دانشگاه اصفهان جغرافیا و برنامه ریزی محیطی 2008-5362 37 2 2026 06 22 Assessment and Evaluation of Flood Risk Factors (Case Study: Dashtiari Watershed) سنجش و ارزیابی عوامل تأثیرگذار در خطر سیلاب (مطالعه موردی: حوضه دشتیاری) 1 28 30496 10.22108/gep.2026.144627.1714 FA صمد فتوحی دانشیار گروه جغرافیای طبیعی، دانشکده جغرافیا و برنامه‌ریزی محیطی، دانشگاه سیستان و بلوچستان، زاهدان، ایران لیلی لطیفی دانشجوی دکتری ژئومورفولوژی، دانشکده جغرافیا و برنامه‌ریزی محیطی، دانشگاه سیستان و بلوچستان، زاهدان، ایران مسعود سعیدی دانشجوی دکتری ژئومورفولوژی، دانشکده جغرافیا و برنامه‌ریزی محیطی، دانشگاه سیستان و بلوچستان، زاهدان، ایران Journal Article 2025 08 26 Abstract Given the increasing frequency of floods in many watersheds, there is a growing need to examine the key factors contributing to flood risk in these areas. Assessing these factors is essential for evaluating vulnerability and exposure and provides critical information for effective flood management. In this context, the present study aimed to assess and evaluate the factors influencing flood risk with the Dashtiari Watershed selected as the case study. This research was descriptive-analytical in nature and applied in purpose. Data were collected through both library research and field surveys. Flood hazard zoning and risk assessment were carried out by using 13 parameters—including slope, elevation, and soil texture—and analyzed with ArcGIS software, the IGR index, and the Random Forest (RF) algorithm. The results indicated that approximately 5.40% of the study area fell into the very high-risk, 33.63% high-risk, 37.94% moderate-risk, 18.70% low-risk, and 4.31% very low-risk categories. Notably, about 50% of residential areas were situated within high-risk zones, which, in the event of flooding, could lead to severe human casualties and substantial economic losses. Keywords: Flood, Natural hazards, Influencing factors, Dashtiari Basin. Introduction Natural hazards, particularly floods, present serious challenges to communities around the world. These events often result in the destruction of infrastructure, social and economic disruption, and widespread environmental degradation. The occurrence and severity of floods are governed by a combination of geomorphological, hydrological, and atmospheric factors. Key determinants include land slope, soil type, land use, and rock permeability, all of which influence the magnitude and intensity of flooding. In Iran, certain regions are inherently more susceptible to flooding due to their geographic and geological characteristics. One such area is the Dashtiari watershed located in the southern part of Sistan and Baluchestan Province. This vast floodplain, which includes the Kajoo River and several other flood pathways, has historically experienced alternating periods of severe drought and extreme rainfall. Over time, local communities have developed adaptive strategies to mitigate the impacts of such hydrological extremes. Hydromorphometric analyses have shown that examining the physiographic attributes of drainage basins can provide valuable insights into their hydrological behavior. This understanding is critical for assessing flash flood risks and designing appropriate control measures. In this context, flood hazard zoning maps serve as essential tools for sustainable land-use planning and enhancing community resilience. Furthermore, hydraulic models play a key role in evaluating flow distribution patterns across floodplain areas, thereby supporting the development of effective flood mitigation strategies. Ultimately, sound water resource management and comprehensive hazard mitigation planning are vital for reducing both human casualties and economic losses. Implementation of early warning systems, construction of flood-resilient infrastructure, and timely evacuation of high-risk zones rank among the most effective responses to flood crises. Special attention to the calibration of relevant risk factors—particularly in flood-prone regions like Dashtiari—represents a crucial step toward sustainable development and minimization of flash flood damage. Materials and Methods This study integrated diverse datasets from multiple sources to develop a comprehensive flood risk analysis model for the study area. The compiled data encompassed topographic parameters—including slope, elevation, and proximity to river networks—hydrological variables, such as annual precipitation, river discharge, and groundwater levels, as well as geological attributes like soil type, permeability, and lithological structure. In addition, land use and vegetation cover data—including vegetation density, land utilization patterns, and anthropogenic modifications—were incorporated to ensure a holistic representation of watershed characteristics. All spatial data were processed and analyzed within the ArcGIS software environment. For flood hazard modeling, the Random Forest (RF) algorithm, a robust machine learning technique, was employed due to its high predictive performance and ability to handle complex interactions among variables. To further evaluate the relative contribution of each predictor to flood susceptibility, the Information Gain Ratio (IGR) was calculated, providing insight into variable importance. The methodological framework proceeded through several sequential stages: (1) collection and preprocessing of spatial data within a GIS framework; (2) selection of influential variables based on statistical relevance; (3) application of the RF model for flood hazard zoning; (4) validation of model performance using standard statistical metrics, including prediction accuracy, sensitivity, and specificity; and (5) interpretation of the resulting outputs to inform flood management strategies tailored to the region. This systematic approach not only enables precise flood risk assessment, but also supports the development of evidence-based mitigation measures. By integrating advanced modeling techniques with sound spatial analysis, the study contributes to sustainable planning and the design of protective interventions aimed at reducing flood-related damages. Research Findings The modeling results revealed a heterogeneous distribution of flood risk across the Dashtiari watershed with varying levels of vulnerability identified among different areas. According to the RF model outputs, approximately 50% of residential and agricultural lands were situated within high- or very high-risk zones, underscoring a significant exposure to potential flood events. Flood hazard zoning further indicated that 5.40% of the total study area was classified as very high risk, predominantly comprising regions adjacent to river channels and areas characterized by gentle slopes. An additional 33.63% of the watershed fell into the high-risk category attributed largely to the dense drainage network and prevalence of impermeable soil types, which together facilitated rapid runoff generation. In contrast, 37.94% of the area was designated as moderate risk, while 18.70% and 4.31% were classified as low and very low risk, respectively. The analysis of contributing factors using the IGR identified annual precipitation as the most influential variable in promoting surface runoff and triggering flood events. The Topographic Wetness Index (TWI) and land elevation also played critical roles with lower-lying areas and zones of water accumulation demonstrating heightened susceptibility. Proximity to river networks emerged as another key determinant of flood risk. Furthermore, soil characteristics—particularly clay content and low permeability—significantly enhanced runoff potential. Land use and vegetation cover also exerted a direct influence on flood dynamics; irrigated agricultural lands and bare surfaces, in particular, displayed increased sensitivity to flooding. The predictive performance of the RF model proved to be robust, achieving an overall accuracy of 85% and a kappa coefficient of 0.82. These metrics confirm the model’s strong capability in delineating flood-prone areas and provide a reliable basis for risk assessment and subsequent planning efforts. Discussion and Conclusion: The modeling results indicated that flood risk distribution in the Dashtiari watershed was heterogeneous with varying levels of vulnerability across different areas. According to the RF model, approximately 50% of residential and agricultural lands were situated in high- or very high-risk zones. Flood hazard zoning further revealed that 5.40% of the study area was classified as very high risk, predominantly comprising areas near rivers and regions with gentle slopes. An additional 33.63% fell into the high-risk category, which was attributed to the dense drainage network and impermeable soils that facilitated flooding. In contrast, 37.94% of the area was designated as medium risk, while 18.70% and 4.31% were categorized as low and very low risk, respectively. The analysis of influential factors using the IGR index identified annual precipitation as the most significant driver of surface runoff and flood generation. The TWI and land elevation also played crucial roles as lower elevations and zones of water accumulation would increase flood susceptibility. Proximity to rivers emerged as another key determinant of flood risk. Furthermore, soil type and permeability substantially influenced flood formation with clayey and impermeable soils contributing to heightened surface runoff. Land use and vegetation cover also exerted a direct effect, particularly irrigated agricultural lands and bare areas, which exhibited increased sensitivity to flood events. The RF model demonstrated strong predictive performance, achieving an overall accuracy of 85% and a kappa coefficient of 0.82 and confirming its reliability in delineating flood-prone areas and supporting its applicability for risk assessment and planning purposes. چکیده امروزه با توجه به وقوع سیلاب‌ در بیشتر حوضه‌های آبخیز، به‌ضرورت بررسی عوامل تأثیرگذار در این حوضه‌ها بیش از پیش توجه شده است. ارزیابی عوامل تأثیرگذار در خطر سیلاب به‌منظور بررسی میزان آسیب‌پذیری و مواجهه با خطر، اطلاعات ارزشمندی را برای مدیریت سیلاب فراهم می‌کند. در این راستا هدف از این تحقیق، سنجش و ارزیابی عوامل تأثیرگذار در خطر سیلاب (مطالعه موردی: حوضه دشتیاری) است. تحقیق حاضر از نظر روش تحقیق توصیفی‌تحلیلی و لحاظ هدف، کاربردی است. روش گردآوری داده‌ها، کتابخانه‌ای و میدانی است. در این تحقیق که با استفاده از 13 عامل (شیب، ارتفاع، بافت خاک و ...) به پهنه‌بندی و ارزیابی خطر سیلاب در منطقه دشتیاری اقدام شد. برای تجزیه‌وتحلیل داده‌ها از نرم‌افزار Arc GIS و شاخص IGR و از الگوریتم جنگل تصادفی (Random forest) استفاده شد. طبق نتایج حدود 5.40 درصد از منطقه مورد مطالعه دارای بیشترین خطر، 33.63 درصد در معرض خطر بالاتر، 37.94 درصد در معرض خطر متوسط، 18.70 درصد در معرض خطر کمتر و 4.31 درصد دارای کمترین خطر هستند. حدود 50 درصد از مناطق مسکونی در مناطق پرخطر واقع شده است که در صورت وقوع سیل منجر به خسارات جانی و مالی شدید می‌شود.

https://gep.ui.ac.ir/article_30496_a5fb8be266b8d05028c0d8ef4722eb2c.pdf

دانشگاه اصفهان جغرافیا و برنامه ریزی محیطی 2008-5362 37 2 2026 06 22 Perceiving Historical-Tourism Places Using Augmented Reality: A Case Study of Saray-e Mirza Jalil in the Historical Bazaar of Tabriz تحلیل کیفی تجربۀ مخاطبان در ادراک مکان‌های تاریخی - گردشگری با استفاده از واقعیت افزوده؛ مطالعۀ موردی سرای میرزا جلیل در بازار تاریخی تبریز 29 60 30706 10.22108/gep.2026.147585.1760 FA الناز ابی زاده ستادیار گروه معماری و شهرسازی، دانشگاه ملی مهارت، تهران، ایران Journal Article 2025 12 01 Abstract With the proliferation of emerging technologies, Augmented Reality (AR) has the potential to redefine the way humans interact with historical–tourism places by serving as an experiential mediator. This study qualitatively explored audiences' experiences of perceiving Iran's historical–tourism places within an AR context, focusing on Saray-e Mirza Jalil, a heritage building in Tabriz Bazaar, as a case study. Adopting a qualitative approach, the research involved the development of a practical AR prototype for the selected heritage site to create an experimental setting. Data were collected through semi-structured interviews and analyzed by using thematic analysis, encompassing 3 stages of coding: basic, organizing, and overarching themes. The findings indicated that AR transformed the visitor experience from a static, information-based encounter into a multidimensional, dynamic, analytical, and participatory process. The thematic analysis revealed 3 overarching categories: "Empowering the observation–perception–learning chain", "Enriching emotional–imaginative and cognitive experiences", and "Intelligent interaction and active user engagement". The results suggested that AR enhanced users' analytical place perception by improving architectural–historical reading, guiding attention, and facilitating situated learning, while simultaneously deepening cognitive and emotional experiences through excitement, presence, and gamified elements. Moreover, interactive and personalized features shifted users from passive observers to active, self-directed participants. This research extended the understanding of human–place relationships in the context of emerging technologies and demonstrated that AR can enhance spatial literacy, place perception, and cultural experience. These findings held practical implications for environmental planning, cultural heritage management, and smart tourism development. Keywords: Augmented Reality (AR), Place Perception, Audience Experience, Cultural Heritage, Saray-e Mirza Jalil, Environmental Planning. Introduction Iranian cities endowed with rich historical and cultural heritage have long been recognized as prominent cultural tourism destinations. However, the true value of these places extends beyond the mere presence of historical artifacts; audience experience and place perception play a pivotal role in effectively leveraging these assets. Today, technological advances and widespread access to digital tools, particularly Augmented Reality (AR), offer novel opportunities to enhance tourist experiences and represent historical places in innovative ways. Nevertheless, many studies focus primarily on the technical, interactive, and marketing aspects of these technologies, often overlooking the cultural, identity-related, semantic, and multisensory dimensions of user experience. AR has the potential to transform tourists' interaction with historical places from passive observation into an active, multidimensional, and meaningful experience. By integrating multimedia information, providing virtual guidance, and enabling personalization, AR allows for a deeper understanding of the historical structure, architecture, and layers of meaning embedded in a place. Yet, the gap between technological capabilities and actual user engagement underscores the need to analyze audience experience and identify factors influencing the adoption of emerging technologies and interaction with them. This study aimed to qualitatively analyze audience experience in perceiving Iran's historical–tourism places with a specific focus on Saray-e Mirza Jalil, a heritage building in Tabriz Bazaar, as a case study. It addressed gaps in prior research across 3 dimensions: 1) identifying the structure of user experience and how passive observation transforms into active, meaningful engagement; 2) examining the role of AR in historical representation and enhancing architectural and semantic comprehension; and 3) analyzing the effects of interaction, active participation, and cognitive–emotional engagement with historical places. Through a qualitative approach and thematic analysis, this study investigated multidimensional user experiences encompassing spatial perception, emotions, imagination, and active participation within an AR context. The findings contribute to understanding human–place relationships while informing smart tourism policy, environmental planning, cultural heritage management, and practical strategies for enhancing audience experience at historical–tourism places. Materials & Methods This study employed a qualitative research method to analyze audience experience of AR application in perceiving Iran's historical–tourism places. To create an experimental setting, a practical AR prototype was developed and implemented at Saray-e Mirza Jalil, a historical building in Tabriz Bazaar, featuring interactive and informational capabilities. These included layered architectural explanations, historical maps and documents, representations of historical figures and digital guides, 3D models, and self-guided tours. Using mobile devices, users could overlay digital layers onto the physical environment, experiencing a hybrid, interactive, and interpretive engagement with both the physical place and augmented data. The research population consisted of visitors to Saray-e Mirza Jalil, with purposive sampling employed. 30 participants were selected based on prior experience with historical places, willingness to participate, and readiness to use AR technology. Sampling continued until theoretical saturation was reached, meaning no new data emerged in the final interviews and core patterns of user experience were established. Data were collected through semi-structured interviews, focusing on 5 dimensions: cognitive (narrative understanding, historical comprehension, architectural interpretation), emotional (excitement, sense of presence, imagination, mental engagement, cultural pride), spatial (place perception, attention to architectural elements, environmental legibility), behavioral–interactive (technology use, ease of interaction, type of user engagement), and evaluative (priorities, values, overall assessment). Thematic analysis was conducted in 3 stages: (1) coding semantic units related to users' experiences into basic themes; (2) grouping similar basic themes into conceptual categories to form organizing themes; and (3) integrating organizing themes into overarching themes, thereby creating a comprehensive framework for analysis and conclusions. To ensure credibility and reliability, all coding stages were reviewed by a second coder to minimize individual bias and each data segment was confirmed or adjusted through consensus. Additionally, findings were triangulated by using direct interview excerpts and field notes. Inter-coder reliability was calculated by using Cohen's Kappa coefficient, yielding a value of 0.84, which indicated substantial agreement. Theoretical saturation was employed as a key validity indicator to ensure consistency, reproducibility, and stability of themes. This methodology enabled a comprehensive analysis of multidimensional user experiences—encompassing perception, emotion, imagination, and active participation—while clarifying the role of AR in historical representation, architectural and semantic comprehension, increased engagement, and cognitive–emotional bonding with cultural heritage. Research Findings Qualitative analysis revealed that AR application at Iran's historical–tourism places created a multidimensional, active, and interactive visitor experience. 3 primary dimensions of user experience emerged: "Empowering the Observation–Perception–Learning Chain", "Enriching Emotional–Imaginative and Cognitive Experiences", and "Intelligent Interaction and Active User Engagement". These interwoven dimensions explained how user experiences were formed at AR-enhanced historical places and could inform environmental planning strategies and smart place management. Empowering the Observation–Perception–Learning Chain:AR transformed user experiences from static, information-based observation into an analytical, participatory process. Users gained a clearer understanding of structural logic and hidden layers of historical buildings, directing attention to architectural details and historical developments. This included analytical comprehension of spatial and architectural aspects, enhanced active observation, experiential learning, and clarity of presented information. Digital layers, 3D models, and multimedia explanations allowed for gradual reconstruction of place understanding and meaningful connection between the building's form and its history without the need for human guides. Enriching Emotional–Imaginative and Cognitive Experiences:AR enhanced emotional attachment, creating a layered, imaginative experience. Users reported excitement, engagement, presence, cultural pride, and a sense of history coming alive. AR also enabled mental reconstruction of past environments, generating a quasi-documentary, creative experience. This interactive dimension fostered curiosity, motivation, sustained learning, and deeper cognitive engagement with the place. Intelligent Interaction and Active User Engagement:AR empowered users to personalize viewpoints, scale, lighting, and informational layers. Features like bookmarking, note-taking, interacting with digital characters, and 3D modeling transformed users from passive observers into active agents, supporting independent learning, spatial–analytical skill development, and creative interaction with the place. These capabilities could inform experience-driven environmental planning and smart tourism management. Overall, findings indicated that AR elevated visits to historical places beyond mere observation and digital information consumption into dynamic, analytical, participatory, and meaningful experiences. Users interacted actively with places while co-creating comprehensive, multilayered experiences encompassing history, culture, architecture, and meaning. This experience could serve as an effective tool for smart tourism planning, heritage management, and improving visitor experience quality. Discussion of Results & Conclusion This study demonstrated that AR as a multilayered experiential tool could transform human engagement with historical places, space, and cultural heritage. Qualitative analysis revealed that user experiences formed a network of spatial perception, imagination, learning, participation, and emotion while dynamically interacting with the physical and historical context of the place. AR served 3 fundamental functions: 1) empowering the observation–perception–learning chain through analytical comprehension, attention guidance, revealing hidden layers, and situated learning; 2) enriching emotional, imaginative, and cognitive experience through emotional presence, mental reconstruction of the past, and creativity; and 3) enabling intelligent interaction and active user participation with technology. These functions converted static, information-based visits into analytical, participatory, dynamic, and meaningful experiences. This study confirmed previous findings while addressing gaps in the literature, particularly in 3 areas: (1) elucidating identity, cultural, and semantic layers of historical experience with AR; (2) explaining perceptual–analytical mechanisms in place reading; and (3) clarifying the link between digital experience and visitor behavior at Iran's historical–tourism places. AR enhanced users' comprehension from descriptive observation to analytical interpretation, allowing deeper understanding of spatial structure, historical layers, and architectural logic—a key aspect of spatial literacy and place perception, central to cultural landscape and human geography studies. Emotional and imaginative AR experiences strengthened human–place connections and fostered user identity, supporting sustainable tourism and improved environmental experience quality. Active user engagement provided new capacities for smart tourism management, transforming historical places into dynamic, interactive, learning-focused environments aligned with modern planning and heritage management approaches. Practically, these findings can inform smart tourism policy, enhance audience experience, and guide the preservation and promotion of Iran's cultural heritage. Theoretically, emerging technologies like AR are not merely technical tools; they can redefine human–place relationships, place experience, meaning-making, and historical reconstruction. Studying user experience with such technologies enriches theory in human geography, cultural landscape, and place perception, facilitating deeper understanding of human behavior in spatial contexts. Ultimately, AR mediates cognitive and perceptual processes while transforming visits to historical places into dynamic, analytical, participatory, and meaningful experiences, thus highlighting the need for interdisciplinary research in this field. Future studies should explore the spatial, social, cultural, economic, and managerial impacts of AR deployment at urban and regional scales, assessing its role in sustainable tourism development and smart environmental planning. چکیده با گسترش فناوری‌های نوین، واقعیت افزوده به‌عنوان یک واسطه تجربه‌ساز می‌تواند شیوه تعامل انسان با مکان‌های تاریخی - گردشگری را بازتعریف کند. پژوهش حاضر با هدف واکاوی کیفی تجربه مخاطبان در ادراک مکان‌های تاریخی – گردشگری در بستر واقعیت افزوده انجام شده و بر سرای میرزا جلیل، یکی از بناهای تاریخی بازار تبریز، به‌عنوان مطالعه موردی متمرکز است. این پژوهش با اتکا بر رویکرد کیفی و ازطریق مصاحبه‌های نیمه‌ساختاریافته گردآوری شده است. داده‌ها با بهره‌گیری از روش تحلیل مضمون شامل کدگذاری سه‌مرحله‌ای مضامین پایه، سازمان‌دهنده و فراگیر تحلیل شده است. یافته‌ها نشان می‌دهد که واقعیت افزوده تجربه بازدید را از مواجهه‌ای ایستا و اطلاعات‌محور به فرایندی چندبعدی، پویا، تحلیلی و مشارکتی تبدیل می‌کند. تحلیل مضمون، سه مقوله فراگیر شامل «توانمندسازی زنجیره مشاهده، درک و یادگیری»، «غنی‌سازی تجربه احساسی – تخیلی و ذهنی» و «تعامل هوشمند و مشارکت فعال کاربر» را آشکار ساخت. نتایج بیانگر آن است که واقعیت افزوده با ارتقای خوانش کالبدی – تاریخی مکان، هدایت توجه و تسهیل یادگیری موقعیت‌مند، ادراک تحلیلی کاربران را تقویت می‌کند و همزمان از طریق ایجاد هیجان، حس حضور و افزودن عناصر بازی‌وار، تجربه شناختی و احساسی آنان را تعمیق می‌بخشد. علاوه بر این، قابلیت‌های تعاملی و شخصی‌سازی، کاربران را از مشاهده‌گر منفعل به مشارکت‌کننده فعال و خودراهبر تبدیل می‌کند. نتایج پژوهش فهم رابطه انسان – مکان را در بستر فناوری‌های نوظهور، گسترش و نشان می‌دهد واقعیت افزوده می‌تواند فرایندهای سواد فضایی، خوانش مکان و تجربه فرهنگی را ارتقا دهد. این یافته‌ها برای برنامه‌ریزی محیطی، مدیریت میراث فرهنگی و توسعه گردشگری هوشمند دارای دلالت‌های کاربردی مهمی است.

https://gep.ui.ac.ir/article_30706_8d03882b746cc192f8b06a8e3de54730.pdf

دانشگاه اصفهان جغرافیا و برنامه ریزی محیطی 2008-5362 37 2 2026 06 22 Spatial Evaluation and Modeling of Suitable Locations for High-Rise Buildings in Bandar Abbas Using FAHP, Machine Learning, and Spatial Stability Analysis ارزیابی و مدل‌سازی فضایی مکان‌های مناسب برای ساختمان‌های بلندمرتبه در بندرعباس با بهره‌گیری ازFAHP، یادگیری ماشین و پایداری مکانی 61 92 30691 10.22108/gep.2026.149230.1792 FA پیام رضایی دانشجوی دکتری گروه مهندسی عمران، واحد قشم، دانشگاه آزاد اسلامی، قشم‌، ایران محسن دادرس استادیار گروه مهندسی عمران، واحد بندرعباس، دانشگاه آزاد اسلامی، بندرعباس‌، ایران مهدی آجدانی استادیار گروه مهندسی کامپیوتر، واحد قشم، دانشگاه آزاد اسلامی، قشم‌، ایران شهلا پاسلار استادیار گروه مهندسی صنایع، واحد بندرعباس، دانشگاه آزاد اسلامی،بندرعباس‌، ایران Journal Article 2026 06 08 Abstract Spatial assessment and modeling of suitable locations for high-rise buildings play a crucial role in environmental planning and controlling urban sprawl. This study aimed to identify suitable areas for high-rise building development in Bandar Abbas, Iran, through an integrated approach combining Multi-Criteria Decision-Making (MCDM), machine learning, and uncertainty analysis. Initially, 37 sub-criteria grouped into 4 categories—physical, infrastructural, socio-economic, and environmental—were defined and weighted by using the Fuzzy Analytic Hierarchy Process (FAHP). Subsequently, Random Forest (RF) and Extreme Gradient Boosting (XGBoost, XGB) algorithms were employed to re-evaluate the importance of criteria and generate spatial suitability maps. To assess model sensitivity, a tornado analysis was conducted, while uncertainty and stability were evaluated by using Monte Carlo simulation and a coefficient-of-variation-based spatial stability index. Additionally, Cohen's Kappa coefficient was calculated to measure the spatial agreement among the final outputs of the three approaches. The results indicated that physical criteria had the highest influence on high-rise building suitability with a fuzzy weight of 0.646. The RF model identified larger suitable areas, whereas FAHP and XGB provided more conservative estimations. Stability and uncertainty analyses demonstrated that the XGB model exhibited higher spatial robustness and reliability compared to RF. Moreover, a substantial agreement was observed between XGB and FAHP results (Kappa=0.83), indicating strong spatial consistency. Overall, the findings suggested that integrating MCDM methods, machine learning algorithms, and uncertainty analysis provided a robust and reliable framework for high-rise building site selection and urban decision-making support. The results of the three models indicated that the northwestern and central-eastern parts of Bandar Abbas were most suitable for high-rise development. Furthermore, the spatial stability index confirmed that XGB performed more consistently than RF, making it a more reliable model for urban planning applications related to high-rise development suitability mapping. Keywords: High-Rise Development, Sensitivity Analysis, Machine Learning, Monte Carlo, Bandar Abbas. Introduction Rapid urbanization and the increasing demand for land in large cities have intensified the need for vertical development and construction of high-rise buildings. In this context, the appropriate siting of such land uses plays a crucial role in reducing pressure on urban land, preventing horizontal urban sprawl, and improving spatial efficiency. Bandar Abbas, due to its geographical location, specific climatic conditions, infrastructural limitations, and environmental sensitivities, requires careful planning for high-rise development. Considering the complexity of factors influencing site selection, the use of integrated approaches combining Multi-Criteria Decision-Making (MCDM), machine learning, and uncertainty analysis can enhance the accuracy and reliability of results. This study aimed to provide a comprehensive framework for evaluating and modeling suitable locations for high-rise buildings in Bandar Abbas. Materials & Methods In this study, 37 sub-criteria were identified and structured into 4 main criteria: accessibility to infrastructure and physical, environmental, and socio-economic factors. The criteria weights were determined by using the Fuzzy Analytic Hierarchy Process (FAHP). After weighting, all spatial layers were standardized and normalized and a Weighted Linear Combination (WLC) method was applied to generate the initial land suitability map. Subsequently, to validate and compare FAHP results, machine learning algorithms, including Random Forest (RF) and Extreme Gradient Boosting (XGB), were employed to re-estimate the importance of criteria. Three final suitability maps were produced based on the FAHP, RF, and XGB models. Sensitivity analysis was performed by using the tornado method. In addition, uncertainty and stability were assessed by using Monte Carlo simulation and a coefficient-of-variation-based spatial stability index. Furthermore, the spatial agreement between the models was evaluated by using the Kappa coefficient. Research Findings The results indicated that in all three methods (FAHP, RF, and XGB), the physical criterion had the highest importance in high-rise building site selection. However, its weight in machine learning models (approximately 0.80) was significantly higher that of FAHP (0.469). This suggested that data-driven models were more sensitive to the physical characteristics of the urban environment and assigned greater importance to these factors in determining land suitability. In contrast, FAHP allocated higher importance to infrastructural and socio-economic criteria, reflecting a more comprehensive expert-based perspective compared to machine learning algorithms. The similarity between RF and XGB results indicated strong convergence and stability of these two models in spatial analysis, while FAHP provided a more balanced distribution of weights among the criteria. Sensitivity analysis revealed that the physical criterion has the greatest impact on the final output, where small changes in its weight significantly altered the spatial suitability pattern. The other criteria showed lower sensitivity and contributed more stable effects within the model. Regarding uncertainty analysis, Monte Carlo simulation results demonstrated that the XGB model exhibited higher spatial stability compared to RF as random changes in criterion weights had less influence on its outputs. Moreover, the high Kappa coefficient (0.83) between XGB and FAHP indicated a strong agreement between these two approaches. Spatially, the northwestern and central-eastern parts of Bandar Abbas were identified as the most suitable areas for high-rise development. Discussion of Results & Conclusion The findings of this study demonstrated that integrating multi-criteria decision-making methods, machine learning techniques, and uncertainty analysis provides a comprehensive, accurate, and reliable framework for high-rise building site selection. Model comparison showed that FAHP offers a more expert-driven and balanced perspective, while RF and XGB rely more on spatial data patterns and observed relationships. However, the XGB model is recommended as the most robust approach due to its higher stability, better consistency with FAHP, and more balanced performance in identifying suitable areas. Overall, this integrated framework can serve as an effective tool for urban planners to control urban sprawl, optimize land use, and guide high-rise development in Bandar Abbas. Future studies are recommended to incorporate climatic variables, population growth scenarios, and temporal datasets to enable more dynamic urban modeling. مکان‌یابی ساختمان‌های بلندمرتبه، نقش مهمی در برنامه‌ریزی محیطی و کنترل گسترش افقی شهرها دارد. پژوهش حاضر با هدف مکان‌یابی این ساختمان‌ها در شهر بندرعباس با استفاده از تلفیق روش‌های تصمیم‌گیری چندمعیاره، یادگیری ماشین و تحلیل عدم قطعیت انجام شد. در این راستا، ابتدا 37 زیرمعیار در قالب چهار معیار کالبدی، زیرساختی، اقتصادی - اجتماعی و فیزیکی تنظیم و با روش تحلیل سلسله‌مراتبی فازی (FAHP) وزن‌دهی شدند. سپس الگوریتم‌های RF و XGBoost به‌منظور وزن‌دهی مجدد معیارها و تولید نقشه‌های مطلوبیت مکانی به کار گرفته شدند. برای ارزیابی حساسیت مدل‌های یادگیری، از تحلیل تورنادو استفاده شد و عدم‌قطعیت و پایداری نتایج با مونت‌کارلو و شاخص پایداری مبتنی بر ضریب تغییرات انجام گرفت. همچنین، ضریب کاپا برای سنجش میزان توافق فضایی بین نقشه‌های نهایی سه روش محاسبه شد. نتایج نشان داد که معیارهای کالبدی با وزن فازی 646/0 بیشترین تأثیر را در مکان‌یابی ساختمان‌های بلندمرتبه دارند. مدل RF پهنه‌های مناسب بیشتری را شناسایی کرد؛ درحالی‌که XGB و FAHP دیدگاه محافظه‌کارانه‌تری ارائه دادند. نتایج تحلیل پایداری و عدم قطعیت بیان‌کننده آن بود که مدل XGB از ثبات فضایی و قابلیت اعتماد بیشتری نسبت به RF برخوردار است. همچنین ضریب کاپای بالا بین XGB و FAHP (83/0) نشان‌دهنده توافق قابل توجه این دو روش بود. در مجموع، یافته‌های پژوهش نشان می‌دهد که تلفیق روش‌های تصمیم‌گیری چندمعیاره، یادگیری ماشین و تحلیل عدم قطعیت می‌تواند چارچوبی کارآمد و قابل اعتماد برای مکان‌یابی ساختمان‌های بلندمرتبه و پشتیبانی از تصمیم‌گیری‌های شهری فراهم سازد. نتایج مدل‌های سه‌گانه بیان‌کننده این است که اراضی واقع در شمال‌غربی و بخش‌های شرقی - مرکزی بندرعباس برای بلندمرتبه‌سازی ساختمانی مناسب‌اند. نتایج شاخص پایداری بیان‌کننده آن است که مدل XGB در مقایسه با RF عملکرد قابل اعتمادتر و سازگارتری در تولید نقشه مطلوبیت مکانی داشته و می‌تواند به‌عنوان مدل پایدارتر برای پشتیبانی از تصمیم‌گیری‌های شهری در زمینه توسعه ساختمان‌های بلندمرتبه استفاده شود.

https://gep.ui.ac.ir/article_30691_9ead7a52c222f71342d93c9d50f3cc36.pdf