مجله مهندسی و مدیریت آبخیز
سال شانزدهم شماره 3 (پاییز 1403)

Landslides and sinkholes damage social, economic, and natural infrastructure. These processes have direct and indirect impacts on important infrastructure, including residential areas, and influence land use change and migration from rural to urban areas. Sinkholes and landslides occur when parts of a soil collapse mainly in more gentle or steeper slopes, which are often triggered by intensive rainfall. One of the main goals in sustainable land management is the identification and control of natural disasters, which on the one hand leads to the quantitative and qualitative improvement of production in the long term, and on the other hand, maintains the quality of the soil and prevents soil degradation. In order to manage better and more stable, it seems necessary to know how to change and identify different forms of erosion such as sinkholes and landslides. Sinkholes and landslides occur when parts of a soil collapse mainly in more gentle or steeper slopes, which are often triggered by intensive rainfall. 

Recent advances in acquiring images from unmanned aerial vehicles (UAV) (UAV) and deep learning (DL) methods inherited from computer vision have made it feasible to propose semi-automated soil landform detection methodologies for large areas at an unprecedented spatial resolution. In this study, we evaluate the potential of two cutting-edge DL deep learning segmentation models, the vanilla U-Net model, and the Attention Deep Supervision Multi-Scale U-Net model, applied to UAV-derived products, to map landslides and sinkholes in a semi-arid environment, the “Golestan Province” (north-east Iran).

Landslides: The performance of the U-Net model shows that it has fewer false positives, but at the same time, it has missed many landslide cells. Meanwhile, the ADSMS U-Net model has performed better in detecting landslide cells, but it attributed many cases to incorrect predictions (which is explained by the low accuracy score). The best F1 score achieved for the ADSMS U-Net model is 0.68. Sinkholes: For all band combinations, the performances of ADSMS U-Net are better than those of the traditional U-Net model. The best overall scores by ADSMS U-Net were obtained when trained on the ALL data. Regarding the effectiveness of the various combinations evaluated in this study, we can observe the contradictory behaviors of the models. The traditional U-Net achieves the best performance using the RGB optical combination, while the ADSMS U-Net can leverage topographic derivative information and optical data, showing the best results with the ALL combination. Moreover, it is evident that the DSHC data alone provides the worst results for both models. In overall, the results show that the ability of ADSMS U-Net to predict landslides is closer to the ground reality compared to U-Net. This model identifies most of the landslides in the test sections. Also, for all combinations of sinkhole bands, ADSMS U-Net performs better than the U-Net model. The best overall scores were obtained by ADSMS U-Net when trained on ALL data.

Since this kind of soil erosion is the main origin of some major soil erosion including gully initiation and extension, applying new technology namely, UAV and deep learning is highly important and recommended. Our framework can successfully map landslides in a challenging environment (with an F1-score of 69 %), and topographical derivates from UAV-derived DSM decrease the capacity of mapping sinkholes and landslides of the models calibrated with optical data. Future research could explore the use of such an approach to map landslides and sinkholes over time to assess time-based changes in the formation and spread of natural hazards.

Salmas Plain represents one of the most critical areas in the country experiencing subsidence. In general, various factors cause land subsidence, but in many areas, the excessive extraction of ground water from aquifers causes land subsidence. The increasing use of ground water, especially in the sites that are accumulated with alluvial deposits, shallow sea or unconsolidated lake, leads to subsidence or collapse of the land. With the excessive extraction of ground water, the water level of the aquifer decreases and the hydrostatic pressure decreases, which makes it possible for the land to subside gradually. Subsidence in plains mostly occurs due to this factor, namely excessive groundwater extraction and compaction of clay and silt layers between aquifers. In this case, even if the water table level rises again, the land cannot return to its original level.

In this study, the susceptibility of land subsidence in Salmas Plain was investigated using layers of influential factors in subsidence with ArcGIS software and fuzzy logic. In the first stage, statistical information on some factors causing subsidence, including groundwater level decline, well extraction rate, aquifer storage coefficient, transmissivity coefficient, precipitation, DEM map, soil texture, and bedrock depth, was collected and raster maps of each of these factors at the aquifer level were prepared. In the next stage, fuzzy layering was performed using fuzzy membership functions based on the impact of decreasing or increasing each of these factors on land subsidence. Subsequently, the maps were combined using fuzzy operators (Gamma OR, AND, SUM, PRODUCT) to obtain a unified map of aquifer subsidence susceptibility. Finally, to select the best combination of operators, the results were compared and evaluated with field observation data and the ROC curve performance index.

The results showed that the OR operator had the lowest conformity with observed subsidence in the area with an AUC of 0.693. Gamma operators with an AUC above 70% had the highest overlap or conformity with observed subsidence in the plain. In this study, the Gamma 0.9 operator was selected as the best fuzzy operator with an AUC of 0.805. The results indicate that the eastern part of the aquifer is critical in terms of subsidence. Approximately 25% of the total area of Salmas Plain, equivalent to 93 square kilometers, has subsidence with very high susceptibility.

Based on the results obtained, it can be said that although the AUC value of the fuzzy operator sum is higher, the Gamma operator with a value of 0.9 has the highest conformity with the ground reality on the fuzzy map, even though it has a lower AUC value. It is essential to mention that the minimum operator AND and Product create a region with low susceptibility, while the maximum operator OR and SUM maximize the susceptible area. They cannot achieve satisfactory performance in preparing a subsidence susceptibility map. Here, they have only been used to demonstrate the inefficiency of fuzzy operators in maximizing or minimizing subsidence susceptibility.

A comprehensive examination of natural hazards in Iran highlights the country's susceptibility to extensive damage from various natural crises. Iran's unique spatial structure has made it one of the world's most vulnerable regions to environmental hazards. This research reviews the application of “resilience” in studies of natural hazards, including floods, drought, and land degradation (single-hazard studies) as well as studies that combine these hazards (combined studies) across different regions of Iran.

This study utilized a conceptual research methodology, performing a systematic review of related research documents published in Iran, such as journal articles, dissertations, and theses. The databases used included the "Scientific Information Center of Academic Jihad (SID)", "Iran’s Research Institute of Science and Information Technology (Irandoc)", "Information Bank of Iran's Publications (Magiran)", "Knowledge Reference (Civilica)", and "Google Scholar". While previous studies have often examined resilience within an established analytical framework (concept analysis), this study employed a conceptual methodology aimed at representing knowledge and analyzing data from multiple disciplines. This approach helps clarify meanings and expand operational definitions. The data were analyzed thematically, enabling a more objective examination of resilience across various scientific fields.

The first study on resilience in Iran was conducted in 1988 in a dissertation at Tarbiat Modares University, and the concept appeared in articles in "Hakim Research Journal" in 2005. However, resilience studies focusing on natural hazards, especially in rural areas of Iran, are relatively recent, mostly emerging since the early 2010s. Based on the search criteria, 1,742 scientific documents were identified, with 57 relevant articles included in the review. Most documents were found in "Google Scholar," while the fewest were retrieved from "SID." The highest number of studies was conducted in 2017 (12 studies), and the lowest in 2012, 2021, and 2022 (one study each). Data from most studies were analyzed using statistical tests with SPSS, PLS, and AMOS software, while ArcGIS was commonly used for spatial data display and zoning to prioritize study areas. Most resilience studies focused on drought (36.8%), while the fewest focused on land degradation (10.5%).

The results show that, in flood-related research, the social dimension (29.4%) received the most attention, whereas the institutional dimension (20.6%) received the least. In drought research, the economic dimension (35.4%) received the most attention, while the institutional dimension (16.7%) received the least. In land degradation studies, the physical dimension (33.4%) was the most frequently examined, with other dimensions receiving 22.2% of the focus. Overall, the physical dimension (30.8%) and social dimension (20.5%) received the most attention. By identifying key resilience components, these findings can improve crisis management, reduce damages, and support the planning of development and educational projects in Iran.

Humanity is facing many environmental challenges. Natural disasters are among these problems, causing the death and injury of hundreds of thousands of people and rendering millions homeless worldwide every year. Geomorphological hazards, particularly mass movements and landslides, are considered some of the most potentially harmful phenomena. Mass movement refers to the outward or downward movement of domain-forming materials under the influence of gravity. These movements are primarily triggered by gravity, natural factors such as heavy rainfall, earthquakes, and soil saturation with water, as well as human activities like deforestation and improper engineering operations. According to the 2012 World Natural Hazards Report, landslides were ranked among the seven most dangerous natural disasters globally. The occurrence of natural hazards, including landslides, exerts considerable pressure on the economic development of countries, especially in developing regions, with financial damages hindering economic growth and prosperity. Iran, with its mountainous terrain, high tectonic and seismic activity, and diverse geological and climatic conditions, has created natural conditions conducive to a wide range of landslides.

The aim of this study is to spatially model landslide susceptibility using machine learning techniques, including random forest, support vector machine, and enhanced regression tree, in Razavi Khorasan province. Initially, the distribution map of landslides in the region was prepared through field visits and data from the national landslides database. In the next step, 70% of the identified landslides were used for model development, while 30% were reserved for model evaluation. The information layers for altitude, slope, slope direction, distance from waterway, waterway density, distance from roads, distance from faults, land use, vegetation index, surface curvature, profile curvature, precipitation, and selected lithological units were prepared and mapped using ArcGIS.

Prioritization of the factors affecting landslide occurrence using the random forest model showed that precipitation and altitude had the greatest impact on landslides in the study area. Additionally, the evaluation of the machine learning models using the relative operating characteristic (ROC) curve indicated that the landslide potential map generated by the random forest method had the highest accuracy (0.97). Based on this map, more than 25% of the area was classified into high and very high-risk zones.

This map can assist environmental planners in construction projects and help prevent land-use changes and construction in high-risk areas. Additionally, public awareness campaigns can reduce harmful human activities in these zones. While controlling landslides may not always be feasible or is often very expensive, proper management can help mitigate or reduce risks. By identifying the key factors in mass movement occurrences and zoning the areas accordingly, it is hoped that this research will contribute to the development of effective risk management plans and reduce the damage caused by landslides.

Today, soil erosion in particular and soil degradation in general as a result of human activities have been raised as a social problem, and the role of the human factor in the emergence and acceleration of the soil degradation process has been clarified in many fields. Obtaining accurate statistics and information about soil erosion and sediment yield in the watersheds is necessary for the implementation of soil conservation programs and methods for determining the resistance to soil erosion and reducing sediment yield. Due to the lack of data about soil erosion and sediment yield in many watersheds of Iran, the use of appropriate empirical methods is inevitable to erosion estimation, sediment yield and especially sediment delivery ratio.

For this purpose, in the present study, empirical models for estimation of the sediment delivery ratio based on easy-to-measure variables and three models of SATEEC, InVEST and WaTEM/SEDEM were used in the Khamsan representative-paired watershed, western Iran. The WaTEM module of the WaTEM/SEDEM model is based on the RUSLE model for estimating soil erosion, and in the SEDEM module is based on the performance of effective physical factors in the sediment transport equation. In many sources, the 137Cs method is mentioned as the only available and reliable method for measuring the components of the sediment budget, including total erosion, total sedimentation, net erosion (sediment yield) and sediment delivery ratio, especially at the hillslope and sub-watershed scales. To evaluation of the used models, the results of calculating the sediment delivery ratio using the 137Cs method obtained from previous researches were used, which are 25.61% and 58.94% for the entire watershed and the average of 15 sub-watersheds, respectively, and its accuracy is based on the observed sediment data at the outlet of the watershed has been confirmed.

Among the studied sediment delivery ratio models, Renfro and Waldo (1983), Williams and Brendt (1972), Roehl (1962) considering one hour excess precipitation, Walling (1983), Ferro (1995), Vanoni (1975) and USDA (1972) provided the closest estimates (±10%) to the 137Cs method for the whole watershed scale and the Renfro (1975), USDA (1972), USDA-SCS (1979), SATEEC and Roehl (1962) considering one hour excess precipitation for the sub-watershed scale provided the closest estimates (±10%) to the 137Cs method for the sub-watershed scale and were selected as suitable sediment delivery ratio models for Khamsan representative-paired watershed. Also, these models can be used to estimate the sediment delivery ratio in watersheds similar to Khamsan representative-paired watershed.

In the sub-watersheds scale, the equations estimations based on easy-to-measure variables except the Roehl (1962) method - excess rainfall of 1.0 h was significantly lower than the SATEEC model, on the other hand, the estimations of all the methods except the Roehl (1962) method - excess rainfall of 0.1 h was significantly higher than WaTEM/SEDEM and InVEST models. This issue emphasizes the importance of considering excess precipitation conditions to estimation of the sediment delivery ratio in the presented equations by Roehl. Also, at the scale of the whole watershed, the estimations of the equations based on easy-to-measure variables, except for the Mutchler and Bowie (1975) method, were lower than the results of the SATEEC model. The great difference of the investigated methods performance in the two scales of the sub-watershed and the whole watershed is due to the undeniable effect of the middle plain of the Khamsan watershed in the drastic reduction of the sediment transportation from the sub-watersheds to the outlet of the watershed.

The rapid growth of cities and the process of industrialization have created numerous environmental problems across many parts of the world. It is essential for planners and managers to be aware of changes in land cover and land use over extended periods to evaluate and predict the impacts caused by these changes. Remote sensing is an effective tool for monitoring land use changes in urban areas and their surroundings. Tehran has expanded significantly over the last few decades due to population growth and migration, leaving substantial effects on the surrounding environment. Consequently, this study presents a model based on the decision tree algorithm to classify and monitor land use changes using images from TM and MSS sensors in the western region of Tehran between 1975 and 2011.

In this study, one MSS sensor image and three TM sensor images from the Landsat satellite, all taken in June, were used along with ancillary data, specifically a digital elevation model extracted from the 1:25000 topographic map of the Mapping Organization. After pre-processing, land cover indices, including vegetation index, DT method, and its combination with the maximum likelihood classification method, were used to extract land use classes. The accuracy of the classified images obtained from the DT was evaluated using the kappa coefficient and overall accuracy, and finally, the changes in different land use classes over time were calculated using the image comparison method.

According to this study's findings, the overall classification accuracy for 2011 is 82%. The results of change monitoring indicate a positive and increasing trend in the density of built-up land over the 36-year period, while other land types have decreased. The density of the built-up land class in 1975, with an area of 2166 hectares (equivalent to 8%), increased to 8125 hectares (29%) by 2011. In total, the percentage of relative change is 21%, equivalent to 5959 hectares. By examining the land use changes in the west of Tehran from 1975 to 2011, shown in the maps, it is evident that urban development and increased demand for various services, coupled with a lack of adequate space, have led to the destruction of green spaces in the western part of Tehran, replaced by other land uses.

This research aimed to monitor land use/cover in the west of Tehran with high classification accuracy using a model based on the DT algorithm combined with the maximum likelihood classification method. Multi-temporal satellite images from the Landsat satellite’s TM and MSS sensors, along with ancillary data, were used to conduct the research. After preparing a land use map for each period, a map depicting land cover and land use changes was extracted. The results of this research indicate that remote sensing data combined with classification techniques have a high capability to extract various types of land use maps and evaluate land use changes. Moreover, Landsat’s MSS and TM sensor data prove to be suitable and cost-effective tools for depicting and analyzing land use/cover changes over time. Additionally, the findings highlight that using a branching or multi-stage method for classifying satellite images offers advantages such as reduced processing time, improved accuracy in small classes, and the ability to use different data sources, feature sets, and algorithms at each decision-making stage.

The concentration time of catchments is one of the most important and common effective features in hydrological studies, particularly in determining the flow discharge for designing watershed management projects. Most of the catchments in the world especially in Iran were not equipped with hydrometric stations, and project managers are forced to use traditional empirical models to estimate concentration time and peak flow. The review of previous studies shows that experimental models for estimating concentration time have unfavorable results due to the change of environmental conditions outside the place where the model is presented. On the other hand, there is not enough information about the effectiveness of experimental models for estimating concentration time in many catchments in Iran, especially in semi-arid areas. The purpose of this study is to evaluate the accuracy of some experimental models for estimating concentration time in the sub-basins of the semi-arid region of the northwest of the country and to identify its determining factors.

This study was conducted in eight sub-basins including Alanagh, Ordakloo, Shekaralichay, Shiramin, Kurjan, Kalaleh and Livar from Urmia Lake and Araz River basins in Northwest Iran. Meteorological and hydrometric data were obtained from the Natural Resources of East Azerbaijan and stations belonging to the Ministry of Energy. The characteristics of the basin such as area, length, slope, height and shape were determined through field studies and drawing maps in the GIS platform. The concentration time was calculated using the hydrograph of the flows in the statistical period of 30 years (from 1367 to 1397) and it was estimated through six experimental models including Kirpich (1940), Kerby (1959), Chow (1962), Federal Aviation Administration (1972), Bransby-Williams (1980) and Ventura (2007). The relationship between concentration time and catchment characteristics was investigated by correlation matrix, Pearson's method. Nash-Sutcliffe efficiency coefficient, average error and root mean square error were used to evaluate the accuracy of the models.

According to the results, Shekaralichay sub basin has the shortest (66 minutes) and the Kalaleh sub basin has the longest concentration time (132 minutes). Bransby-Williams model had the lowest error (6.8 %) and the highest efficiency coefficient (73%); while the estimation error (36.2 %) and the Nash-Sutcliffe efficiency of Federal Aviation Administration model were 36.2% and-14.4% respectively. The slope was the most important main factor on the estimation of concentration time of the assessment in the Kirpich model (r= 0.83), Chow (r= 0.82) and Bransby-Williams (r= 0.73). Federal Aviation Administration model (1972) and Ventura model (2007) have a weak estimate in sub-basins with low slope and length.

The results showed that among the physical characteristics of the basin, the area, slope and length of the sub-basin play a more important role in changes in concentration time. This study showed that the slope percentage of the basin is the most important factor in reducing concentration time, peak discharge and increasing the speed of flooding in the studied sub-basins, so it is suggested to use soil protection plans in order to increase the concentration time for sub-basins that have a higher slope percentage. The evaluation of concentration time estimation models in eight catchments showed that the Bransby-Williams (1980) model with an average error of 6.80% and Nash-Sutcliffe efficiency coefficient of 73% provides the best estimation among others, so the use of this model in similar basins which do not have measuring stations, it is suggested.

One of the natural disasters that threaten residential areas and roads in mountainous regions is mountain falls, including rockfalls, avalanches, and landslides. Avalanches carry large amounts of snow, rocks, ice, and debris downstream in mountainous areas. The occurrence of avalanches is a significant natural hazard that results in considerable human and financial losses, making the study of factors influencing avalanches and their simulation crucial for managing this phenomenon.

In this study, the factors affecting snow avalanches (both terrestrial and meteorological) were examined using the RAMMS simulation model in the Central Alborz Velayat Rood (Dizin Road), Alborz Province. Topographic and geomorphological factors, such as slope, aspect, curvature, topographic position index (TPI), terrain roughness index (TRI), and topographic wetness index (TWI), were analyzed using a digital elevation model (DEM) with a 6×6 cm² pixel size obtained by drone. Meteorological factors, including rainfall, temperature, and wind, were also considered. The RAMMS simulation model was then used to estimate avalanche components such as speed, pressure, and height within the study area.

The results indicated that slope and aspect, as topographic indicators, have a significant impact on snow avalanche formation and occurrence. The largest portion of the region, covering 5.7 hectares (54.6% of the study area), with a northeast aspect and slopes of 60 to 120%, was identified as having the highest avalanche potential. Additionally, the RAMMS simulation model results showed that the average and maximum avalanche speeds in the region were 5.3 m/s and 16 m/s, respectively. The average effective avalanche pressure was 7 kPa, with a maximum of 45 kPa. The estimated avalanche height indicated that the average avalanche height in the runout area (residential areas) was 4.5 meters, with a maximum height of 10 meters, categorizing it as a large avalanche.

Understanding avalanches and their dynamic characteristics is essential for predicting and controlling this hazardous natural phenomenon. Identifying avalanche types (wet, slab, or powder) can greatly assist experts in managing and proposing control methods. In this study, maps, meteorological data, and geomorphological parameters such as curvature, TPI, TRI, and TWI, along with field observations, were used to identify accumulation areas, track zones, and runout zones. The study identified the key factors influencing avalanche occurrence in the region, including high slopes (60-120%), slope orientation (north and northeast), and climatic factors such as precipitation and temperature. The average avalanche height in the runout area (residential areas) was 4.5 meters, with a maximum of 10 meters. This study indicates a high potential for avalanches and associated damage in the area, underscoring the need for management and control programs to mitigate possible harm.