Dr. Hossein Bagheri is a leading expert in Geospatial Artificial Intelligence (GeoAI), specializing in the development of advanced machine learning and deep learning frameworks. His research focuses on pioneering methodologies for fusing multi-sensor remote sensing data to create high-fidelity geospatial information. He has a proven track record of translating complex algorithmic innovations into practical, impactful solutions for large-scale urban and environmental analysis.
Research Interests:
Remote Sensing, Machine Learning, GeoAI, Environmental monitoring,
Dr. Hossein Bagheri is a leading expert in Geospatial Artificial Intelligence (GeoAI), specializing in the development of advanced machine learning and deep learning frameworks. His research focuses on pioneering methodologies for fusing multi-sensor remote sensing data to create high-fidelity geospatial information. He has a proven track record of translating complex algorithmic innovations into practical, impactful solutions for large-scale urban and environmental analysis.
Research Interests:
Remote Sensing, Machine Learning, GeoAI, Environmental monitoring,
- Ph.D., Ph.D., Dr.-Ing. , Technical University of Munich (TUM), [ - ]
Articles
Publication Date: 2026
Trees, Forests and People (26667193)24
Accurate estimation of aboveground biomass (AGB) is essential for sustainable forest management and climate change monitoring; however, conventional remote sensing approaches often rely on a single data type and provide limited interpretability. This study introduces a novel framework that systematically integrates multi-sensor remote sensing (Sentinel-1 SAR, Sentinel-2, and Landsat-8 optical data), process-based climate variables (Climate-FVS), and LiDAR-derived topographic indices into deep learning models. By coupling these fusion scenarios with explainable AI (XAI) analyses, the framework not only improves predictive accuracy but also reveals the relative contribution of ecological drivers such as precipitation, temperature, and short-wave infrared bands. Seven U-Net-based architectures, including U-Net3+, TransU-Net, and Attention U-Net, were trained and evaluated using RMSE, MAE, and R² metrics. Results showed that U-Net3+ achieved the highest performance under the complete fusion scenario, with an RMSE of 28.10 Mg/ha, an MAE of 17.49 Mg/ha, and an R² of 0.89. XAI analyses highlighted that SWIR1, SWIR2, and Red bands were the most influential predictors, while climatic variables significantly improved model generalization in topographically complex areas. The highest errors occurred at vegetation boundaries and steep terrain. These findings demonstrate that multi-source data fusion combined with interpretable deep learning provides a robust pathway for both accurate AGB estimation and a deeper understanding of its environmental drivers, directly supporting carbon accounting and sustainable forest management. © 2026 The Author(s)
Publication Date: 2025
Trees, Forests and People (26667193)19
This study investigates the multifaceted factors influencing wildfire risk in Iran, focusing on the interplay between climatic conditions and human activities. Utilizing advanced remote sensing, geospatial information system (GIS) processing techniques such as cloud computing, and machine learning algorithms, this research analyzed the impact of climatic parameters, topographic features, and human-related factors on wildfire susceptibility assessment and prediction in Iran. Multiple scenarios were developed for this purpose based on the data sampling strategy. The findings revealed that climatic elements such as soil moisture, temperature, and humidity significantly contribute to wildfire susceptibility, while human activities—particularly population density and proximity to powerlines—also played a crucial role. Furthermore, the seasonal impact of each parameter was separately assessed during warm and cold seasons. The results indicated that human-related factors, rather than climatic variables, had a more prominent influence during the seasonal analyses. This research provided new insights into wildfire dynamics in Iran by generating high-resolution wildfire susceptibility maps using advanced machine learning classifiers. The generated maps identified high-risk areas, particularly in the central Zagros region, the northeastern Hyrcanian Forest, and the northern Arasbaran forest, highlighting the urgent need for effective fire management strategies. © 2025 The Authors
Publication Date: 2025
Heliyon (24058440)11(1)
This study focuses on generating high-resolution annual solar energy potential maps (ASMs) using global Digital Elevation Models (DEMs) to aid in solar panel placement, especially in urban areas. A framework was developed to enhance the resolution of these maps. Initially, the accuracy of ASMs derived from various DEMs was compared with LiDAR-derived ASMs. The evaluations indicated that the Copernicus DEM provided a highly accurate ASM. Subsequently, deep learning algorithms were trained to improve the resolution of the LiDAR-derived ASM. The results demonstrated that the Enhanced Deep Super-Resolution (EDSR) Network outperformed the U-Net-based model. The trained EDSR model was then utilized to enhance the resolution of the Copernicus ASM. Comparing the enhanced-resolution map of Copernicus respective to LiDAR showed that the EDSR model provided the necessary generalizability to improve the accuracy and resolution of the Copernicus ASM, particularly in urban areas. The investigations revealed that the improved resolution map with a resolution of 6 m, achieving RMSE of 35.75 [Formula presented] and a correlation of 0.87 respective to LiDAR data, was capable of locating solar panels on buildings, whereas the original Copernicus-derived maps with a 30 m resolution had RMSE of 51.26 [Formula presented] and a correlation of 0.72 for such placement purposes. © 2024 The Authors
Publication Date: 2025
Remote Sensing Applications: Society and Environment (23529385)38
Three-dimensional reconstruction of buildings, particularly at Level of Detail 1 (LOD1), plays a crucial role in various applications such as urban planning, urban environmental studies, and designing optimized transportation networks. This study focuses on assessing the potential of LiDAR data for accurate 3D building reconstruction at LOD1 and extracting morphological features from these models. Four deep semantic segmentation models — U-Net, Attention U-Net, U-Net3+, and DeepLabV3+ — were used, applying transfer learning to extract building footprints from LiDAR data. The results showed that U-Net3+ and Attention U-Net outperformed the others, achieving IoU scores of 0.833 and 0.814, respectively. Various statistical measures, including maximum, range, mode, median, and the 90th percentile, were used to estimate building heights, resulting in the generation of 3D models at LOD1. As the main contribution of the research, the impact of segmentation accuracy on the quality of 3D building modeling and the accuracy of morphological features like building area and external wall surface area was investigated. The results showed that the accuracy of building identification (segmentation performance) significantly affects the 3D model quality and the estimation of morphological features, depending on the height calculation method. Overall, the UNet3+ method, utilizing the 90th percentile and median measures, leads to accurate height estimation of buildings and the extraction of morphological features. © 2025 Elsevier B.V.
