracy (OA) values of 98.14%, 90%, and 91% in Region of Interest-1 (ROI-1), ROI-2, and ROI-3, respectively. Additionally, the Kappa Coefficients (KC) for these regions were 0.96, 0.97, and 0.85, respectively. The NRI index provides an excellent OA result when vegetation, water, and slope features are carefully eliminated. Finally, it can be inferred that the simultaneous use of the sentinel-1 & 2 and slope data in feature space leads to increased BF accuracy.

Forest fires are natural events that occur in numerous ecosystems worldwide and cause significant damage to human, ecological and socio-economic factors. It is also crucial to obtain useful information on the distribution and density of burned areas on large scale. An efficient way to map large regions is through remote sensing (RS). Nevertheless, the complex scenario and similar spectral signature of features in multispectral bands can lead to many false positives, making it difficult to extract the burned areas accurately. Multispectral data from Sentinel-2 satellite images allow the development of novel burned area indices, as more spectral data is recorded in the Red-Edge region. This research aims to develop a new burned area detection index (BADI) at 20 m spatial resolution in the google earth engine platform to detect the wildfire-affected areas in southwest of Iran using Sentinel-2 satellite imagery. The BADI spectral index has been specially designed to take benefit of the Sentinel-2 spectral bands and use a spectral combination of bands that are reasonable for post-fire burned regions detection. The final results indicated that the proposed index by applying a post-processing stage works well in the case of the study area to identify the burned areas. At the same time, it can satisfactorily suppress the complicated and irrelevant changes in the scene. Furthermore, the BADI index is rapid and can provide the burned areas map in near real-time. According to the Copernicus Emergency Management Service (EMS) reference data, maps of the burned areas were produced with a kappa coefficient of 0.92 and an overall accuracy of 92.15%, which demonstrated a good result in comparison to similar spectral indices.

Image matching is a branch of computer vision that forms the basis of most photogrammetry and remote sensing data processes. Outstanding results from recent activities in this field have shown the effectiveness of trainable descriptors. Currently, the existing datasets used to train these types of descriptors are collected based on front-view photos, and because of extractable features in these images, learned descriptors face challenges in the vertical image matching process. Also, a training dataset derived from top-view images has not been developed for the learned descriptors. To overcome these limitations, a training dataset based on remote sensing images is presented. Training patches in this collection are extracted from 61,223 images with 130 different scenes and different kind of platforms including fixed and multi-rotor unmanned aerial vehicles (UAV). To evaluate the performance of the dataset, mean average precision (mAP) criterion, ratio of descriptor distances, and point cloud density were considered. The effectiveness of the developed dataset was proved with 15 https://github.com/farhadinima75/UAVPatches .