دانشگاه فردوسی مشهد جغرافیا و مخاطرات محیطی 2322-1682 15 2 2026 06 22 An Automatic Method for Flood Event Selection and Characterization: A Case Study of Kajo River Basin, Iran روش خودکار برای انتخاب رویدادهای سیلاب و شناسایی ویژگی‌های آن‌ها (مطالعۀ موردی: حوضۀ آبریز رودخانه کاجو، ایران) 1 26 48008 10.22067/geoeh.2026.97442.1642 FA فاطمه گنجی گوهری گروه جغرافیای طبیعی، دانشکده جغرافیا و برنامه‌ ریزی محیطی، دانشگاه سیستان و بلوچستان، زاهدان، ایران 0009-0001-0791-4898 حمید نظری پور گروه جغرافیای طبیعی، دانشکده جغرافیا و برنامه‌ ریزی محیطی، دانشگاه سیستان و بلوچستان، زاهدان، ایران 0000-0002-9655-6091 محمدرضا پودینه گروه جغرافیای طبیعی، دانشکده جغرافیا و برنامه‌ ریزی محیطی، دانشگاه سیستان و بلوچستان، زاهدان، ایران محسن حمیدیان پور گروه جغرافیای طبیعی، دانشکده جغرافیا و برنامه‌ ریزی محیطی، دانشگاه سیستان و بلوچستان، زاهدان، ایران علیرضا قائمی گروه جغرافیای طبیعی، دانشکده جغرافیا و برنامه‌ ریزی محیطی، دانشگاه سیستان و بلوچستان، زاهدان، ایران رضا تیموری گروه مهندسی آب، دانشکده مهندسی آب و خاک، دانشگاه علوم کشاورزی و منابع طبیعی گرگان، گرگان، ایران Journal Article 2026 01 01 The selection of flood events and determination of their characteristics (such as start and end times, peak discharge, volume, and duration) are crucial initial steps in flood analysis. In this study, to accurately extract key flood information, the automated Peaks Over Threshold (POT) model was used for flood sampling, and an automated method based on the Master Recession Curve (MRC) analysis was developed to determine flood characteristics. This process was implemented using a graphical user interface (GUI) and a toolbox in MATLAB. Results demonstrated that the proposed method performs well in the Kajo River basin, despite its variable hydrological regime. The developed toolbox can be easily applied to other watersheds for flood sampling and event description, thereby helping to reduce uncertainties in subsequent flood analyses, such as multivariate frequency and trend analyses. The practical applications of this approach in flood management, hydraulic structure design, flood risk assessment, and climate change studies in river basins are significant. However, the method's dependence on the quality and continuity of daily flow data, the need to adjust threshold parameters and independence criteria in basins with different hydroclimatic conditions, and challenges related to accurate baseflow separation are notable limitations of this research. For future studies, it is recommended that this method be validated in basins with diverse hydrological regimes, including snow-dominated basins and areas affected by human activities such as dam construction.<br /><strong>Introduction</strong><br />Floods rank among the most devastating natural disasters, causing extensive financial and human losses worldwide, including in Iran. Climate change and the increased frequency of extreme climatic events have further underscored the importance of accurate flood analysis and risk management. The core of any flood analysis whether multivariate frequency analysis, trend assessment, or hydraulic structure design depends on the precise identification of individual flood events and the quantitative extraction of their key characteristics (flood features). These features typically include the event's start and end time, peak discharge (peak), total direct runoff volume, and flood duration. Conventional methods for extracting these features are often based on manual or semi-automated procedures that rely on the analyst's subjective judgment, posing challenges such as time consumption, lack of full reproducibility, and difficulty in transferability to different watersheds. Particularly when dealing with large volumes of streamflow time-series data, these methods become inefficient and prone to introducing systematic uncertainties in subsequent analytical stages. The two main methodological challenges in this path are the objective and optimal selection of flood events from continuous data series and the accurate separation of the flood hydrograph from baseflow to calculate the true volume. The common approach for flood sampling is the Peaks Over Threshold method, which itself heavily depends on determining an optimal threshold and criteria for event independence. On the other hand, simplified methods such as connecting a straight line between start and end points are often used to define event bounds and calculate volume, which lack sufficient accuracy. Therefore, the development of an automatic, objective, and generic method that can standardize the complete process of event selection and feature extraction with minimal user intervention is a research and operational necessity. This study aims to address this methodological gap by proposing a comprehensive automatic framework and testing it in a basin with a complex and variable hydrological regime the Kajo River basin in southeastern Iran. The Kajo basin, influenced by two distinct rainfall systems (Mediterranean and Monsoonal), the presence of an upstream dam, and a history of destructive floods, provides an ideal testing ground for evaluating the efficiency and robustness of the proposed method.<br /><strong>Material and Methods</strong><strong> </strong><br />This research was conducted using an integrated and automated methodological framework that covers all stages from flood event selection to feature extraction. The study area is the Kajo River basin in southeastern Iran, with an approximate area of 6,178 square kilometers. The primary data used were daily streamflow time series from two hydrometric stations, Chandokan (located in the upstream section) and Pirsohrab (located in the downstream section), over a 50-year period. The methodological core is based on two main pillars. The first stage is the automatic selection of flood events using the Peaks Over Threshold model. To eliminate the inherent subjectivity in threshold determination, an automated algorithm based on the Anderson-Darling goodness-of-fit test was implemented. This algorithm systematically evaluates candidate thresholds and selects as optimal the threshold for which the exceedance series shows the best fit with the Generalized Pareto Distribution. Simultaneously, the well-known independence criteria from the United States Water Resources Council were applied to distinguish independent events, and the validity of the final extracted flood sample was evaluated using Kendall's correlation coefficient test and the calculation of the Dispersion Index. The second stage is the automatic identification of characteristics for each flood event, which itself comprises two key processes: first, the start and end times of each flood are determined algorithmically based on ratios of the peak discharge and a fixed time window; subsequently, for the precise separation of the net flood hydrograph from baseflow and calculation of the actual volume, the Master Recession Curve method is employed. In this step, using the matching strip method, the master recession curve is extracted from the entire data period and fitted with Wittenberg's analytical equation. Deep baseflow is defined as the average of minimum flows during the dry period and is used as a baseline. Finally, this complex methodological framework is implemented in a user-friendly toolbox with a graphical interface in the MATLAB software environment, named SFE_IFC, to make the analysis process fully automated and executable on large datasets.<br /><strong>Results and Discussion </strong><br />The application of the proposed framework to the Kajo River basin data yielded significant quantitative and qualitative results. The optimal flood thresholds for the Chandokan (upstream) and Pirsohrab (downstream) stations were calculated as 52 and 177.6 m³/s, respectively. This significant difference clearly illustrates the impact of increasing basin area and hydrological changes along the river course. Statistical evaluations confirmed the correct performance of the algorithm; Dispersion Index values close to one and non-significant Kendall's correlation coefficients at the 95% confidence level validated the assumption of event independence and the adequacy of the modeling. Over the 50-year period, 102 flood events were automatically identified at the Chandokan station and 82 events at the Pirsohrab station. Analysis of the extracted features revealed a strong positive correlation between flood volume and its duration. The coefficient of determination for this relationship was approximately 0.61 for the Pirsohrab station and 0.36 for the Chandokan station. This result indicates that flood volume is a primary determinant of the flood duration period in this basin. One of the important findings of this research was elucidating the role of different rainfall regimes on flood characteristics. Results showed that approximately 84% of floods at Pirsohrab and 65% at Chandokan occurred during the Mediterranean rainfall season (cold season). The visual outputs from the toolbox (separated hydrographs) clearly demonstrated that the proposed method can correctly identify start and end times for both single-peak and complex multi-peak floods. Furthermore, the fit of the master recession curve with a coefficient of determination greater than 0.9 proved the method's high accuracy in modeling the recession limb and, consequently, in calculating volume. Despite the construction of the Zirdan Dam upstream, the observation of large floods downstream (at the Pirsohrab station) indicates the intensity of flood processes in this basin and the necessity for considering complementary management measures.<br /><strong>Conclusion </strong><br />This study successfully developed and validated a generic automatic method for the objective selection of flood events and the accurate extraction of their characteristics from daily streamflow data. The presented framework, by integrating an automatic optimal threshold selection algorithm within the Peaks Over Threshold model and the Master Recession Curve method, systematically addresses two major methodological problems in flood analysis. Implementing this framework in a user-friendly MATLAB toolbox has transformed it into a powerful operational tool for researchers and engineers, capable of drastically reducing analysis time while simultaneously enhancing reproducibility and transparency. Applying the method in the Kajo River basin not only proved its efficacy in a complex hydrological environment influenced by two climatic regimes but also provided valuable hydrological insights into the occurrence patterns and characteristics of floods in this important basin, including the dominance of cold-season floods in terms of frequency and the superiority of warm-season floods in terms of volume and duration. The findings emphasize the urgent need for focused attention on flood management and the development of flood control infrastructure in basins similar to Kajo. While the proposed method represents an important step toward standardization, its dependence on the quality and continuity of input data and the need for empirical adjustment of some parameters for basins with vastly different conditions remain its limitations. For future studies, validation of the method in basins with diverse hydroclimatic regimes, its integration with radar rainfall data and rainfall-runoff models for causal analysis, and investigation of the impacts of climate change on flood characteristics using this framework are recommended. انتخاب رویدادهای سیلاب و تعیین ویژگی‌های آن، نخستین گام‌های حیاتی برای تحلیل‌های سیلاب محسوب می‌شوند. در این بررسی، برای به‌دست‌آوردن دقیق اطلاعات کلیدی سیلاب، از مدل خودکار آستانه‌ای، برای نمونه‌برداری از سیلاب‌ها استفاده و یک روش خودکار برای تعیین ویژگی‌های سیلاب با به‌کارگیری روش تحلیل منحنی فروکش اصلی(MRC) ارائه گردید. همچنین از یک رابط کاربری گرافیکی (GUI) و جعبه ابزار برای این فرآیند در متلب استفاده شده است. نتایج نشان داد که روش پیشنهادی برای حوضه آبریز رودخانه کاجو با رژیم هیدرولوژیکی متغیر، عملکرد خوبی دارد. جعبه ابزار توسعه‌یافته را می‌توان به راحتی در سایر حوضه‌های آبخیز برای نمونه‌برداری سیلاب و توصیف وقایع سیلاب به کار برد. کاربرد عملی این رویکرد در تحلیل‌های مدیریت سیلاب، طراحی سازه‌های هیدرولیکی، ارزیابی ریسک سیل و مطالعات تغییر اقلیم در حوضه‌های آبریز قابل توجه است. با این حال، وابستگی روش به کیفیت و پیوستگی داده‌های جریان روزانه، نیاز به تنظیم پارامترهای آستانه و معیارهای استقلال در حوضه‌های با شرایط هیدرواقلیمی متفاوت، و چالش‌های مربوط به تفکیک دقیق جریان پایه از جمله محدودیت‌های قابل توجه این پژوهش محسوب می‌شوند. برای پژوهش‌های آینده، پیشنهاد می‌شود این روش در حوضه‌های با رژیم‌های هیدرولوژیکی متنوع اعتبارسنجی گردد. https://geoeh.um.ac.ir/article_48008_43034f3b0edfd1513d2ad87c7e387ac9.pdf

دانشگاه فردوسی مشهد جغرافیا و مخاطرات محیطی 2322-1682 15 2 2026 06 22 Prioritization of Flood Potential in Sub-Basins Flowing from the Northern and Southern Slopes to Tabriz Metropolis Using SWARA-MABAC Methods and Morphometric Indices اولویت بندی سیل خیزی زیرحوضه‌های جاری از دامنه‌های شمالی و جنوبی به کلان شهر تبریز با استفاده از روش های SWARA-MABAC و شاخص های مورفومتری 27 43 48143 10.22067/geoeh.2026.98037.1655 FA محمدحسین رضائی مقدم گروه ژئومورفولوژی، دانشکده برنامه‌ریزی و علوم محیطی، دانشگاه تبریز، تبریز، ایران 0000-0003-2626-4656 شهرام روستائی گروه ژئومورفولوژی، دانشکده برنامه‌ریزی و علوم محیطی، دانشگاه تبریز، تبریز، ایران 0000-0003-0664-1688 توحید رحیم پور گروه ژئومورفولوژی، دانشکده برنامه‌ریزی و علوم محیطی، دانشگاه تبریز، تبریز، ایران عبدالحمید سرتیپی گروه ژئومورفولوژی، دانشکده برنامه‌ریزی و علوم محیطی، دانشگاه تبریز، تبریز، ایران Journal Article 2026 02 24 On the northern and southern slopes of Tabriz, several small and large drainage basins direct their runoff toward the city. Urban expansion toward the outlets of these basins, together with extensive land‑use changes and modifications to natural watercourses in recent decades, has increased the risk of flooding. This study aimed to evaluate and prioritize the flood‑susceptibility potential of drainage basins draining into the metropolitan area of Tabriz using morphometric analyses and multi‑criteria decision‑making (MCDM) models. To this end, the morphometric characteristics of the northern and southern basins—including basin area, number of streams, drainage density, stream frequency, drainage texture, drainage intensity, circularity ratio, form factor, compactness coefficient, basin relief, and ruggedness number—were extracted using a Geographic Information System (GIS). The SWARA method was then applied to determine the weights of the criteria affecting flood susceptibility, followed by the MABAC model to rank the basins based on their flood‑susceptibility potential. The results revealed that among the northern basins, Basin N1 exhibited the highest flood susceptibility, with a final weight of 0.206. Similarly, among the southern basins, Basin S6 was identified as the most sensitive unit, with a weight of 0.366. These basins mainly owe their high flood‑susceptibility potential to larger area, high drainage density, and steep slopes.<br /><strong>Introduction</strong><br />Several small and large drainage basins flow toward the city of Tabriz. The city’s expansion toward the outlets of these basins, along with extensive land‑use changes and the alteration of natural watercourses over recent decades, has increased the flood hazard. During precipitation events, these basins collect and convey rainfall-derived waters that manifest as surface runoff and ephemeral flows within the basins. Subsequently, following the steep and generally uniform topographic gradient of the slopes, these flows are directed downstream toward lower-lying areas and the urban extent of Tabriz. Meanwhile, the rapid and unregulated physical expansion of the Tabriz metropolis toward the outlets of these basins, together with the continuous intensification of construction activities—including residential and commercial buildings as well as transportation networks—within and across the corridors of secondary watercourses, and without adequate consideration of hydrological and geomorphological constraints in urban planning and design processes, has substantially increased the risk of various hydro-geomorphological hazards. These hazards include sudden and destructive flash floods as well as unstable slope movements in the affected areas.<br /><strong>Material and Methods</strong><br />The study area of the present research comprises the basins toward the metropolitan area of Tabriz. Tabriz, the capital of East Azerbaijan Province, is located at geographical coordinate's 38° 02′– 38°10′ north latitude and 46° 11′ – 46° 23′ east longitude, with a mean elevation of approximately 1366 m above sea level. According to records from the synoptic meteorological station at Tabriz International Airport, the city’s mean annual precipitation during the period 1992–2024 is approximately 255 mm.<br />In the present study, a Digital Elevation Model (DEM) with a spatial resolution of 12.5 meters was employed to delineate the catchment areas of the region and to calculate their morphometric parameters. Subsequently, two different methods were employed for data analysis: (1) the MABAC model and (2) the SWARA model. The Multi-Attributive Border Approximation area Comparison (MABAC) method was introduced by Pamučar and Ćirović in 2015. This method is among the most recent multi-criteria decision-making (MCDM) techniques and is widely used for ranking alternatives. The SWARA model is applied as a multi-criteria decision-making technique to calculate the weights of criteria and sub-criteria.<br /><strong>Results and Discussion</strong><br />In the present study, the morphometric characteristics of the northern and southern basins of Tabriz were analyzed, including parameters such as basin area, stream number, drainage density, stream frequency, drainage texture, drainage intensity, circularity ratio, form factor, compactness coefficient, relief, and ruggedness number. These parameters were extracted using a Geographic Information System (GIS). The research findings indicate that the morphometric characteristics of drainage basins play a decisive role in intensifying or mitigating flood susceptibility. In this regard, criteria such as basin area, stream density, and elevation were identified as the most influential factors affecting runoff generation and the occurrence of flash floods. The results obtained from the SWARA model for criteria weighting reveal that parameters related to basin size and shape (e.g., basin area, compactness coefficient, and circularity ratio), as well as drainage network characteristics (such as drainage density and stream number), contribute most significantly to increased flood susceptibility.<br /><strong>Conclusion</strong><br />This study was conducted with the aim of assessing and prioritizing the flood susceptibility potential of basins into the metropolitan area of Tabriz, through the integration of morphometric analysis and the combined application of the MABAC and SWARA multi criteria decision making methods. Basin prioritization using the MABAC model indicated that Basin N1, located on the northern slopes of Tabriz, was identified as the most flood-prone basin, with the highest final weight (0.206). This basin, due to a combination of relatively large area, high drainage density, and considerable slope, provides favorable conditions for rapid runoff generation and an increased flood hazard. In contrast, basins with negative final weights (such as Basins 2 and 4 in the northern sector) exhibit lower flood susceptibility, which is mainly attributed to moderating morphological characteristics such as elongated basin shape and low drainage density. In the southern sector, Basin N6 ranked first with a final weight of 0.366. The predominance of this basin is primarily due to its larger area, significant elevation differences, and high drainage density, all of which are key factors contributing to increased runoff volume and flow velocity. These findings indicate that even in semi-arid climatic regions, the presence of large and steep mountainous basins can significantly intensify the risk of flash floods, particularly during wet seasons. From an applied perspective, the results of this research can serve as an effective tool for urban, environmental, and watershed managers and planners. The proposed prioritization facilitates the optimal allocation of resources and the implementation of both structural and non-structural measures—such as the construction of sediment retention dams, land-use pattern modification, conservation and restoration of vegetation cover, and the design of early warning systems—in high-risk basins. Particularly in light of the rapid and unregulated physical expansion of Tabriz toward the outlets of these basins, the incorporation of hydrological and geomorphological considerations into urban development plans and the enforcement of construction controls within stream buffer zones are increasingly imperative. In conclusion, this study confirms the effectiveness of integrating Geographic Information Systems (GIS), quantitative morphometric analyses, and multi-criteria decision-making methods in basin-scale natural hazard assessment. It is recommended that future studies enhance model accuracy and reliability by incorporating hydro climatic parameters (such as rainfall intensity and duration), anthropogenic factors (including land-use change and building density), and higher-resolution datasets. Such an approach would not only contribute to improved flood risk management in the metropolitan area of Tabriz but could also serve as a methodological framework for other cities located adjacent to mountainous drainage basins.<br /><strong>Acknowledgements</strong><strong> </strong><br />We are grateful to all the scientific consultants of this paper. در دامنه‌های شمالی و جنوبی تبریز، حوضه‌های آبریز کوچک و بزرگی وجود دارند که جریان آب آن‌ها به سوی شهر هدایت می‌شود. گسترش و توسعه شهر در جهت خروجی این حوضه‌ها، همراه با تغییرات گسترده در کاربری اراضی و دست‌کاری مسیرهای طبیعی عبور آب طی چند دهه اخیر، موجب افزایش خطر وقوع سیلاب شده است. این پژوهش با هدف ارزیابی و اولویت‌بندی پتانسیل سیل‌خیزی حوضه‌های آبریز منتهی به کلان‌شهر تبریز با استفاده از تحلیل‌های مورفومتری و مدل‌های تصمیم‌گیری چندمعیاره انجام شده است. بدین منظور، ابتدا ویژگی‌های مورفومتری حوضه‌های شمالی و جنوبی تبریز شامل پارامترهای مساحت، تعداد آبراهه، تراکم زهکشی، فراوانی آبراهه، بافت زهکشی، شدت زهکشی، نسبت دایره‌ای، نسبت‌ شکل‌پذیری، ضریب فشردگی، اختلاف ارتفاع و عدد ناهمواری با استفاده از سیستم اطلاعات جغرافیایی استخراج شدند. سپس با به‌کارگیری روش SWARA وزن معیارهای مؤثر در سیل‌خیزی تعیین گردید. در ادامه، با استفاده از مدل MABAC، حوضه‌های آبریز بر اساس پتانسیل وقوع سیل اولویت‌بندی شدند. نتایج نشان داد که در بین حوضه‌های شمالی، حوضه N1 با وزن نهایی 0.206 بالاترین پتانسیل سیل‌خیزی را دارا می‌باشد. در بین حوضه‌های جنوبی نیز حوضه S6 با وزن 0.366 به‌عنوان حساس‌ترین حوضه شناسایی شد. این حوضه‌ها عمدتاً به دلیل دارا بودن مساحت بزرگ‌تر، تراکم زهکشی بالا و شیب تند، از پتانسیل سیل‌خیزی بالایی برخوردار هستند. https://geoeh.um.ac.ir/article_48143_17b3cfe4f4ddd1f23555acfdd525c446.pdf

دانشگاه فردوسی مشهد جغرافیا و مخاطرات محیطی 2322-1682 15 2 2026 06 22 Analysis of Early Warning Signals of Desertification Risk Using Time Series of Remote Sensing Based Indices (Case Study: Mashhad County, Razavi Khorasan Province, Iran) تحلیل سیگنال‌های پیش هشدار خطر بیابان‌زایی با استفاده از سری‌های زمانی شاخص‌های مبتنی بر سنجش از دور (مطالعه موردی: شهرستان مشهد، استان خراسان رضوی) 44 62 48207 10.22067/geoeh.2026.97882.1652 FA عبدالحسین بوعلی گروه مدیریت مناطق خشک و بیابانی، دانشکده منابع طبیعی و محیط زیست، دانشگاه فردوسی مشهد، مشهد، ایران 0009-0002-8014-1067 مرتضی اکبری گروه مدیریت مناطق خشک و بیابانی، دانشکده منابع طبیعی و محیط زیست، دانشگاه فردوسی مشهد، مشهد، ایران 0000-0002-8637-266X Journal Article 2026 02 22 Desertification is one of the major challenges in arid and semi‑arid regions of the world, exerting profound impacts on ecological sustainability, agriculture, and local livelihoods. This study aims to analyze early‑warning signals of desertification risk using time‑series remote sensing indices and field data in Mashhad County, Iran. For this purpose, the NDVI, NDSI, and TGSI indices were employed over the period 2005–2024. Satellite images, after geometric and radiometric corrections, were used alongside field observations for the required calculations. Breakpoint detection in the time series was performed using the BFAST software package, and statistical indicators (lag‑1 autocorrelation, skewness, and standard deviation) were evaluated in R. To examine the breakpoint behavior, trends were analyzed for 100 pixels, including 50 degraded and 50 non‑degraded pixels identified during field surveys. The results indicated that the main breakpoint occurred in 2013, coinciding with a sharp decline in vegetation cover and an increase in soil salinity. The NDVI, NDSI, and TGSI indices showed the strength of changes based on the Kendall’s tau test as 0.71, 0.58, and 0.53, respectively. These differences highlight the prominent role of vegetation cover and soil salinity in driving desertification in the region. The findings further revealed that NDVI and NDSI, together with lag‑1 autocorrelation, exhibited the highest efficiency in the early detection of ecosystem changes. Overall, the results underscore the importance of integrating remote sensing indices with statistical approaches to develop early‑warning systems and support sustainable natural resource management for desertification control.<br /><strong>Introduction </strong><br />   Desertification is widely recognized as one of the most severe environmental challenges affecting arid and semi‑arid regions across the globe, with far‑reaching consequences for ecological stability, agricultural productivity, and human livelihoods. Driven by a complex interplay of climatic pressures such as declining precipitation, rising temperatures, and increased evapotranspiration and human‑induced factors including overgrazing, deforestation, groundwater depletion, and unsustainable land‑use practices, desertification has intensified in recent decades, particularly in the Middle East and Iran where fragile ecosystems are highly vulnerable to disturbance. According to global assessments, hundreds of millions of people are directly affected by land degradation, and billions more live in regions at risk, underscoring the urgent need for effective monitoring and early‑warning systems. Traditional assessment methods, which often rely on static indicators, are insufficient for capturing the dynamic and nonlinear nature of ecosystem degradation. Consequently, remote sensing time‑series analysis has emerged as a powerful tool for detecting subtle environmental changes before they evolve into irreversible degradation. Indices such as the Normalized Difference Vegetation Index (NDVI), Normalized Difference Salinity Index (NDSI), and the Terrain Gradient Stability Index (TGSI) provide valuable insights into vegetation dynamics, soil salinity, and land surface conditions. When combined with statistical early‑warning indicators such as lag‑1 autocorrelation, skewness, and standard deviation these datasets can reveal critical slowing down, reduced resilience, and approaching ecological thresholds. Mashhad County, located in northeastern Iran, has experienced increasing environmental stress due to prolonged droughts, declining groundwater levels, and expanding agricultural and urban activities. These pressures make the region an ideal case study for evaluating early‑warning signals of desertification using long‑term remote sensing data and advanced statistical techniques.<br /><strong>Material and Methods</strong><strong> </strong><br />   This study utilized a comprehensive time‑series dataset of three remote sensing indices NDVI, NDSI, and TGSI spanning the period from 2005 to 2024 to assess early‑warning signals of desertification in Mashhad County. In this study, Satellite images were used for calculations after being validated with reference data. The Breaks for Additive Seasonal and Trend (BFAST) algorithm was employed to decompose each time series into trend, seasonal, and abrupt change components, enabling the detection of structural breakpoints associated with ecosystem shifts. To evaluate early‑warning indicators of declining resilience, three statistical metrics lag‑1 autocorrelation, skewness, and standard deviation were calculated using the R statistical environment. These indicators are widely used to detect critical slowing down, a hallmark of systems approaching tipping points. A total of 100 representative pixels were selected for detailed analysis, including 50 degraded and 50 non‑degraded sites identified through extensive field surveys. This sampling strategy allowed for a robust comparison between stable and vulnerable areas. In addition to temporal analysis, spatial patterns of change were examined across the entire study area to map the distribution and intensity of degradation processes. The integration of remote sensing indices, statistical early‑warning metrics, and field‑based validation provided a comprehensive framework for assessing both temporal dynamics and spatial heterogeneity of desertification risk.<br /><strong>Results and Discussion </strong><br />   The time‑series analysis revealed a distinct and consistent breakpoint in NDVI and NDSI during the period 2013–2014, marking a critical transition in ecosystem behavior. This abrupt change corresponded to a sharp decline in vegetation cover and a significant increase in soil salinity, indicating a shift toward reduced ecosystem productivity and heightened degradation pressure. In contrast, TGSI did not exhibit a clear structural breakpoint, suggesting that this index may be less sensitive to short‑term environmental fluctuations. Following the breakpoint, NDVI displayed a pronounced downward trend in degraded areas, while NDSI increased steadily, reflecting intensified salinization and reduced vegetation health. Non‑degraded areas, however, maintained relatively stable trends throughout the study period, highlighting the contrasting resilience levels across the landscape. Statistical early‑warning indicators further supported these findings: lag‑1 autocorrelation showed a strong and persistent increasing trend across all indices, with high positive Kendall’s tau values, indicating reduced system resilience and approaching instability. Although skewness and standard deviation also exhibited occasional increases, their sensitivity and discriminatory power were notably weaker than that of autocorrelation. Spatial analysis revealed that the western and southern regions of Mashhad experienced the most significant changes in NDVI, NDSI, and TGSI, corresponding to areas with intensive agricultural activity, groundwater depletion, and higher exposure to climatic stress. Kendall’s tau values of 0.71 for NDVI, 0.58 for NDSI, and 0.53 for TGSI underscore the dominant role of vegetation dynamics and soil salinity in driving desertification processes. These results align with previous studies demonstrating that increasing salinity, declining vegetation cover, and reduced ecological resilience are key precursors to land degradation in arid environments.<br /><strong>Conclusion</strong><br />   This study demonstrates that integrating remote sensing time‑series indices with statistical early‑warning indicators provides a robust and effective framework for detecting emerging desertification risks in arid and semi‑arid regions. The identification of a major breakpoint in 2013–2014 highlights a critical shift toward ecological instability in Mashhad County, driven primarily by vegetation decline and increasing soil salinity. Among the indices analyzed, NDVI and NDSI exhibited the highest sensitivity to environmental changes, while lag‑1 autocorrelation emerged as the most reliable early‑warning metric, effectively capturing reduced ecosystem resilience prior to major transitions. Spatial analysis further revealed that the western and southern parts of Mashhad are the most vulnerable to degradation, emphasizing the need for targeted land management interventions in these areas. Overall, the findings underscore the importance of remote sensing‑based early‑warning systems for sustainable land management, enabling policymakers and environmental managers to implement timely mitigation strategies and prevent the progression of desertification. By providing a scientifically grounded approach to monitoring ecosystem health, this study contributes to the development of proactive management frameworks aimed at preserving ecological stability and supporting long‑term environmental sustainability in vulnerable regions. بیابان‌زایی یکی از چالش‌های اساسی در مناطق خشک و نیمه‌خشک جهان است که بر پایداری اکولوژیکی، کشاورزی و معیشت جوامع تأثیر عمیق می‌گذارد. این پژوهش با هدف تحلیل سیگنال‌های پیش‌هشدار خطر بیابان‌زایی با استفاده از سری زمانی شاخص ­های سنجش از دوری و داده ­های میدانی در شهرستان مشهد انجام شد. برای این منظور، از شاخص‌های سنجش از دور NDVI، NDSI و TGSI طی دوره زمانی سال‌های 2005 تا 2024 میلادی استفاده گردید. در این پژوهش تصاویر ماهواره‌ای پس از صحت ­سنجی با داده ­های مرجع برای محاسبات مورد استفاده قرار گرفتند. تحلیل نقاط شکست سری‌های زمانی با استفاده از مجموعه نرم افزاری BFAST و ارزیابی شاخص‌های آماری (خودهمبستگی lag-1، چولگی، انحراف معیار) در نرم افزار R انجام شد. در این پژوهش به منظور بررسی نقطه شکست، روند تغییرات در 100 پیکسل شامل 50 پیکسل مناطق تخریب یافته و 50 پیکسل مناطق تخریب نیافته که در طی بازدید میدانی تعیین شده بود، انجام گرفت. نتایج نشان داد که نقطه شکست اصلی در سال 2013 رخ داد است، که با کاهش شدید پوشش گیاهی و افزایش شوری خاک همراه بود. به ترتیب شاخص ­های NDVI ، NDSI و TGSI قدرت تغییرات را  براساس آزمون آماری کندال تای 71/0 ، 58/0 و  53/0 نشان دادند.  این اختلاف در مقادیر نشان داد که پوشش گیاهی و شوری خاک  نقش برجسته­ ای در بیابان ­زایی منطقه ایفا می­ کنند. نتایج نشان داد که شاخص‌های NDVI  و NDSI همراه با خودهمبستگی lag-1 بیشترین کارایی را در شناسایی زودهنگام تغییرات اکوسیستمی دارند. یافته‌ها اهمیت ترکیب شاخص‌های سنجش از دور با روش‌های آماری را در توسعه سیستم‌های هشدار سریع و مدیریت پایدار منابع طبیعی برای کنترل بیابان‌زایی نشان می‌دهد. https://geoeh.um.ac.ir/article_48207_3dfc20df8217703e7233c770fb10b97a.pdf

دانشگاه فردوسی مشهد جغرافیا و مخاطرات محیطی 2322-1682 15 2 2026 06 22 Flood Risk Assessment, Prediction and Zonation in the Kashafrud Basin Using Machine Learning Algorithms ارزیابی، پیش بینی و پهنه بندی خطر سیلاب در حوضه کشف رود با استفاده از الگوریتم های یادگیری ماشین 63 92 47958 10.22067/geoeh.2026.96351.1624 FA محمد علی زنگنه اسدی گروه آب‌وهواشناسی و ژئومورفولوژی دانشکده جغرافیا و علوم محیطی، دانشگاه حکیم سبزواری، سبزوار، ایران 0000-0003-2027-4853 لیلا گلی مختاری گروه آب‌وهواشناسی و ژئومورفولوژی دانشکده جغرافیا و علوم محیطی، دانشگاه حکیم سبزواری، سبزوار، ایران 0000-0003-3797-0222 مهناز ناعمی تبار گروه آب‌وهواشناسی و ژئومورفولوژی دانشکده جغرافیا و علوم محیطی، دانشگاه حکیم سبزواری، سبزوار، ایران Journal Article 2025 11 06 Floods, as one of the most destructive natural hazards, cause irreparable damage to infrastructure and human communities on an annual basis. The present study aims to produce an accurate flood hazard zonation map for the Kashafrud basin by employing two advanced machine learning algorithms, namely Logistic Model Tree (LMT) and Random Forest (RF). In this research, thirteen influential parameters including elevation, curvature, precipitation, drainage density, aspect, soil and geological characteristics, slope, distance from streams, land use, normalized difference vegetation index (NDVI), topographic wetness index (TWI), and stream power index (SPI) were analyzed as input variables. Data corresponding to 145 recorded flood locations were identified and divided into training (70%) and validation (30%) datasets. Following the weighting of the thematic layers, flood susceptibility maps were generated and classified into five hazard classes using the natural breaks classification method. Model performance was evaluated using the Receiver Operating Characteristic (ROC) curve, which indicated that the LMT model, with an AUC value of 0.897, exhibited higher predictive accuracy than the RF model, which achieved an AUC of 0.811. The flood hazard zonation results reveal that high-risk areas are predominantly concentrated in regions characterized by gentle slopes, impermeable formations, low elevations, and floodplains. Areas classified as very high and high risk are mainly located in the central and outlet sections of the basin. These findings not only confirm the effectiveness of machine learning algorithms in flood hazard zonation but also underscore the necessity of strategic management planning and targeted basin management practices in this flood-prone region.<br /><strong>Introduction</strong><br />Floods are among the most significant and destructive natural hazards worldwide, annually affecting millions of people and causing severe damage to human life, critical infrastructure, agriculture, industry, and both urban and rural areas (Organization, 1999; Das, 2019; Daneshparvar, 2021). In recent years, climate change, rapid population growth, and accelerated urban expansion have contributed to increases in both the frequency and severity of flood events, underscoring the urgent need for effective flood risk management and mitigation strategies. The Kashafrud Basin, located in northeastern Iran, is particularly vulnerable to severe flooding due to its geomorphological characteristics, diverse geological structures, variable precipitation patterns, and intensive human activities. Effective flood risk management in this basin requires accurate identification of flood-prone areas, comprehensive analysis of influential factors, and the application of advanced modeling techniques. Accordingly, this study aims to identify and classify flood-prone areas within the Kashafrud Basin by applying machine learning algorithms specifically the Logistic Model Tree (LMT) and Random Forest (RF) to evaluate the relative contributions of both natural and anthropogenic factors within a data-driven scientific framework.<br /><strong>Material and Methods</strong><br />In this study, 145 flood and non-flood locations were identified using data obtained from the Watershed Management Organization, complemented by satellite imagery interpretation. The dataset was randomly divided into two subsets, comprising 70% for model training and 30% for validation. Key explanatory variables, including land use, elevation classes, drainage density, lithology, slope, aspect, distance from waterways, rainfall, soil types, and vegetation and moisture indices (TWI, SPI, and NDVI), were derived from topographic maps, geological maps, and 2024 satellite imagery. All spatial datasets were processed and analyzed using ArcGIS 10.4 and ENVI 5.4 software, generating both quantitative and qualitative indicators for each variable.<br />To address potential multicollinearity among the independent variables, the Variance Inflation Factor (VIF) and Tolerance (TOL) indices were calculated. Variables exhibiting high interdependence were excluded to enhance model reliability and predictive performance. Flood susceptibility modeling was conducted using two advanced machine learning algorithms: the LMT model, which integrates decision tree structures with logistic regression, and the RF model, which employs an ensemble of decision trees. Model performance was evaluated using sensitivity, specificity, the Kappa coefficient, the Receiver Operating Characteristic (ROC) curve, and the Area Under the Curve (AUC). Flood hazard zonation maps were produced using the natural breaks classification method, and the model outputs were converted into thematic flood risk maps.<br /><strong>Results and Discussion</strong><br />Multicollinearity analysis confirmed that none of the selected variables exceeded the threshold values (VIF > 10 or TOL < 0.1), indicating a robust and stable modeling framework. The results identified thirteen key factors including slope, elevation, curvature, distance from rivers, drainage density, rainfall, soil type, land use, and vegetation and moisture indices as significant determinants of flood risk. Areas characterized by low to moderate slopes and elevations below 1000 m were found to exhibit increased surface runoff velocity, thereby intensifying flood risk and severity, particularly in the central and outlet sections of the basin. Terrain curvature and dense drainage networks significantly influenced runoff concentration and flow pathways, highlighting the critical role of topography and drainage structure in flood dynamics.<br />Resistant geological formations and impermeable clay-rich soils substantially reduced infiltration capacity, leading to increased surface runoff and the occurrence of sudden and hazardous flood events. In addition, weak vegetation cover and inappropriate land use changes particularly urban and industrial development indirectly exacerbated both runoff volume and flow velocity. Low NDVI values reflected degraded vegetation conditions and a heightened susceptibility to flooding.<br />Flood hazard zonation maps generated using the LMT model indicated that approximately <strong>81.23%</strong> of the basin falls within moderate to very high flood risk classes. Similarly, the RF model estimated that <strong>68.44%</strong> of the study area is exposed to comparable risk levels, with high-risk zones predominantly concentrated in the central and outlet regions of the basin. Statistical validation demonstrated that the LMT model achieved higher predictive accuracy (<strong>AUC = 0.897</strong>) compared to the RF model (<strong>AUC = 0.811</strong>), although both models exhibited satisfactory and reliable performance for flood susceptibility assessment. High sensitivity, specificity, and Kappa coefficient values further confirmed the robustness and operational applicability of both modeling approaches.<br />The findings indicate that geomorphological and hydrological factors such as slope, elevation, drainage density, soil characteristics, and impermeable geological formations play a more dominant role in flood occurrence than anthropogenic factors, including land use and human-induced disturbances. Nevertheless, the influence of human activities on flood intensity and spatial distribution remains significant, emphasizing the necessity for integrated and sustainable watershed management strategies. These results are consistent with recent international studies, which highlight the primary contribution of natural variables to flood risk while acknowledging the amplifying effects of human interventions.<br /><strong>Conclusion </strong><br />By integrating remote sensing data, GIS-based spatial analysis, and advanced machine learning algorithms, this study presents a comprehensive and innovative framework for flood risk assessment, analysis, and spatial zonation in one of the most critical watersheds in northeastern Iran. The results demonstrate that the LMT model outperforms the RF model in accurately identifying flood-prone areas within the Kashafrud basin. The resulting flood hazard maps provide valuable decision-support tools for prioritizing crisis management actions, land use planning, and the development of resilient water infrastructure.<br />While accurate consideration of natural and geomorphological factors is fundamental to effective flood risk management, regulating human activities such as land use planning, construction of flood resilient infrastructure, restoration of vegetation cover, and implementation of early warning systems is equally essential for minimizing flood-related damages. Emphasizing data-driven and machine learning based approaches, this research contributes to bridging the gap between applied scientific research and practical flood management needs in Iranian catchments. Future studies incorporating advanced algorithms, uncertainty analysis, and higher-resolution satellite data are strongly recommended to further enhance the resilience of aquatic and urban ecosystems. سیلاب به‌عنوان یکی از مخرب‌ترین مخاطرات طبیعی، سالانه خسارات جبران‌ناپذیری به زیرساخت‌ها و جوامع انسانی وارد می‌کند. پژوهش حاضر با هدف پهنه‌بندی دقیق خطر سیلاب در حوضه آبخیز کشف‌رود، از دو الگوریتم پیشرفته یادگیری ماشین LMT و RF بهره برده است. در این مطالعه، ۱۳ پارامتر مؤثر شامل ارتفاع، انحنا، بارش، تراکم زهکشی، جهت شیب، ویژگی‌های خاک و زمین‌شناسی، شیب، فاصله از آبراهه، کاربری اراضی، شاخص پوشش گیاهی‌(NDVI)، رطوبت توپوگرافی (TWI) و توان آبراهه (SPI) به‌عنوان متغیرهای ورودی مورد تحلیل قرار گرفت. داده‌های مربوط به ۱۴۵ نقطه سیلابی شناسایی و به دو مجموعه آموزش (۷۰٪) و اعتبارسنجی (۳۰٪) تقسیم شد. پس از وزن‌دهی لایه‌ها، نقشه‌های پتانسیل وقوع سیل با روش شکست طبیعی به پنج طبقه خطر طبقه‌بندی گردید. ارزیابی دقت مدل‌ها با استفاده از منحنی ROC نشان داد که مدل LMT با مقدار AUC برابر 897/0، دقت بالاتری نسبت به مدل RF با AUC برابر 811/0 دارد. نتایج پهنه‌بندی حاکی از تمرکز مناطق پرخطر در نواحی با شیب کم، سازندهای غیرقابل نفوذ، ارتفاعات پایین و دشت‌های سیلابی است. نواحی با خطر بسیار زیاد و زیاد عمدتاً در بخش‌های مرکزی و خروجی حوضه متمرکز شده‌اند. https://geoeh.um.ac.ir/article_47958_1204aa07efb15ce7cbe8a5a185a41b21.pdf

دانشگاه فردوسی مشهد جغرافیا و مخاطرات محیطی 2322-1682 15 2 2026 06 22 Modeling and Forecasting Riverbed Changes and Riparian Land Use Patterns along the Aras River in Ardabil Province مدل‌سازی و پیش‌بینی تغییرات بستر و الگوهای کاربری اراضی حاشیه‌ای رودخانه ارس در محدوده استان اردبیل 93 117 47961 10.22067/geoeh.2026.96349.1623 FA رضا طلائی پژوهشکده حفاظت خاک و آبخیزداری کشور، سازمان تحقیقات، آموزش و ترویج کشاورزی، تهران،‌ ایران 0009-0008-4853-2459 علیرضا حبیبی پژوهشکده حفاظت خاک و آبخیزداری کشور، سازمان تحقیقات، آموزش و ترویج کشاورزی، تهران،‌ ایران 0000-0003-4815-4028 Journal Article 2025 11 09 The Aras River, as a transboundary meandering river system, has experienced pronounced morphodynamic and riparian land-use changes driven by the combined effects of natural processes and human interventions. This study aimed to conduct a quantitative spatial analysis of riverbed dynamics and riparian land-use changes along the Aras River during the period 2010–2024, and to predict future trends up to the 2031 horizon within Ardabil Province, Iran. To this end, multi-temporal Landsat 5 TM (2010), Landsat 8 OLI (2017) and Sentinel-2 satellite imagery were integrated with field observations related to changes in riverbed and land-use, along with GIS techniques, to detect spatiotemporal patterns of channel migration and land-use transformation. Future changes were predicted using an integrated Cellular Automata and Markov model. The results reveal a substantial reduction in the riverbed area, declining from 1,821.6 ha in 2010 to 939.47 ha in 2024, whereas agricultural lands (31,968 ha) and residential areas (829.99 ha) exhibited the highest spatial persistence. Concurrently, tree-shrubland expanded markedly, increasing from 921.6 to 1,684.7 ha over the study period. Model validation confirmed the high predictive capability of the CA–Markov model, with an overall accuracy of 99%, a Kappa coefficient of 0.97, and an AUC value of 0.926. Model projections indicate that by 2031 approximately 25.5% of the riverbed will be converted into tree and shrubland and 2.92% into agricultural land, reflecting continued lateral channel migration, riverbed morphological alteration, and a tendency toward relative channel desiccation. The intensification of bank erosion and channel displacement poses increasing risks to agricultural lands and human settlements. These findings highlight the need for integrated riparian land-use management, bank stabilization strategies, and flood risk mitigation in transboundary river systems.<br /><strong>Introduction</strong><br />Rivers, as dynamic geomorphological systems, play a crucial role in sediment transport, floodplain formation, and the stability of surrounding ecosystems. Meandering rivers, due to the complexity of hydrodynamic and sedimentary processes, have long been a focus of scientific research. Their morphological changes are influenced by hydrological, tectonic, and human-induced factors, and understanding them requires a multi-scale approach. Remote sensing technologies and geographic information systems (GIS) enable tracking river course changes and quantifying morphometric indices over decades. The Aras River, with its meandering course and numerous bends in the northern part of Ardabil Province, is a prominent example of this dynamism. Its channel migration has led to bed changes, lateral erosion, and the threat to agricultural lands, natural resources, and settlements. Land-use changes and unsustainable exploitation of natural resources further intensify these hydrodynamic processes. This study aims to analyze the morphological changes of the Aras River from 2010 to 2024 and predict future trends until 2031. Using multi-temporal satellite imagery combined with spatiotemporal CA–Markov modeling, dynamic changes in river morphology and surrounding land-use were quantified, and hazard maps identifying high-risk areas in agricultural lands, rangelands, woody and shrub vegetation, shrublands and residential zones were produced. The findings provide valuable insights for sustainable water and soil resource management, mitigation of flood and erosion hazards, and enhancement of environmental and economic security in border regions.<br /><strong>Material and Methods</strong><br />To analyze the morphological and land-use dynamics of the Aras River and predict future trends up to 2031, an integrated modeling approach was applied using the TerrSet environment. The study area covers a 73-km reach of the Aras River in the Moghan Plain, Ardabil Province, characterized by high fluvial dynamism and geomorphological sensitivity. Multitemporal satellite imagery, including Landsat 5 TM (2010), Landsat 8 OLI (2017), and Sentinel-2 (2024) with 10–30 m spatial resolution, served as the primary dataset. Supplementary data from Google Earth and field surveys were used for validation and accurate delineation of land-use boundaries. Classification accuracy was assessed through overall accuracy, Kappa coefficient and the ROC curve. The CA–Markov hybrid model was employed as the core predictive framework. The transition probability matrix was computed using 2010–2017 data to simulate the 2024 scenario, which was validated against the actual 2024 map. Upon confirming model reliability, the 2031 land-use projection was generated. The model integrates Markov analysis (temporal transitions) with Cellular Automata (spatial dynamics), enabling detailed simulation of landscape evolution. Finally, the simulated changes in land-use, as well as in the extent and channel of the Aras River, were integrated with the potential economic losses to examine the bidirectional interactions between natural fluvial processes and human activities, along with their associated economic impacts on agricultural lands, natural resources, and infrastructures. This methodological framework provides a scientific foundation for sustainable land management and flood-risk mitigation in the Moghan Plain located along the northern border of Ardabil Province.<br /><strong>Results and Discussion</strong><br />The Aras River, located in the northern part of Ardabil Province, exhibits a highly dynamic and meandering course that plays a crucial role in shaping floodplains and controlling sediment deposition. Continuous erosion along concave banks and deposition on convex ones have resulted in channel migration and significant local morphological changes. Fertile agricultural lands and diverse vegetation have developed along the river margins, which, although productive, remain highly vulnerable to natural river dynamics and land loss. To mitigate erosion and protect infrastructure, several embankments have been constructed; however, these structures, while effective in the short term, often modify natural flow regimes and intensify downstream erosion.<br />Land-use analyses from 2010, 2017, and 2024 reveal a clear expansion of human-modified zones—especially agricultural, residential, and industrial areas—while the riverbed area has shrunk significantly by 2024. Between 2010 and 2017, the main river channel migrated primarily northward, affecting around 373.35 hectares. Some of the abandoned river areas were converted to tree and shrub cover, while the riverbed width decreased by approximately 64.78 hectares. Agricultural and residential lands showed the highest persistence (≈99%), while rangelands experienced the most substantial changes. From 2010 to 2024, agricultural and residential areas remained largely stable (99% and nearly 100%, respectively), whereas the riverbed underwent the most drastic transformations, retaining only 734 out of 1,868 hectares. Some of these transformed zones were converted to agricultural, rangeland, or tree/shrub areas. These results emphasize the stability of anthropogenic land-uses and the high dynamism of riverine environments over the 14-year period.<br />Using the CA–Markov model, the 2024 land-use simulation based on 2010–2017 data showed high predictive accuracy (Overall accuracy = 99%, Kappa = 0.97, AUC = 0.926). The model predicted that by 2031, agricultural (99.89%), residential (98.86%), and tree/shrub (97.66%) areas will remain highly stable, while the riverbed will continue to be the most dynamic feature, with only a 68.33% probability of persistence. Around 25.5% of the riverbed is expected to convert to tree/shrub cover and 2.9% to agriculture, indicating continued channel adjustment and further desiccation in the future.<br />The economic assessment estimated total losses of about 22,971 billion rials by 2031, mainly due to riverbed conversion to natural cover (19,193 billion rials). Changes in agricultural (60 ha; 1,811 billion rials) and residential (18 ha; 530 billion rials) lands also contributed substantially. Based on the estimated value of 30 billion rials per hectare in 2025, agricultural and residential zones represent the highest economic risk, whereas rangelands and natural covers serve critical ecological functions. Among the six studied reaches, reaches 3 and 5 are most vulnerable, with estimated losses of 7,106 and 6,145 billion rials, respectively. Therefore, priority management should focus on these sectors through channel stabilization, protective vegetation, and adaptive land-use planning. Historical floods in Parsabad and Aslanduz (Modarres, Sarhadi & Burn, 2016; Khoshnoodmotlagh et al., 2020) confirm recurring flood hazards.<br /><strong>Conclusion</strong><br />The Aras River in Ardabil Province is experiencing active morphological and functional transformation driven by meandering channel migration, bank erosion, and sediment deposition. Analysis of satellite imagery and CA–Markov modeling (2010–2024) revealed that agricultural and residential lands remain largely stable, whereas the riverbed is highly dynamic, with significant portions transforming into vegetated and agricultural areas. Model validation (Overall accuracy = 99%, Kappa = 0.97, AUC = 0.926) confirmed its reliability, and predictions for 2031 indicate continued riverbed contraction alongside high stability of cultivated and settled areas.<br />These results highlight the strong interaction between fluvial processes and surrounding land-use dynamics, demonstrating that channel migration directly influences the stability of agricultural and residential lands. Without integrated management, ongoing morphological evolution may intensify economic losses and ecological degradation. Therefore, implementing targeted floodplain planning, riverbank stabilization, and systematic monitoring especially in high-risk reaches is essential for ensuring long-term sustainability. This study provides a scientific basis for sustainable water and soil management, erosion and flood risk mitigation, and rational land-use planning in the transboundary Moghan Plain. رودخانه ارس به‌عنوان یکی از سامانه‌های رودخانه‌ای مئاندری و مرزی ایران، تحت تأثیر هم‌زمان فرآیندهای طبیعی و مداخلات انسانی، دچار تغییرات سریع مورفودینامیکی و کاربری اراضی در پهنه‌های ساحلی خود شده است. این پژوهش با هدف تحلیل کمّی-فضایی تغییرات بستر رودخانه و الگوی کاربری اراضی در کرانه‌های رودخانه ارس طی دوره ۲۰۱۰ -۲۰۲۴ و پیش‌بینی روندهای آتی تا افق ۲۰۳۱ در محدوده استان اردبیل انجام شد. بدین منظور، از تصاویر ماهواره‌ای چندزمانه Landsat 5 TM (2010)، Landsat 8 OLI (2017) و Sentinel-2 در تلفیق با داده‌های برداشت میدانی مربوط به تغییرات در بستر و کاربری اراضی و سامانه اطلاعات جغرافیایی برای استخراج تغییرات مکانی-زمانی استفاده شد. پیش‌بینی تغییرات آینده با بهره‌گیری از مدل ترکیبی سلول‌های خودکار و مارکوف صورت گرفت. نتایج نشان داد که مساحت بستر رودخانه طی دوره مطالعه کاهش چشمگیری داشته و از ۱۸۲۱٫۶ هکتار در سال ۲۰۱۰ به ۹۳۹٫۴۷ هکتار در سال ۲۰۲۴ رسیده است، در حالی که اراضی کشاورزی (۳۱۹۶۸ هکتار) و کاربری‌های مسکونی (۸۲۹٫۹۹ هکتار) بالاترین سطح پایداری مکانی را حفظ کرده‌اند. هم‌زمان، پوشش درختی-درختچه‌ای و بوته‌زار با افزایشی قابل توجه از ۹۲۱٫۶ به ۱۶۸۴٫۷ هکتار گسترش یافته است. نتایج اعتبارسنجی مدل سلول‌های خودکار-مارکوف نشان‌دهنده توان پیش‌بینی بسیار بالای آن است (صحت کلی 99 درصد، ضریب کاپا 97/0 و AUC برابر با 926/0). پیش‌بینی‌های مدل نشان می‌دهند که تا سال ۲۰۳۱ حدود 5/25 درصد از بستر رودخانه به پوشش‌های درختی-درختچه‌ای و بوته‌زار و ۲٫۹۲ درصد به اراضی کشاورزی تبدیل خواهد شد که بیانگر تداوم مهاجرت جانبی کانال، تغییر مورفولوژی بستر و گرایش به خشک‌شدگی نسبی پهنه رودخانه‌ای است. تشدید فرسایش کرانه‌ای و جابه‌جایی مسیر رودخانه تهدیدی فزاینده برای اراضی کشاورزی و سکونتگاه‌های انسانی منطقه محسوب می‌شود. یافته‌های این پژوهش بر ضرورت اتخاذ رویکردهای یکپارچه در مدیریت کاربری اراضی ساحلی، تثبیت کرانه‌ها و کاهش خطرات سیلاب در رودخانه‌های مرزی تأکید دارد. https://geoeh.um.ac.ir/article_47961_8d745f13fd904a9719c97f7ef22d99bc.pdf

دانشگاه فردوسی مشهد جغرافیا و مخاطرات محیطی 2322-1682 15 2 2026 06 22 Analyzing the Factors Influencing Avalanche Occurrence Using Multivariate Statistical Methods (Case Study: Karaj Dam Watershed) تحلیل عوامل مؤثر بر وقوع بهمن با استفاده از روش‌های آماری چندمتغیره (مطالعۀ موردی: حوزه آبخیز سد کرج) 118 140 48077 10.22067/geoeh.2026.97613.1644 FA مجید کاظم زاده گروه مرتع و آبخیزداری، دانشکده منابع طبیعی و محیط زیست، دانشگاه فردوسی مشهد، مشهد، ایران 0000-0002-9150-353X زهرا نوری گروه مهندسی احیا مناطق خشک و کوهستانی، دانشکده منابع طبیعی، دانشگاه تهران، کرج، ایران Journal Article 2026 01 30 Avalanches are among the most significant natural hazards in the mountainous regions of the Central Alborz, particularly in areas with tourism activities and sensitive infrastructure, where they pose a serious threat to human life and economic systems. The objective of this study is to identify and analyze the factors influencing avalanche occurrence in the upstream area of the Karaj Dam watershed using multivariate statistical methods. Avalanche occurrence data for the period 2023–2024 were collected based on field observations and local evidence, including 150 avalanche points and 142 non-avalanche points used as control data. In addition, 20 topographic, morphometric, geological, land use, and climatic variables were extracted using a 10 m resolution Digital Elevation Model (DEM), station-based climatic records, and thematic maps. Principal Component Analysis (PCA) was applied to reduce data dimensionality and extract dominant underlying patterns. Hierarchical clustering using Ward’s method was performed to group variables with similar characteristics, and Discriminant Function Analysis (DFA) was used to differentiate avalanche-prone and non-avalanche areas. The PCA results led to the extraction of six independent components that collectively explained 81.51% of the total variance. Based on the clustering outcomes, the variables influencing avalanche occurrence were categorized into three homogeneous clusters. The DFA results showed that Relative Slope Position, Valley Depth, Topographic Wetness Index, and Distance to River had the highest discriminative power between avalanche and non-avalanche areas. The overall classification accuracy of the model was 85.9%, while the cross-validation accuracy reached 84.5%, demonstrating the high effectiveness of multivariate statistical approaches in predicting avalanche occurrence.<br /><strong>Extended</strong> <strong>Abstract</strong><strong> </strong><br /><strong>Introduction</strong><br />Avalanches are one of the most significant natural hazards in the mountainous regions of the Central Alborz, particularly in areas with tourism activities and sensitive infrastructure. Avalanches pose a serious threat to human lives and economic activities, especially in areas with tourism and critical infrastructure. Every year, avalanches cause considerable damage to residential areas, transportation routes, and vital infrastructures. Therefore, identifying areas prone to avalanches and implementing appropriate measures to control and mitigate them plays a crucial role in managing and reducing the impacts of this natural phenomenon. Several factors influence the formation and occurrence of avalanches, with the most important being snow cover characteristics, meteorological factors such as the type and amount of precipitation, temperature, and morphological features including slope, aspect, elevation, slope curvature, land use, and geological conditions. In recent years, global climate changes, which are primarily associated with rising temperatures, have significantly altered weather patterns in terms of spatial and temporal distribution. This warming affects the formation, stability, and transformation of snow cover, which in turn influences avalanche occurrence. The increase in temperature enhances the freeze-thaw cycles, contributing to the formation of weak layers in snow masses. Therefore, the objective of this research is to identify and analyze the factors affecting avalanche occurrence in the upstream area of the Karaj Dam watershed, including the sub-watersheds of Velayat Rud, Kasil Nesa, and Varange Rud, using multivariate statistical methods.<br /><strong>Material and Methods</strong><br />In this study, the factors influencing avalanche occurrence were analyzed, focusing on morphological, topographical, climatic, and environmental characteristics. For this purpose, 20 factors affecting avalanche occurrence were considered, including morphometric, topographical, geological, geomorphological, land use, and climatic variables. Avalanche occurrence data were collected based on field observations, historical evidence, and surveys conducted between 2023 and 2024. A total of 150 avalanche-prone points from 50 potential areas were identified, and 142 non-avalanche points were selected as control data. To reduce data dimensions and identify the most significant factors, Principal Component Analysis (PCA) was used. For clustering areas with similar features, hierarchical clustering with the Ward method was employed. Then, to differentiate and compare avalanche-prone and non-prone areas, Discriminant Function Analysis (DFA) was applied. Statistical tests such as KMO, Bartlett’s test, Wilks' Lambda, and cross-validation were used to evaluate the models. The statistical and processing analyses in this study were performed using ArcGIS, SAGA GIS, SPSS, and R software.<br /><strong>Results and Discussion</strong><br />The results of the Principal Component Analysis (PCA) showed that six principal components were extracted, which explained 81.51% of the total data variance. The first component indicated the relative position of the slope and the distance from rivers, the second component showed elevation, precipitation, and temperature, and the third component represented plan curvature and slope length index. Other components highlighted slope, terrain ruggedness index, aspect, and profile curvature, showing the role of topography and solar radiation in avalanche occurrence. Hierarchical clustering analysis identified three homogeneous clusters: The first cluster included topographic ruggedness index (TRI), slope, aspect, valley depth (VD), topographic wetness index (TWI), analytical hillshading (AH), elevation, and precipitation, representing areas with unstable slopes and high snow accumulation. The second cluster included plan curvature (PL), convergence index (CI), topographic position index (TPI), profile curvature (PRL), distance from rivers (RD), relative position of slopes (RSP), vector ruggedness index (VRM), slope length (LS), and distance from faults (DTF), which indicated areas influenced by morphometric and geomorphological features. The third cluster included only temperature, which separately represented the climatic effects. This clustering analysis indicates that avalanche occurrence is influenced by a combination of morphometric, environmental, and climatic factors, and it allows for accurate identification of avalanche-prone areas. Discriminant Function Analysis (DFA) showed significant differences between avalanche and non-avalanche groups (Wilks' Lambda = 0.432, p = 0.001). The relative position of the slope (RSP), valley depth (VD), and topographic wetness index (TWI) were identified as the most important factors distinguishing avalanche-prone areas. The DFA model correctly classified 85% of the points, and its accuracy was 84% in cross-validation. The results suggest that the selected topographic and environmental indices well represent the spatial pattern of avalanche occurrence. Moreover, 88.7% of avalanche-prone areas were found to have pasture land use. In terms of geology, the majority of the area lies in the Karaj formation (56.1%), Quaternary deposits (9.2%), and the Shemshak formation (5.7%). Avalanche occurrence is the result of a complex interaction of topographic, environmental, and climatic factors, and the models can be useful for avalanche risk prediction and management. The results obtained can be applied in the preventive management of hazards, land use planning, and reducing human and financial losses, especially in high-traffic tourist areas such as the Dizin Ski Resort.<br /><strong>Conclusion</strong><br />The results of this study showed that avalanche occurrence in the upstream area of the Karaj Dam watershed is the result of a complex and multidimensional interaction of topographic, geomorphological, and climatic factors, and no single variable can explain the spatial pattern of this phenomenon. Principal Component Analysis (PCA) revealed the dominant structures between environmental variables by reducing data dimensions, showing that the elevation-precipitation-temperature gradient, the relative position of slopes to the drainage network, surface geometry of the slope, and moisture conditions are among the most significant components affecting snow mass instability. The extraction of six independent components, which explain more than 80% of the total data variance, indicates the high potential of this method for simplifying and interpreting complex environmental data. Hierarchical clustering results showed that the factors affecting avalanche occurrence could be categorized into three main groups: factors related to terrain roughness and slope, geomorphological factors guiding avalanche movement, and the climatic factor of temperature. Discriminant Function Analysis (DFA) confirmed the key role of some variables in distinguishing between avalanche-prone and non-prone areas. The relative position of the slope, valley depth, topographic wetness index, and distance from rivers were identified as the most important distinguishing factors. In contrast, variables such as vector ruggedness index, slope aspect, and distance from faults showed limited influence on distinguishing avalanche-prone areas. Therefore, this study demonstrates that the combined use of multivariate statistical methods can be an effective tool for identifying key factors, reducing uncertainty, and improving avalanche risk zoning in mountainous areas.<br /><strong>Acknowledgements</strong><strong> </strong><br />The authors would like to thank financial supporting of Ferdowsi University of Mashhad (Poject No. 64099) in the present study. بهمن یکی از مهم‌ترین مخاطرات طبیعی در مناطق کوهستانی البرز مرکزی است که به‌ویژه در نواحی دارای کاربری گردشگری و زیرساخت‌های حساس، تهدیدی جدی برای جان انسان‌ها و فعالیت‌های اقتصادی محسوب می‌شود. هدف این پژوهش، شناسایی و تحلیل عوامل مؤثر بر وقوع بهمن در بالادست حوزه آبخیز سد کرج با بهره‌گیری از روش‌های آماری چندمتغیره است. در این پژوهش، داده‌های وقوع بهمن طی سال‌های ۲۰۲۳ تا ۲۰۲۴ بر اساس مشاهدات میدانی و شواهد محلی گردآوری شد که شامل ۱۵۰ نقطه بهمنی و ۱۴۲ نقطه غیر‌بهمنی به‌عنوان داده‌های کنترل بود. همچنین، ۲۰ متغیر توپوگرافی، مورفولوژیکی، زمین‌شناسی، کاربری اراضی و اقلیمی با استفاده از مدل رقومی ارتفاع 10 متری، داده‌های ایستگاهی هواشناسی و نقشه‌های موضوعی استخراج گردید. برای کاهش ابعاد داده‌ها و استخراج الگوهای اصلی از تحلیل مؤلفه‌های اصلی (PCA) استفاده شد. همچنین، خوشه‌بندی سلسله‌مراتبی به روش Ward برای گروه‌بندی متغیرهای مشابه و تحلیل تابع تشخیصی (DFA) برای تفکیک مناطق وقوع و عدم وقوع بهمن به‌کار رفت. نتایج تحلیل مؤلفه‌های اصلی منجر به استخراج شش مؤلفه مستقل شد که در مجموع ۵۱/۸۱ درصد از واریانس کل داده‌ها را تبیین کردند. بر اساس نتایج خوشه‌بندی، متغیرهای مؤثر بر وقوع بهمن در قالب سه خوشه‌ی همگن طبقه‌بندی شدند. تحلیل تابع تشخیصی نشان داد که شاخص موقعیت نسبی شیب، عمق دره، شاخص رطوبت توپوگرافی و فاصله تا رودخانه بیشترین قدرت تمایز را بین مناطق وقوع و عدم وقوع بهمن داشتند. دقت کلی مدل ۹/۸۵ درصد و دقت اعتبارسنجی متقابل ۵/۸۴ درصد به‌دست آمد که نشان‌دهنده کارایی بالای روش‌های آماری چندمتغیره در پیش‌بینی وقوع بهمن است. https://geoeh.um.ac.ir/article_48077_0c6702e7c69ad5df080432c20b524db6.pdf

دانشگاه فردوسی مشهد جغرافیا و مخاطرات محیطی 2322-1682 15 2 2026 06 22 Identification of Threat Drivers and Assessment of Key Pressures Affecting Water Resources in Khuzestan Province, Iran شناسایی پیشران‌های تهدید و ارزیابی فشارهای کلیدی موثر بر منابع آبی استان خوزستان 141 162 48150 10.22067/geoeh.2026.97973.1654 FA پروانه سبحانی گروه محیط زیست، دانشکده منابع طبیعی، دانشگاه لرستان، خرم آباد، ایران 0000-0001-9878-3768 افشین دانه کار گروه محیط زیست، دانشکده منابع طبیعی، دانشگاه تهران، کرج، ایران 0000-0003-0641-9286 Journal Article 2026 02 21 Due to the expansion of human activities, various water resource groups in the country, especially in tropical regions, have been subjected to numerous pressures and threats. In the present study, threat drivers and assessment of key pressures affecting rivers, inland wetlands, and coastal-marine wetlands of Khuzestan province were identified using DPSIR model. According to the results, among the threatening drivers, highest priority was assigned to climate change and drought (0.143) and lowest priority was assigned to unbalanced aquaculture development (0.011). The highest sensitivity coefficient was related to Karun river (0.102) and lowest sensitivity coefficient was related to Bondoun wetland (0.014). The intensity coefficient of pressures on water resources showed that highest importance coefficient was related to pressure of "reduction of flows (flood and inflow)" and lowest importance coefficient was related to pressure of "recreation in sensitive time and place". The results showed that Karun river, with a weighting factor of 0.342, shows highest impact of identified pressures. Among coastal-marine wetlands, highest-pressure weighting factor is assigned to Khor Musa (0.0913) due to water source pollution and industrial wastewater disposal, and lowest pressure weighting factor is assigned to Bahmanshir estuarine wetland. Among the inland wetlands, the highest-pressure weighting factor is assigned to Shadegan wetland (0.297). According to the results, the most appropriate responses to control the drivers of threats and pressures from water resources in Khuzestan Province include "developing an adaptive management program", "developing an environmental management program", "developing an integrated ecosystem management program" and "developing a responsible nature tourism program".<br /><strong>Introduction </strong><br />Water resources are one of the most important natural ecosystems in tropical regions that have long played a significant role in the development of their surrounding communities. Today, in addition to meeting human needs, water resources play an important role in the tourism industry, especially water tourism and water recreational activities. Some of these water resources, such as inland and coastal-marine wetlands, are dynamic habitats that form a vital link between land and water and have major functions for storing and supplying water, protecting beaches, controlling erosion and flooding, natural water purification, a place for spawning and breeding aquatic animals, and a habitat for migratory birds. Rivers are also flowing water resources that play an important role in moderating the climate, meeting the water needs of the habitat, and a place for many aquatic species to live. Despite the importance of various natural water resources in the country, especially in tropical regions, unfortunately, many of these valuable resources have been subjected to numerous pressures and threats due to the expansion of human activities. Excessive exploitation of these water resources for agricultural and aquaculture purposes has led to the drying up and loss of habitat for many species. In addition, the construction of upstream dams, increased types of pollution from wastewater, waste, and oil spills have led to threats and increased threats to these natural ecosystems. The ecological conditions and quality of water resources depend on the characteristics of their surrounding environment, including the type and ratio of land uses, so any excessive and unplanned development leads to threats and damage to these sensitive ecosystems. Therefore, given the importance of water resources in providing ecological and valuable services for human health and other living beings, it is essential to identify and analyze the pressures on these resources and take measures to control these threatening factors. In this regard, Khuzestan Province in southwestern Iran is one of the provinces with extensive biological resources, especially in relation to running water and wetland ecosystems. Due to population growth and the development of human activities, unfortunately, these valuable ecosystems have been affected by numerous threats and pressures, including the construction of dams upstream, water source pollution, and wastewater and waste disposal. Therefore, the present study identified the threat drivers and assessed the key pressures affecting various water source groups (river, inland wetland, and coastal-marine wetland) of Khuzestan Province within the framework of the DPSIR conceptual model.<br /><strong>Material and Methods </strong><br />In the present study, the DPSIR model was used to identify the drivers of threats and pressures in each of the water resource groups (river, inland wetland, and coastal-marine wetland). This step consists of three stages, in the first stage, the description, systematic analysis, and recognition of the drivers of threats and pressures on the selected water resources were carried out based on the DPSIR model. In the second stage, the identified drivers and pressures were scored in five categories: very high, high, medium, low, and very low in the numerical range of 1, 3, 5, 7, and 9, and finally, in the third stage, the identified pressures were ranked and prioritized. The DPSIR model consists of drivers (D), pressures (P), situations (S), impacts (I), and responses (R), which respectively represent the "drivers," social, demographic, and economic growth activities that lead to changes in system behavior, and "pressures" refer to human-related activities or processes that express the extent of their adverse impacts on the environment. "Status" also refers to changes in ecological integrity, which includes changes in the physical, biological, and chemical conditions of a particular area, in other words, changes in the state of the ecosystem due to pressures on the system. "Impact" refers to human well-being in the socio-economic system, which is affected by changes in the state of environmental conditions, and finally "response" refers to political actions and programs that are guided by institutions, society, and government.<br /><strong>Results and Discussion </strong><br />As the results showed, the main drivers threatening water resources include climate change and drought, unbalanced water extraction, inappropriate water allocation and water rights, inefficient water quality management, unsustainable agricultural development, unsustainable industrial development, unsustainable urban development, unsustainable infrastructure development, unbalanced extraction of living resources, unauthorized resource extraction, unsustainable water source management, unsustainable tourism development, and unbalanced aquaculture development, which have led to the creation of 35 pressures on selected water resources in the region. According to the results, among the threatening drivers, the highest priority is assigned to climate change and drought (0.143) and the lowest priority is assigned to unbalanced aquaculture development (0.011). Among the water resources, the highest sensitivity coefficient is related to the Karun river (0.102) and the lowest sensitivity coefficient is related to the Bondun wetland (0.014). The intensity coefficient of the pressures imposed on water resources also showed that the highest importance coefficient and the highest rank were related to the pressure of "reducing flows (flood and inflow)", and in contrast, the lowest importance coefficient and the lowest rank were related to the pressure of "recreation in sensitive time and place". The results of examining the probability of the occurrence of pressure imposed on each of the water resources also showed that the Karun river, with a weight coefficient of 0.342, shows the greatest impact of the pressures identified in the region. Also, among the coastal-marine wetlands, the highest-pressure weight coefficient was assigned to Khormousa (0.0913) due to the pressures of water pollution in the basin (upstream), pollution of the water source, and industrial wastewater disposal, and the lowest pressure weight coefficient was assigned to Bahmanshir estuarine wetland. Among the inland wetlands, the highest-pressure weight coefficient was assigned to Shadegan wetland (0.297).<br /><strong>Conclusion </strong><br />According to the results obtained, in order to control the drivers of threats and pressures arising from water resources in Khuzestan Province, the most appropriate responses in this regard include "developing an adaptive management plan", "developing an environmental management plan", "developing an integrated ecosystem management plan", and "developing a responsible nature tourism plan". Accordingly, controlling these threats requires strategic planning and management strategies in the region. The results of this study can help decision-makers and planners in developing a management plan and taking strategic measures in this province by providing a range of threat drivers and the degree of vulnerability of water resources in Khuzestan province.<br /><strong>Acknowledgements</strong><strong> </strong><br />This article is based on a research project titled "Identifying Strategies for Creating, Enhancing, and Developing Water Tourism in Khuzestan Province," conducted between the Khuzestan Water and Power Organization (Office of Applied Research) and the Faculty of Agriculture and Natural Resources at the University of Tehran (Notification No. 984429/1100). The authors also deem it necessary to express their sincere gratitude to Mr. Dr. Baharlouei Bordshahi, the head of the Office of Applied Research, for his valuable assistance. به دلیل گسترش فعالیت‌های انسانی، انواع گروه‌های منبع آبی در کشور بویژه در مناطق گرمسیری تحت فشارها و تهدیدهای فراوانی قرار گرفته‌اند. در مطالعه حاضر به شناسایی پیشران‌های تهدید و ارزیابی فشارهای کلیدی موثر بر رودخانه‌ها، تالاب‌های داخلی و تالاب‌های ساحلی-دریایی استان خوزستان در چارچوب مدل مفهومی DPSIR پرداخته شد. مطابق نتایج، از بین پیشران‌های تهدیدکننده، بالاترین اولویت به تغییرات آب و هوایی و ﺧﺸﮑﺴﺎﻟﯽ (0.143) و پایین‌ترین اولویت به توسعه نامتعادل آبزی پروری (0.011) اختصاص یافت. بیشترین ضریب حساسیت نیز مربوط به رود کارون (0.102) و کمترین ضریب حساسیت مربوط به تالاب بندون (0.014) می‌باشد. ضریب شدت فشارهای وارد شده بر منابع آبی نشان داد که بیشترین ضریب اهمیت مربوط به فشار "ﻛﺎﻫﺶ ﺟﺮﻳﺎن‌ها (ﺳﻴﻼﺑﻲ و ورودی)" و کمترین ضریب اهمیت مربوط به فشار " تفرج در زمان و مکان حساس" است. نتایج نشان داد که رود کارون با ضریب وزنی 0.342، بیشترین تأثیرپذیری از فشارهای شناسایی شده در منطقه را نشان می‌دهد. از بین تالاب‌های ساحلی-دریایی بیشترین ضریب وزنی فشار به خورموسی (0.0913) ناشی از آﻟﻮدﮔﻲ منبع آبی و دفع پساب صنعتی، و کمترین ضریب وزنی فشار به تالاب مصبی بهمنشیر اختصاص یافته است. از بین تالاب‌های داخلی نیز، بیشترین ضریب وزنی فشار مربوط به تالاب شادگان (0.297) است. مطابق نتایج، از عمده‌ترین پاسخ‌های مناسب در راستای کنترل پیشران‌های تهدید و فشارهای ناشی از منابع آبی در استان خوزستان، می‌توان به "تدوین برنامه مدیریت سازشی"، "تدوین برنامه مدیریت محیط زیستی"، " تدوین برنامه مدیریت یکپارچه زیست بومی" و " تدوین برنامه طبیعت گردی مسئولانه" اشاره کرد. https://geoeh.um.ac.ir/article_48150_eb5e7787874a212cada4f3ae7be441e6.pdf

دانشگاه فردوسی مشهد جغرافیا و مخاطرات محیطی 2322-1682 15 2 2026 05 06 Comparative Analysis of Logistic Regression and Random Forest Algorithms for Landslide Susceptibility Mapping: A Case Study of the Haraz Road, Mazandaran, Iran بررسی مقایسه ای روش آماری رگرسیون لجستیک و الگوریتم یادگیری ماشین جنگل تصادفی در پهنه بندی حساسیت مکانی وقوع زمین لغزش (منطقه مورد مطالعه: جاده هراز، مازندران) 163 186 48170 10.22067/geoeh.2026.97881.1651 FA جلال سمیعا گروه جغرافیا، دانشکده علوم انسانی و اجتماعی، دانشگاه مازندران، بابلسر، ایران 0000-0002-3339-6801 Journal Article 2026 02 21 Landslides are among the most significant geomorphological hazards, posing serious threats to transportation networks and infrastructure. This study aims to compare the performance of a reference statistical method, Logistic Regression (LR), and a machine learning algorithm, Random Forest (RF), for spatial landslide susceptibility mapping along the Haraz Road corridor in Mazandaran Province, northern Iran. In this study, a balanced dataset consisting of 10,000 pixels representing landslide and non-landslide locations, together with a set of environmental conditioning factors associated with landslide occurrence, was prepared. To improve model stability, ten independent training datasets and ten independent testing datasets were generated and used in the modelling process for both approaches. Model performance was evaluated using overall accuracy, the Kappa coefficient, and the area under the receiver operating characteristic curve (AUC). Finally, the best-performing models from both approaches were selected based on the AUC values obtained from the testing datasets, and landslide susceptibility maps were produced accordingly. The results indicate that both methods exhibit considerable capability in predicting landslide occurrences. The LR model achieved AUC values of 0.90 for the training datasets and 0.89 for the testing datasets, whereas the RF model yielded corresponding values of 0.93 and 0.89. Analysis of factor importance revealed that lithology and land use are the most influential factors controlling landslide occurrence along the Haraz Road. Moreover, the spatial distribution of susceptible zones in the produced landslide susceptibility maps shows a notable degree of consistency, suggesting that the quality and characteristics of input data may play a more decisive role than the complexity of the modelling algorithm. The findings of this study can contribute to landslide hazard management and to improving safety conditions along the Haraz Road.<br /><strong>Introduction </strong><br />Landslides are among the most significant natural disasters affecting susceptible mountainous regions, causing substantial damage to infrastructure, environment, and human life. Reliable identification of landslide-prone areas is therefore essential for hazard and risk management.<br />The Haraz Road in northern Iran represents one of the most critical transportation routes connecting capital to Mazandaran province, and it is frequently affected by landslides due to susceptible geological structures and lithology, steep slopes, climatic variability, and human interferences. These conditions highlight the necessity of developing accurate and scientifically robust landslide susceptibility models for effective risk reduction and planning.<br />In recent decades, landslide susceptibility modelling has increasingly relied on quantitative modelling approaches, particularly statistical approaches. Logistic Regression (LR) has been widely used as a reference statistical technique due to its simplicity, interpretability, and solid probabilistic foundation. In recent years, machine learning algorithms such as Random Forest (RF) have gained attention because of their ability to model nonlinear relationships, handle complex interactions among variables, and achieve high predictive accuracy without strict statistical assumptions. Despite the growing popularity of machine learning approaches, there remains ongoing debate regarding whether advanced algorithms significantly outperform traditional statistical approaches when datasets are properly prepared and validated.<br />Therefore, the main objective of this research is to implement a comparative evaluation of Logistic Regression and Random Forest models for spatial landslide susceptibility modelling along the Haraz road. The specific aims include: (1) evaluating the predictive performance and accuracy of both models using the same multiple training and testing datasets, (2) identifying the relative importance of environmental conditioning factors affecting landslide occurrence, and (3) generating landslide susceptibility maps suitable for hazard management and infrastructure planning.<br /><strong>Material and Methods </strong><br />In order to model spatial landslide susceptibility using LR and RF, a comprehensive landslide inventory was prepared from field studies and existing geological data sources. Based on this inventory, an equal and balanced dataset consisting of 10,000 pixels was prepared, including 5,000 pixels with landslides and 5,000 non-landslide pixels. This balanced dataset was used to reduce classification bias and improve model reliability. Also, a set of environmental factors related to the 10000 pixels was used including lithology, land use, normalized difference vegetation index (NDVI), slope gradient, topographic position index (TPI), topographic wetness index (TWI), stream power index (SPI), plan and profile curvature, aspect-derived indices (northness and eastness), and distance-based factors such as proximity to roads, rivers, and faults.<br />To enhance model robustness and reduce uncertainty associated with random sampling, ten independent training datasets and ten corresponding testing datasets were generated using repeated random partitioning from the balanced dataset. Then, both LR and RF were implemented using identical datasets to ensure fair comparison. Logistic Regression modelling was performed using a binomial generalized linear modelling framework, while Random Forest modelling involved ensemble classification through multiple decision trees. For the Random Forest algorithm, hyperparameter tuning was performed to specify the optimal number of trees and the number of predictor variables per node of trees, thereby improving predictive performance and minimizing overfitting.<br />At the end, model performance was evaluated using multiple statistical metrics, including overall accuracy, Cohen’s kappa coefficient, and the area under the receiver operating characteristic curve (AUC). These metrics were computed for each of the ten training and testing datasets, and the mean and standard deviation values were reported to assess model stability and reliability. The best-performing model for each approach was selected based on the highest AUC values obtained from validation datasets and based on that, landslide susceptibility maps were generated.<br /><strong>Results and Discussion </strong><br />The results indicate that both LR and RF approaches achieved high predictive performance with strong classification capability. The mean AUC values of approximately 0.90 for 10 training datasets and 0.89 for 10 testing datasets in LR, demonstrate very good performance and accuracy. The RF approach achieved slightly higher training AUC values (approximately 0.93), while testing AUC values remained similar to LR (approximately 0.89), indicating comparable generalization performance between the two Approaches. Accuracy and kappa statistics also showed consistent results across the 10 training and 10 testing datasets for both models, confirming the robustness and reliability of the employed approaches.<br />Results related to the variable importance revealed that lithology and land use were the most influential factors in both approaches affecting landslide occurrence in the Haraz road. The dominance of lithology highlights the critical role of weak geological formations and weathered materials such as Shemshak formation and Scree deposits in controlling landslide susceptibility along the Haraz road.<br />The landslide susceptibility maps generated from the best-performing models in LR and RF showed similar spatial patterns, with high and very high susceptibility zones concentrated along steep slopes within regions characterized by weak lithological formations. The similarity between maps generated by LR and RF suggests that both models successfully captured the dominant controlling factors and their spatial relationships with landslides distribution along the Haraz road.<br />The relative similar predictive performance obtained by LR and RF in this study suggests that data quality, balanced sampling design, and rigorous validation procedures may play a more critical role than algorithm complexity alone in achieving reliable landslide susceptibility prediction. This finding supports previous research findings indicating that advanced machine learning methods do not always significantly outperform well-established statistical models when datasets are properly prepared. Therefore, LR remains a valuable reference approach, while RF provides complementary advantages in handling nonlinear relationships and variable interactions.<br /><strong>Conclusion </strong><br />This study demonstrates that both LR and RF models provide reliable and robust outcomes for landslide susceptibility modelling. Although RF exhibited slightly higher performance during training phase, both models yielded similar predictive accuracy on independent testing datasets, indicating similar practical applicability for landslide susceptibility assessment. The selection of lithology and land use as dominant landslide influencing factors emphasizes the importance of geological formation and human land-use practices in influencing slope instability along mountainous roads.<br />From an applied point of view, the susceptibility maps generated in this research offer valuable tools and information for infrastructure management, land use planning, and disaster risk reduction strategies. The decision makers and planners can use these maps to prioritize landslides monitoring programs, engineering interventions, and land use regulation in high and very high risk zones, thereby improving road safety and reducing socio-economic and environmental losses.<br />Future research could consider dynamic triggering factors such as rainfall intensity, temporal land-use changes, and detailed topographic changes through radar remote sensing imageries to improve predictive capability and support early warning landslide systems. Additionally, the integration of remote sensing time-series data, advanced hybrid models, and deep learning approaches may further improve the performance and accuracy of landslide susceptibility models and provide a more reliable foundation for landslide hazard and risk assessments in complex mountainous environments. زمین‌لغزش‌ها به‌عنوان یکی از مهم‌ترین مخاطرات ژئومورفولوژیکی، تهدیدی جدی برای شبکه‌های حمل‌ونقل و زیرساخت‌ها به­ شمار می روند. پژوهش حاضر با هدف مقایسه کارایی روش مرجع آماری رگرسیون لجستیک و الگوریتم یادگیری ماشین جنگل تصادفی در پهنه‌بندی حساسیت مکانی وقوع زمین‌لغزش در حاشیه جاده هراز انجام شده است. در این مطالعه و در ابتدا، مجموعه داده‌ای متعادل شامل ۱۰٬۰۰۰ پیکسل نقاط دارای زمین‌لغزش و فاقد زمین‌لغزش همراه با مجموعه عوامل محیطی موثر در وقوع زمین­ لغزش تهیه شد. به‌ منظور افزایش پایداری مدل‌ها، ۱۰ مجموعه داده آموزشی و ۱۰ مجموعه داده آزمایشی مستقل ایجاد و برای هر دو روش در فرآیند مدل سازی استفاده گردید. ارزیابی عملکرد مدل‌ها با استفاده از شاخص‌های دقت کلی، ضریب کاپا و AUC انجام شد. در انتها، بهترین مدل‌ها در هر دو روش بر اساس مقادیر AUC به ­دست ­آمده در داده ­های آزمایشی انتخاب و نقشه‌های پهنه­ بندی حساسیت وقوع زمین‌لغزش بر اساس آنها تولید شده است. نتایج نشان ­داده که هر دو روش از عملکرد قابل­ ملاحظه­ ای در پیش­ بینی زمین ­لغزش ها برخوردار می ­باشند. مقادیر AUC در روش رگرسیون لجستیک برای داده‌های آموزشی 0.90 و برای داده‌های آزمایشی 0.89 و در روش جنگل تصادفی به‌ترتیب 0.93 و 0.89 محاسبه شده است. تحلیل اهمیت عوامل موثر نشان­ داده که زمین‌شناسی و کاربری اراضی مهم‌ترین عوامل  وقوع زمین‌لغزش در جاده هراز می­ باشند. توزیع مناطق مستعد وقوع زمین ­لغزش در نقشه‌های پهنه ­بندی زمین ­لغزش نیز تطابق قابل‌توجهی را نشان داده که می­تواند بیانگر نقش تعیین‌کننده داده ­های ورودی نسبت به پیچیدگی الگوریتم مدل سازی باشد. یافته‌های این پژوهش می‌تواند در مدیریت مخاطره زمین­ لغزش و افزایش ایمنی جاده هراز کاربرد مؤثری داشته باشد. https://geoeh.um.ac.ir/article_48170_e26c55ed95245ec1d682ea371f2c9947.pdf

دانشگاه فردوسی مشهد جغرافیا و مخاطرات محیطی 2322-1682 15 2 2026 06 22 Predicting Wildfire Occurrence Risk in the Forests and Rangelands of Mazandaran Province by Comparing Regression and Machine Learning Models پیش‌بینی خطر وقوع آتش سوزی در جنگل ها و مراتع استان مازندران با مقایسۀ مدل‌های رگرسیونی و یادگیری ماشین 187 204 47981 10.22067/geoeh.2026.96904.1629 FA اردوان قربانی گروه مرتع و آبخیزداری، دانشکده کشاورزی و منابع طبیعی، دانشگاه محقق اردبیلی، اردبیل، ایران 0000-0001-7201-1225 محدثه امیری گروه مرتع و آبخیزداری، دانشکده کشاورزی و منابع طبیعی، دانشگاه محقق اردبیلی، اردبیل، ایران 0009-0004-8089-3327 Journal Article 2025 12 11 Wildfire is a major driver of terrestrial ecosystem degradation, and its impacts have intensified in the forests and rangelands of Mazandaran Province due to climate change and human activities. This study assessed wildfire susceptibility and evaluated predictive model performance using four data-mining methods: Artificial Neural Network (ANN), Generalized Linear Model (GLM), Generalized Additive Model (GAM), and Maximum Entropy (MaxEnt), applied individually and in an ensemble framework. Climatic, topographic, vegetation, and human-related variables were used as model inputs, and performance was evaluated using the Area Under the Curve (AUC), True Skill Statistic (TSS), sensitivity, and specificity. Results showed that the ensemble model achieved the highest accuracy and stability, while MaxEnt performed best among single models. Variable-importance analysis indicated that Normalized Difference Vegetation Index (NDVI) and precipitation were the dominant factors controlling the spatial pattern of wildfire susceptibility, followed by elevation and slope. The highest susceptibility occurred in areas with moderate elevation, slopes greater than 35%, moderate to dense vegetation cover, and annual precipitation below 500 mm. The ensemble map showed a widespread and relatively uniform distribution of high-risk zones, confirming the multifactorial and complex nature of wildfire occurrence and highlighting the value of multi-model approaches for preventive and sustainable natural-resource management.<br /><strong>Introduction</strong><br />Wildfires represent one of the most significant ecological hazards affecting Iran’s natural ecosystems, resulting from the combined influences of climatic conditions, vegetation characteristics, topographic complexity, and human activities. Global evidence indicates that climate change through reduced precipitation, rising temperatures, and increased soil and vegetation dryness has intensified both the frequency and severity of wildfire events. Over the past two centuries, forests worldwide have undergone extensive degradation driven by fire, and numerous studies have highlighted the pivotal roles of natural and anthropogenic factors in shaping landscape-level fire susceptibility. In Iran, wildfires have emerged a critical environmental concern, destroying thousands of hectares of forest annually and placing the country among the most affected in the region. The ecological consequences of wildfire include biodiversity loss, alterations in soil physical and chemical properties, accelerated erosion, and increased carbon emissions. Given these impacts, the application of remote sensing, geographic information system (GIS), and data-driven modelling approaches has become essential for predicting fire occurrence and generating reliable susceptibility maps. Previous studies have demonstrated the effectiveness of neural networks, statistical methods, and machine-learning algorithms in wildfire forecasting. Building upon this body of research, the present study aims to map wildfire susceptibility in Mazandaran Province, identify the most influential environmental drivers, evaluate the performance of individual and ensemble models, and determine the most accurate predictive framework. The study is based on the hypothesis that climatic factors exert a dominant influence on wildfire occurrence and that ensemble modelling provides superior predictive performance compared to single-model approaches.<br /><strong>Material and Methods</strong><br />To develop the wildfire susceptibility map, spatially explicit records of historical fire occurrences were collected through field surveys. After removing spatial autocorrelation, the fire locations were randomly divided into training and testing subsets. Absence data were generated using a pseudo-absence approach guided by preliminary MaxEnt outputs to ensure representative background conditions. A comprehensive set of environmental predictors was compiled, including topographic, meteorological, anthropogenic, and vegetation-related variables. Elevation, slope, aspect, topographic wetness index, and distance to rivers were derived from digital elevation models and remote-sensing data. Climatic variables—temperature, precipitation, relative humidity, and wind speed were interpolated across the study area using geostatistical methods based on observations from 32 meteorological stations. Anthropogenic factors included distance to roads and settlements, while vegetation conditions and moisture stress were characterized using the normalized difference vegetation index (NDVI) and Temperature Vegetation Dryness Index (TVDI). Wildfire susceptibility across forest and rangeland ecosystems of Mazandaran Province was modeled using four data-driven approaches: Artificial Neural Networks (ANN), Generalized Linear Models (GLM), Generalized Additive Models (GAM), and Maximum Entropy (MaxEnt). Model performance was evaluated using k-fold cross-validation, the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and the true skill statistic (TSS). To reduce uncertainty and improve predictive robustness, an ensemble model was developed using AUC-weighted averaging of the individual models, and wildfire susceptibility was classified into five categories. Finally, the relative contribution of each predictor was quantified using game-theoretic Shapley values to identify the key environmental drivers of wildfire occurrence.<br /><strong>Results</strong> <strong>and Discussion</strong><br />The results indicate that the models exhibited varying levels of performance in predicting wildfire susceptibility. Both the MaxEnt and ensemble models achieved excellent accuracy, with AUC values of 0.92 and 0.94, respectively. Among the evaluation metrics, the ensemble model produced the highest TSS value (0.74), reflecting a strong ability to distinguish between fire presence and absence. In terms of sensitivity, the GAM and ensemble models performed best, whereas the GLM and ensemble models exhibited the highest specificity. Variable-importance analysis revealed that vegetation, climatic, and topographic factors exerted the greatest influence on wildfire susceptibility. NDVI showed high importance in the GAM and MaxEnt models, while precipitation was identified as the dominant variable by the ANN and ensemble models. Elevation and other topographic attributes also played significant roles in fire occurrence and spread, whereas anthropogenic variables had comparatively weaker effects. Spatial patterns of wildfire risk varied among models but generally converged in identifying high-risk zones. The ensemble susceptibility map showed a widespread distribution of wildfire risk across the region, highlighting the multifactorial nature of fire occurrence. Response curves indicated that increasing NDVI is generally associated with higher wildfire probability, although this relationship weakens at very high NDVI values due to reduced fine fuel availability. Fire susceptibility was highest at moderate elevations and increased with slope, particularly on slopes exceeding 10%. An inverse relationship was observed between precipitation and wildfire susceptibility, with drier areas exhibiting the highest risk.<br /><strong>Conclusion</strong><br />The findings demonstrate that ensemble modeling provides a robust and reliable framework for wildfire susceptibility mapping, outperforming individual algorithms in terms of predictive accuracy and spatial consistency. Wildfire risk in Mazandaran Province is controlled by a complex interaction among vegetation characteristics, climatic conditions, and topographic features, underscoring the need for integrated management strategies. The susceptibility maps produced in this study offer a valuable scientific basis for land-use planning, proactive fire management, and prioritization of high-risk areas, particularly given that preventive measures are generally more cost-effective and sustainable than post-fire suppression and restoration. Overall, this study emphasizes the importance of multi-model and systematic approaches for mitigating wildfire risk in ecologically sensitive landscapes, and the ensemble-based maps developed here can serve as effective decision-support tools for regions with similar environmental conditions. آتش‌سوزی از مهم‌ترین عوامل تخریب اکوسیستم‌های زمینی است که در جنگل‌ها و مراتع استان مازندران تحت تأثیر همزمان تغییرات اقلیمی و فعالیت‌های انسانی تشدید شده است. این پژوهش با هدف تعیین حساسیت به آتش‌سوزی و ارزیابی کارایی مدل‌های پیش‌بینی خطر، از مدل‌های داده کاوی شامل شبکه عصبی مصنوعی (ANN)، مدل خطی تعمیم‌یافته (GLM)، مدل افزایشی تعمیم‌یافته (GAM) و حداکثر آنتروپی (MaxEnt) به‌صورت منفرد و اجماعی بهره گرفت. متغیرهای اقلیمی، توپوگرافی، پوشش گیاهی و انسانی به‌عنوان ورودی مدل‌ها در نظر گرفته شدند و عملکرد مدل‌ها با شاخص‌های سطح زیر منحنی (AUC)، آماره مهارت واقعی (TSS)، حساسیت و اختصاصی بودن ارزیابی شد. بر اساس نتایج، مدل اجماعی با تلفیق خروجی مدل‌های منفرد، بالاترین دقت و پایداری را در پیش‌بینی خطر آتش‌سوزی ارائه داد و در میان مدل‌های منفرد، حداکثر آنتروپی عملکرد بهتری داشت. تحلیل اهمیت متغیرها نقش غالب شاخص اختلاف نرمال شده پوشش گیاهی (NDVI) و بارندگی را در تبیین الگوی مکانی خطر آشکار ساخت و پس از آن ارتفاع و شیب نیز تأثیر معنی‌داری داشتند. بیشترین حساسیت در مناطق با ارتفاع متوسط، شیب بیش از 35 درصد، پوشش گیاهی متوسط تا متراکم و بارندگی کمتر از ۵۰۰ میلی‌متر مشاهده شد. نقشه مدل اجماعی همچنین پراکنش گسترده و نسبتاً یکنواخت خطر را نشان داد که مؤید ماهیّت چندعاملی و پیچیده پدیده آتش‌سوزی است و کارایی رویکردهای چندمدلی را در مدیریت پیشگیرانه و پایدار منابع طبیعی تأیید می‌کند. https://geoeh.um.ac.ir/article_47981_44007f34034857cf878ebe2f0fa15e06.pdf

دانشگاه فردوسی مشهد جغرافیا و مخاطرات محیطی 2322-1682 15 2 2026 06 22 Interpretability of Machine Learning Algorithms in Drought Susceptibility Mapping (Case Study: Khozestan Province, Iran) تفسیرپذیری الگوریتم‌های یادگیری ماشین در تهیۀ نقشه‌های حساسیت به خشکسالی (مطالعه موردی: استان خوزستان، ایران) 205 228 48133 10.22067/geoeh.2026.97359.1640 FA سیده زینب شگرخدایی گروه جغرافیا، دانشکدۀ ادبیات و علوم انسانی، دانشگاه رازی، کرمانشاه، ایران امان اله فتح نیا گروه جغرافیا، دانشکدۀ ادبیات و علوم انسانی، دانشگاه رازی، کرمانشاه، ایران 0000-0001-8754-1624 سیدوحید رضوی ترمه گروه نقشه برداری، دانشگاه یاسوج، یاسوج، ایران پژوهشگر دانشگاه Sejong ، سئول، کره جنوبی Journal Article 2026 02 03 In this study, a drought susceptibility map for Khuzestan Province, Iran, was produced using the Random Forest (RF) algorithm combined with the Shapley Additive Explanations (SHAP) interpretability approach. Drought occurrence data were derived from the Standardized Precipitation Index (SPI) for the period 2018–2022. A total of 18 environmental and climatic factors-including relative humidity, wind speed, evapotranspiration, minimum and maximum temperatures, elevation, slope, slope aspect, topographic wetness index (TWI), land cover, normalized difference vegetation index (NDVI), soil water content, river density, water level, sandy soil, soil bulk density, clay content, and soil texture-were used as input variables for modeling drought susceptibility. Model performance was evaluated using the Receiver Operating Characteristic (ROC) curve, yielding an Area Under the Curve (AUC) value of 0.987, which demonstrates the excellent predictive performance of the Random Forest algorithm. According to the drought susceptibility map, 31.03% of the study area falls in the “very low” class, 8.18% in “low”, 8.30% in “moderate”, 10.88% in “high”, and 41.58% in the “very high” susceptibility class. The highest spatial frequency was observed in the “very low” and “very high” categories. Based on SHAP results, elevation, maximum temperature, topographic wetness index, wind speed, and slope were identified as the most influential factors contributing to drought occurrence. This research highlights the advantages of interpretable machine learning approaches in drought susceptibility assessment, providing valuable insights for environmental planners and decision-makers to mitigate the adverse impacts of drought.<br /><strong>Extended Abstract</strong><br /><strong>Introduction</strong><br />The frequent occurrence of droughts in recent decades has posed significant challenges to the sustainability of ecosystems, agriculture, and socio-economic development. Drought is defined as a prolonged deficiency of precipitation over a specific period and represents a complex phenomenon characterized by reductions in soil moisture, declines in surface water flows, decreases in groundwater levels, and simultaneous increases in temperature. Droughts are typically classified into meteorological, agricultural, hydrological, socio-economic, and environmental types, and they exert both direct and indirect impacts, including global warming, deforestation, and urbanization. Machine learning algorithms, including SVM, BRT, RF, XGBoost, KNN, and CNN, exhibit strong capabilities in drought prediction, particularly the Random Forest (RF) algorithm, which offers high accuracy and computational efficiency. However, the “black-box” nature of these models limits physical interpretability, making the application of explainable artificial intelligence (XAI) techniques and the Shapley method essential for analyzing both the positive and negative effects of features and their interactions. Recent studies have demonstrated that integrating the Random Forest (RF) algorithm with the Shapley method enhances both the accuracy of drought prediction and the analysis of climatic, soil, topographic, and socio-economic factors. This study aims to improve the interpretability of the RF algorithm and to develop a drought sensitivity map for Khuzestan Province, with its novelty lying in the application of a spatially interpretable approach to analyze the factors influencing drought occurrence.<br /><strong>Material and Methods</strong><br />Khuzestan Province, located in southwestern Iran, covers an area of approximately 64,236 km² with elevations ranging from 0 to 3,740 m, and experiences a climate spectrum from arid to humid. Despite its major rivers and extensive water resources, the region has been increasingly affected by environmental crises due to recurrent droughts and overexploitation of water. The development of water-intensive industries, expansion of large-scale agricultural lands, and cultivation of high-water-demand crops have intensified pressure on water resources. Additionally, declining precipitation in neighboring provinces and increased groundwater extraction have exacerbated drought conditions, impacting approximately 98.7% of the province between 2012 and 2021.<br />To model drought susceptibility, a total of 19 climatic, topographic, soil, and hydrological variables were considered, including precipitation, relative humidity, wind speed, evapotranspiration, minimum and maximum temperatures, elevation, slope and aspect, topographic wetness index, land cover, vegetation index, soil water content, river density, water table level, sand and clay fractions, bulk density, and overall soil texture. Climatic data were derived from the annual averages of 22 meteorological stations across Khuzestan Province (2018–2022), while soil data were obtained from the USDA database via Google Earth Engine (GEE). Elevation, slope, slope direction, and watershed power index layers were generated from the SRTM digital elevation model with a resolution of 30 x 30 meters, and the vegetation index was generated from NDVI based on Landsat 8 images from 2019 to 2022. River density was calculated using the Line Density tool in ArcMap, and the water surface layer was extracted from well water table depth data (piezometric data from the National Water Resources Management Company). The land cover layer was generated by integrating Sentinel-1 and Sentinel-2 imagery based on the 13 classes defined by Ghorban et al. (2020). Subsequently, all layers were transformed into raster format with a spatial resolution of 250 m using the Inverse Distance Weighting (IDW) method to facilitate drought modeling.<br />The Standardized Precipitation Index (SPI) was employed to produce the drought map, as it quantifies precipitation deficits across various time scales and is recognized as a reliable indicator for drought monitoring. Drought susceptibility was modeled using the Random Forest algorithm, which is capable of handling large, multivariate datasets while providing robust predictive performance. To enhance model interpretability, the Shapley Additive Explanations (SHAP) method was applied to assess the contribution of each input variable to drought prediction. Model accuracy was assessed using the Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) indices, while the coefficient of determination (R²) was calculated as a dimensionless measure of model fit. The performance of the drought susceptibility map was further evaluated using the Receiver Operating Characteristic (ROC) curve and the Area Under the Curve (AUC) index, where values approaching 1 indicate a strong capability of the model to accurately predict vulnerable areas.<br /><strong>Results and Discussion</strong><br />In this study, the Random Forest model was employed to predict drought occurrences in Khuzestan Province based on environmental and climatic variables. Model performance was assessed using a 10-fold cross-validation approach, with 70% of the data allocated for training and the remaining 30% for testing. The results demonstrated that the model exhibited excellent performance, with an R² of 0.925, MAE of 0.035, and RMSE of 0.137, indicating a high level of accuracy in drought prediction. Interpretability analysis using the Shapley method revealed that precipitation, elevation, topographic wetness index, relative humidity, and wind speed were the most influential factors, with minimum precipitation exerting a particularly significant effect. The model output was generalized in ArcMap software and using the natural failure method, the drought sensitivity map was classified into five classes: "very low, low, medium, high, and very high." The results showed that 45.26% of the area was in the "very low" class, 4.25% in "low," 8.5% in "medium," 25.8% in "high," and 36.4% in "very high." An AUC value of 0.972 further confirmed the model’s outstanding performance. These results highlight the Random Forest model as a powerful tool for drought identification, capable of capturing complex and nonlinear relationships while processing both numerical and categorical data, thereby providing effective support for environmental management and decision-making.<br /><strong>Conclusion</strong><br />Recent drought events have inflicted substantial impacts on both the environment and human activities. In this study, a drought susceptibility map for Khuzestan Province was developed using the Random Forest model in combination with the Shapley Additive Explanations (SHAP) method for interpretability. The model demonstrated excellent predictive performance, with an AUC of 0.972. Analysis revealed that precipitation, elevation, relative humidity, the topographic wetness index, and wind speed were the most influential factors driving drought occurrence, whereas soil texture and land cover exhibited comparatively lower impacts. The most prevalent categories in the drought susceptibility map were “very low” and “very high.” Areas including Andimeshk and Dezful were classified as “very low,” while parts of Mahshahr and Ahvaz fell into the “very high” category. Drought susceptibility mapping serves as an effective tool for identifying vulnerable regions and informing management strategies, thereby enhancing the resilience of both ecosystems and human communities. در این پژوهش، با بهره‌گیری از الگوریتم جنگل تصادفی و روش تفسیرپذیری شپلی، نقشه حساسیت خشکسالی استان خوزستان تهیه شد. برای تهیه نقشه وقوع خشکسالی، از داده‌های شاخص بارش استاندارد شده در بازه زمانی 2018 تا 2022 استفاده شد. در این مطالعه از 18 فاکتور ( رطوبت‌نسبی، سرعت باد، تبخیروتعرق، کمینه و بیشینه دما، ارتفاع، شیب، جهت شیب، شاخص رطوبت توپوگرافی، پوشش اراضی، شاخص پوشش گیاهی، محتوای آب‌خاک، تراکم رودخانه، سطح آب، خاک ماسه، چگالی ظاهری خاک، خاک رس، بافت خاک) به‌عنوان متغیرهای ورودی برای مدل استفاده شد. ارزیابی کارایی مدل با شاخص منحنی مشخصه عملکرد سیستم (ROC) نشان داد که دقت پیش‌بینی 0.987 بود که بیانگر عمکرد عالی مدل در نقشه خروجی مدل جنگل تصادفی بود. براساس نتایج حاصل از نقشه حساسیت به خشکسالی، 31.03 درصد منطقه مطالعاتی در طبقه "خیلی کم"، 8.18 درصد در طبقه "کم"،8.30 درصد منطقه در طبقه "متوسط"، 10.88درصد منطقه در طبقه زیاد و 41.58 درصد منطقه در طبقه "خیلی زیاد" قرار گرفته است. بیشترین فراوانی مشاهده شده طبقات مرتبط با دو طبقه "خیلی کم" و "خیلی زیاد" بوده است. براساس روش شپلی فاکتورهایی مانند ارتفاع، دمای بیشینه، شاخص رطوبت توپوگرافی، سرعت باد و شیب بیشترین تاثیر را در وقوع خشکسالی داشته‌اند. این مطالعه مزایای استفاده از روش­های یادگیری ماشین تفسیرپذیر را نشان می‌دهد. https://geoeh.um.ac.ir/article_48133_6a145993f2368093351db6e20881c53f.pdf

دانشگاه فردوسی مشهد جغرافیا و مخاطرات محیطی 2322-1682 15 2 2026 06 22 Forecasting and Trend Analysis of Precipitation in the Navrud Watershed Based on CMIP6 Climate Models پیش نگری و روند بارش‌های حوضه آبخیز ناورود بر اساس مدل‌های اقلیمی CMIP6 229 251 48106 10.22067/geoeh.2026.97677.1647 FA موسی عابدینی گروه جغرافیای طبیعی، دانشکده علوم اجتماعی، دانشگاه محقق اردبیلی، اردبیل، ایران 0000-0002-4243-4670 طیبه بابایی اولم گروه جغرافیای طبیعی، دانشکده علوم اجتماعی، دانشگاه محقق اردبیلی، اردبیل، ایران Journal Article 2026 02 03 Accurate precipitation forecasting is crucial for water resource management and preparedness for climatic events. In this regard, this research utilized five selected CMIP6 models to project the behavior of daily precipitation for stations within the Navrud watershed. The model data were downscaled using three bias correction methods (Local Intensity Scaling, Distribution Mapping, and Power Transformation) in the CMhyd software. During the evaluation of these methods’ accuracy based on statistical indices (R², RMSE, MAE, MSE, and NSE), the Local Intensity Scaling method demonstrated the best performance in reproducing the observed precipitation patterns. Finally, the MPI-ESM1-2-HR, MRI-ESM2-0, MIROC6, CNRM-CM6-1, and EC-Earth3-Veg models were chosen as the top-performing models and were employed in an ensemble model to predict precipitation for the next 30-year period (2025–2054) under SSP2-4.5 and SSP5-8.5 scenarios. The results indicated that the spatial pattern of precipitation in the Navrud watershed during the baseline period (1985–2014) decreased from east (Kharjagil station) to west (Nav station), and the ensemble model successfully simulated this spatial precipitation pattern. However, the ensemble model’s projection for the future period (2025–2054), while maintaining the overall precipitation distribution pattern, showed that according to the SSP2-4.5 scenario, precipitation at Khelian, Kharjgil, and Nav stations will decrease by 10.9, 9.3 and 5.5% respectively, and according to the SSP5-8.5 scenario, it will decrease by 8.3%, 8.01%, and 3.4% compared to the baseline period. Furthermore, drought periods are expected to intensify from May to October. Trend analysis using the modified Mann-Kendall test and Sen’s slope estimator indicated the absence of a statistically significant trend in the daily precipitation of the stations.<br /><strong>Introduction</strong><br />Climate change, as one of the most significant global challenges, has widespread impacts on the environment, human livelihoods, and economic sustainability. This phenomenon can particularly lead to social and economic instability in regions whose economies depend on climate-sensitive sectors such as agriculture. Therefore, understanding past, present, and future climate trends, variability, and changes is essential for designing strategies to mitigate the adverse effects of this phenomenon. Among the various climate elements, precipitation, as a key hydro-climatic variable, plays a vital role in agricultural production and water resources management, directly influencing the social, economic, and environmental sustainability of regions. Climate change can cause significant fluctuations in precipitation patterns, resulting in floods, droughts, and landslides. These fluctuations also affect drinking water supply, groundwater and surface water resources, and agricultural productivity, posing challenges to water resources management. Accordingly, accurate and long-term precipitation projections form a fundamental basis for water resource planning, agricultural infrastructure design, natural disaster risk management, and the development of mitigation policies.<br /><strong>Material and Methods</strong><br />In this study, daily precipitation data from three rain gauge stations Khalian, Kharejgil, and Nav were obtained from the Gilan Regional Water Company. To simulate future precipitation, five models from the CMIP6 ensemble, including MPI-ESM1-2-HR, MRI-ESM2-0, MIROC6, CNRM-CM6-1, and EC-Earth3-Veg, were selected from the ESGFDL database. Daily precipitation simulations for the historical period were then conducted using the outputs of the selected models and three bias-correction methods Local Intensity Scaling, Distribution Mapping, and Power Transformation implemented in the CMhyd software. After applying these methods, the optimal bias-correction technique and the best-performing models were identified based on evaluation metrics, including the coefficient of determination (R²), mean absolute error (MAE), mean squared error (MSE), root mean squared error (RMSE), and Nash–Sutcliffe efficiency (NSE). Subsequently, to integrate the results of the selected models and reduce uncertainty, the Homadi model was applied using a rank-based approach and weighting of the top-performing models. Future precipitation projections for a 30-year period (2025–2054) were then generated based on the Homadi model outputs under two climate change scenarios: SSP2-4.5 and SSP5-8.5. To analyze the spatiotemporal variations in precipitation, mean monthly precipitation graphs for the studied stations were plotted, and annual precipitation totals were spatially mapped. Finally, the significance of precipitation trends during the baseline period and future horizon was assessed using the modified Mann Kendall test and Sen’s slope estimator.<br /><strong>Results and Discussion </strong><br />In this study, precipitation simulation and projection for the Navrud watershed were carried out using five selected CMIP6 climate models and three bias-correction methods. To evaluate model performance, network-averaged statistical error metrics derived from station observations were employed. The results indicated that the Local Intensity Scaling bias-correction method outperformed the Power Transformation and Distribution Mapping approaches in terms of accuracy. Accordingly, the MPI-ESM1-2-HR, MRI-ESM2-0, MIROC6, CNRM-CM6-1, and EC-Earth3-Veg models, when corrected using the Local Intensity Scaling method, showed the highest consistency with observed data and were subsequently incorporated into the Homadi ensemble model based on their respective ranks and assigned weights for further analyses. Analysis of the spatial distribution of mean annual precipitation during the baseline period (1985–2014) revealed a decreasing gradient from east to west across the watershed, with the Kharejgil station identified as the wettest location and the Nav station as the driest. The Homadi model successfully reproduced this spatial pattern for the historical period with acceptable accuracy. Future precipitation projections for the period 2025–2054 under the SSP2-4.5 and SSP5-8.5 scenarios indicate that, although the spatial pattern of precipitation is largely preserved, the absolute values of annual precipitation are expected to decline at all stations. Notably, the reduction is projected to be slightly more pronounced under the intermediate scenario than under the pessimistic scenario. Such changes may have significant implications for regional water resources and intensify water stress in the study area. The analysis of monthly precipitation suggests a relative stability in the seasonal precipitation regime across all three stations. Autumn months, particularly October and November, continue to contribute the largest share of annual precipitation, whereas summer months—especially July and August remain the driest period of the year. Nevertheless, month-to-month variations differ depending on the climate scenario, and a relative increase in precipitation is projected for some colder months. The results of the modified Mann–Kendall test applied to daily precipitation at all three stations indicate that no statistically significant increasing or decreasing trends are evident in either the historical or future periods. The very small Sen’s slope values, Z-statistics close to zero, and high p-values suggest that precipitation variability is largely random and does not follow a distinct climatic trend. Overall, the findings confirm the effectiveness of the Homadi model in representing precipitation patterns and highlight the necessity of climate-adaptive planning to ensure sustainable water resource management in the Navrud watershed.<br /><strong>Conclusion </strong><br />Analysis of climatic data for the Navrud watershed during the period 2005–2024 indicates that the region has experienced notable changes in precipitation, temperature, and soil moisture. The spatial pattern of precipitation reveals a concentration of rainfall in the eastern and northern parts of the watershed, with decreasing amounts toward the central and southwestern areas. Temporally, precipitation exhibited an increasing trend up to approximately 2020, followed by a marked decline in recent years. In contrast, temperature shows a consistent and sustained upward trend across the entire watershed, with an average increase of approximately 2.76 °C. This warming has been more pronounced in the eastern part of the basin, intensifying evapotranspiration processes and contributing to a reduction in soil moisture. Despite a relative increase in mean soil moisture, its interannual variability has been considerable, particularly during low-precipitation years. The De Martonne aridity index further characterizes the climate of the watershed as ranging from arid to semi-arid in the central and eastern regions, while indicating relatively more humid conditions in the northwestern part of the basin. The results of precipitation simulations using CMIP6 models demonstrate that the Local Intensity Scaling bias-correction method provides the best performance, and that the selected models are capable of accurately reproducing the spatial precipitation pattern of the baseline period. Projections for the future period (2025–2054) suggest that, although the overall precipitation pattern is preserved, a gradual decline in precipitation is expected at all stations, with a more pronounced reduction under the SSP2-4.5 scenario. The modified Mann–Kendall test further confirms the absence of statistically significant trends in daily precipitation. Overall, while the seasonal precipitation regime is largely maintained, reductions in summer precipitation and shifts in seasonal distribution may pose serious challenges for water resources management and climate change adaptation in the Navrud watershed. پیش‌بینی دقیق بارش برای مدیریت منابع آبی و آمادگی در برابر رویدادهای اقلیمی، امری حیاتی است. در همین راستا، این پژوهش به منظور پیش‌نگری رفتار بارش روزانه ایستگاه‌های موجود در حوضه آبخیز ناورود، از پنج مدل منتخب CMIP6 بهره گرفته شد. داده‌های مدل‌ها با بهره‌گیری از سه روش تصحیح اریبی ‏(Local Intensity Scaling، ‏Distribution Mapping‏ و ‏Power Transformation) در نرم‌افزار ‏CMhyd‏ مقیاس‌کاهی شدند. در ارزیابی دقت این روش‌ها بر اساس شاخص‌های آماری ‏R²، RMSE، MAE، MSE‏ و NSE‏ روش Local Intensity Scaling بهترین عملکرد را در بازتولید الگوی مشاهداتی بارش نشان داد. در نهایت، مدل‌های MPI-ESM1-2-HR، MRI-ESM2-0، MIROC6، CNRM-CM6-1 و EC-Earth3-Veg به عنوان مدل‌های برتر انتخاب و در ترکیب مدل همادی برای پیش‌بینی بارش دوره ۳۰ سال آینده (2054–2025) تحت سناریوهای SSP2-4.5 و SSP5-8.5 به‌کار گرفته شدند. نتایج نشان داد الگوی مکانی بارش در حوضه آبخیز ناورود طی دوره پایه (2014–1985) از شرق (ایستگاه خرجگیل) به سمت غرب (ایستگاه ناو) کاهش یافته است و مدل همادی نیز توانست این الگوی مکانی بارش را به خوبی شبیه‌سازی نماید. اما پیش‌نگری مدل همادی برای دوره آینده (2054–2025)، ضمن حفظ الگوی کلی توزیع بارش، نشان داد بر اساس سناریوی SSP2‑4.5، بارش در ایستگاه‌های خلیان، خرجگیل و ناو به ترتیب ‏10.9، 3.9 و 5.5 ‏درصد و طبق سناریوی SSP5‑8.5 به میزان ‏8.3، 8.01 و 3.4 ‏درصد نسبت به دوره پایه کاهش خواهد یافت. همچنین انتظار می‌رود دوره‌های کم‌آبی از اردیبهشت تا مهر شدت بیشتری پیدا کنند. تحلیل روند تغییرات با آزمون‌های من–کندال اصلاح‌شده و شیب سن نیز حاکی از عدم وجود روند معنادار آماری در بارش روزانه ایستگاه‌ها است. https://geoeh.um.ac.ir/article_48106_9ee0a1fb4564cf1216ed7c41b649e26e.pdf

دانشگاه فردوسی مشهد جغرافیا و مخاطرات محیطی 2322-1682 15 2 2026 06 22 Risk Analysis of the Impacts of Rising Temperature on the Phenological Cycle of Walnut in Western Iran Using CMIP6 Climate Models تحلیل مخاطرات ناشی از افزایش دما بر چرخه فنولوژیکی گردو در غرب ایران با استفاده از مدل‌های اقلیمی CMIP6 252 275 48169 10.22067/geoeh.2026.97532.1643 FA صنم کوهی گروه جغرافیای طبیعی، دانشکده جغرافیا و برنامه‌ریزی محیطی، دانشگاه سیستان و بلوچستان، زاهدان، ایران محسن حمیدیان پور گروه جغرافیای طبیعی، دانشکده جغرافیا و برنامه‌ریزی محیطی، دانشگاه سیستان و بلوچستان، زاهدان، ایران 0000-0001-7389-172X محمود خسروی گروه جغرافیای طبیعی، دانشکده جغرافیا و برنامه‌ریزی محیطی، دانشگاه سیستان و بلوچستان، زاهدان، ایران 0000-0002-2571-470X حمزه احمدی گروه آب و هواشناسی و ژئومورفولوژی، دانشکده جغرافیا و علوم محیطی، دانشگاه حکیم سبزواری، سبزوار، ایران Journal Article 2026 01 25 Climate change is one of the most significant challenges facing the agricultural sector, with profound implications for horticultural production. This study aimed to investigate the impacts of climate change on the phenological stages of walnut trees in the major walnut-producing provinces of western Iran. To this end, meteorological data from 35 meteorological stations during the baseline period (1990–2014) and outputs from 15 CMIP6 global climate models under two Shared Socioeconomic Pathway scenarios (SSP2-4.5 and SSP5-8.5) were employed for two future time horizons: the near future (2021–2040) and the far future (2041–2060). To enhance prediction accuracy and reduce uncertainty, a weighted averaging approach using the best-performing models was applied. The findings indicate a significant temperature increase across the study region, with minimum temperatures projected to rise by 1.0 to 3.3°C and maximum temperatures by 0.8 to 2.1°C in future periods. These thermal changes will lead to substantial shifts in the walnut phenological calendar. Budbreak will occur 2 to 14 days earlier and flowering 3 to 15 days earlier compared to the baseline period. The onset of fruit growth will advance by 3 to 11 days in the near future and 8 to 17 days in the far future, while fruit maturation will occur 6 to 17 days and 16 to 29 days earlier, respectively. Consequently, the harvest date will be advanced by 7 to 33 days. In contrast, the winter dormancy period (9 to 29 days) and leaf senescence (8 to 37 days) will be delayed. Spatially, the most pronounced phenological shifts are projected for the cold regions in the north and east, while the smallest changes are expected in the warmer southern and southwestern areas. These uneven and premature alterations pose serious risks, including increased vulnerability to late spring frosts, heat and water stress during critical growth stages, and disruption to the product supply chain. The results underscore the urgent need to develop adaptive strategies, such as utilizing late-flowering cultivars, to mitigate the adverse impacts of climate change on walnut orchards in the region.<br /><strong>Introduction </strong><br />Climate change, driven by rising greenhouse gases, poses a significant threat to global ecological balance, with agriculture being particularly vulnerable due to its direct climate dependency. Horticultural and nut crops like walnuts are especially sensitive to thermal fluctuations, which can jeopardize their yield and quality. Consequently, analyzing the impact of climate change on plant phenology is a crucial step for optimizing resource use and mitigating future risks. Existing literature, including regional studies in Iran and globally, underscores this vulnerability, demonstrating that rising temperatures advance key phenological stages—such as flowering and fruit maturation—in various tree species, including walnuts, pistachios, and apples. These shifts can disrupt dormancy cycles, increase exposure to late frosts, and alter water requirements.<br />However, a research gap exists in conducting a comprehensive, high-resolution assessment of climate change impacts across all major phenological stages of walnut trees specifically in western Iran, utilizing the latest CMIP6 climate models and SSP scenarios. Previous studies have often focused on specific regions, stages, or older model generations. This study aims to address this gap by systematically evaluating the effects of projected climate change on the seven key phenological stages of walnut trees in this vital cultivation area. It seeks to determine the extent and spatial variability of changes in the timing and length of these stages under different future scenarios. The research is guided by two primary questions: 1) Does the selection of different climate models influence the projected outcomes? and 2) What specific effects will climatic changes have on the duration and scheduling of the phenological stages of walnut trees in western Iran? By employing an ensemble of CMIP6 models under SSP2-4.5 and SSP5-8.5 scenarios for near- and far-future periods, this study provides a nuanced, forward-looking analysis intended for agricultural planners, policymakers, and environmental researchers to inform adaptive strategies and safeguard walnut production against imminent climatic threats.<br /><strong>Material and Methods</strong><strong> </strong><br />This study employed an integrated observational and simulation-based design to assess the impact of future temperature changes on the phenological stages of walnut trees in five major walnut-producing provinces of western Iran (Kurdistan, Hamadan, Lorestan, Kermanshah, and Ilam). The foundational data consisted of daily minimum and maximum temperature observations from 35 meteorological stations for the 1990–2014 baseline period. For future projections, data from 15 CMIP6 climate models under two scenarios, SSP2-4.5 and SSP5-8.5, were used for two future periods: 2021-2040 and 2041-2060. Following the extraction and spatiotemporal alignment of the model and observational data, the performance of each climate model was evaluated against the baseline observations using multiple statistical indices (CC, RMSE, NRMSE, ANMBD, AARD). Subsequently, the models were weighted by applying normalization, ranking via the Pomeral-Romero method, and entropy-based weighting to create a weighted multi-model ensemble, thereby reducing projection uncertainty. In the next phase, the daily temperature data (both observed and projected) were converted to Julian dates. By applying established temperature thresholds and growing degree-day requirements for each of the seven walnut phenological stages, the occurrence date for each stage was calculated for all stations and time periods. Finally, the quantitative results of these temporal shifts were spatially analyzed and mapped using a GIS environment.<br /><strong>Results and Discussion </strong><br />Analysis of walnut tree phenological stages during the baseline period (1990-2014) revealed that the growing season in the region varies from approximately March to November (9 months). Bud break occurs from mid-March in warm, low-altitude areas (e.g., Ilam) to early May in cold, high-altitude regions (e.g., Kurdistan, Hamadan), indicating a 50-day difference in the start of the growing season. Flowering occurs about 10 days after bud break, spanning from late March to mid-May. Other phenological stages also occur later from south to north and from lowlands to highlands; for instance, harvest takes place from late August in warm areas to mid-October in colder zones. Leaf senescence and the onset of dormancy begin earlier in cold regions (mid-October) and later in warm regions (mid-November).<br />Following the selection and weighting of the best-performing CMIP6 models for temperature simulation, projections were made under the SSP2-4.5 and SSP5-8.5 scenarios. The results indicate a temperature increase across the entire region, but this warming is markedly more intense in cold, high-altitude areas (e.g., Saqqez, Zarrineh) compared to warm, low-altitude zones (e.g., Dehloran). This asymmetric warming directly impacted the walnut phenological cycle.<br />After projecting the temperature changes, the changes in the phenological stages of walnut trees in two future periods (2021–2040 and 2041–2060) were investigated, based on the temperature changes obtained under the SSP5-8.5 scenario, which showed relatively greater temperature variations. In the future periods (2021-2040 and 2041-2060), under the SSP5-8.5 scenario, all growth stages are projected to occur earlier. Bud break will advance by 2 to 14 days, flowering by 3 to 15 days, fruit set by 3 to 17 days, and fruit maturation by 6 to 29 days compared to the baseline period. Consequently, the harvest date will also shift earlier by 7 to 33 days. The most significant shifts were observed in the cold, high-altitude areas of the north and east (Kurdistan and Hamadan provinces), while the smallest changes occurred in the warm southwestern parts. Conversely, the stages of leaf senescence (delayed by 8 to 37 days) and dormancy onset (delayed by 9 to 29 days) are projected to occur significantly later. These asynchronous shifts in the agricultural calendar for walnuts entail risks such as increased vulnerability to late spring frosts and disruptions to harvest scheduling and market logistics.<br /><strong>Conclusion </strong><br />This study assessed climate change impacts on walnut phenology in western Iran. By employing a weighted multi-model ensemble of 15 CMIP6 models under the SSP2-4.5 and SSP5-8.5 scenarios, the projections indicate that minimum temperatures will increase by 1–3.3°C and maximum temperatures by 0.8–2.1°C by the middle of the 21st century. These changes will significantly advance key phenological stages: bud break by 2–14 days, flowering by 3–15 days, and harvest by 7–33 days, with the most pronounced shifts occurring in colder, higher-altitude northern and eastern regions. Conversely, leaf senescence and dormancy onset will be delayed by 8–37 days. The earlier onset of sensitive stages, particularly flowering, heightens the risk of damage from late spring frosts, while higher growing-season temperatures increase the potential for heat stress and greater irrigation demand. These findings underscore the urgent need for adaptive strategies in orchard management to mitigate future climatic risks to walnut production. این مطالعه تأثیر تغییرات اقلیمی بر چرخه رشد درخت گردو در غرب ایران را بررسی کرد. برای این کار، از داده‌های هواشناسی ایستگاهی و پیش‌نمایی ۱۵ مدل اقلیمی تحت دو سناریوی انتشار برای دوره‌های آینده استفاده شد. دقت پیش‌نماییها با میانگین‌گیری وزنی مدل‌های برتر افزایش یافت.نتایج حاکی از افزایش قابل توجه دما در منطقه مورد مطالعه است، به‌طوری‌که دمای کمینه و بیشینه بترتیب حدود ۱ تا 3/3 و 0/8 تا 1/2 درجه سلسیوس در دوره‌های آینده افزایش خواهد یافت. این تغییرات، موجب جابجایی تقویم فنولوژیکی گردو خواهد شد. به‌طور مشخص، زمان رخداد جوانه‌زنی ۲ تا ۱۴ و گلدهی ۳ تا ۱۵ روز زودتر از دوره پایه رخ خواهد داد. همچنین، آغاز رشد میوه در آینده‌نزدیک ۳ تا ۱۱ و در آینده‌دور ۸ تا ۱۷ روز، و تکمیل نمو میوه به ترتیب ۶ تا ۱۷ و ۱۶ تا ۲۹ روز زودتر آغاز می‌شود. در نتیجه، زمان برداشت محصول نیز ۷ تا ۳۳ روز پیشی خواهد گرفت. در مقابل، دوره رکود و خواب زمستانه درخت ۹ تا ۲۹ و نیز زمان خزان برگ‌ها ۸ تا ۳۷ روز دیرتر آغاز خواهد شد. تغییرات در مراحل رشد درختان در مناطق سردسیری منطقه بیشتر و در مناطق گرمسیری منطقه (جنوب و جنوب‌غرب) کمتر خواهد بود. https://geoeh.um.ac.ir/article_48169_286142eb41f5c745b97e5dd10782e008.pdf