Estimation of radon as an earthquake precursor: A neural network approach

An artificial neural networks (ANN) approach combined with Fourier Transform based selection of time period in the time series Radon Emission Data has been presented and shown to improve event prediction rates and reduce false alarms in Earthquake Event Identification over the traditional multiple linear regression techniques. The paper presents a neural networks system using radial basis function (RBF) network as an alternative to traditional statistical regression technique in isolating Radon Emission Anomaly caused by seismic activities. The RBF model has been developed to accept and predict earthquakes events based on a known data set of Radon Emanation, Metrological parameters and actual earthquake events. Subsequently, the model was tested and evaluated on a future data set and a prediction rate of 87.8%, if a reduced false alarm was achieved, the results obtained are better than the traditional techniques.     

Seismic-induced rockfalls and landslide dam following the October 30, 2016 earthquake in Central Italy

On October 30, 2016, a seismic event and its aftershocks produced diffuse landslides along the SP 209 road in the Nera River Gorge (Central Italy). Due to the steep slopes and the outcropping of highly fractured and bedded limestone, several rockfalls were triggered, of which the main event occurred on the slope of Mount Sasso Pizzuto. The seismic shock acted on a rock wedge that, after an initial slide, developed into a rockfall. The debris accumulation blocked the SP 209 road and dammed the Nera River, forming a small lake. The river discharge was around 3.6 m³/s; the water overtopped the dam and flooded the road. By a preliminary topographic survey, we estimated that the debris accumulation covers an area of about 16,500 m², while the volume is around 70,000 m³. The maximum volume occupied by the pre-existing talus mobilized by the rockfall is about 20% of the total volume. Besides blocking the road, the rockfall damaged a bridge severely, while, downstream of the dam, the water flow caused erosion of a road embankment. A rockfall protection gallery, a few hundred meters downstream of the dam, was damaged during the event. Other elastic nets and rigid barriers were not sufficient to protect the road from single-block rockfalls, with volumes around 1–2 m³. Considering the geological and geomorphological conditions, as well as the high seismicity and the socioeconomic importance of the area, a review of the entire rockfall protection systems is required to ensure protection of critical infrastructure and local communities.     

Suitability of airborne laser scanning for the assessment of forest protection effect against rockfall

Rockfall simulation models are now able to quantify the protective effect of forest with the integration of rock impacts on trees. Those models require spatially explicit forest characteristics which are costly to acquire in operational conditions. The present study compares rockfall simulation results obtained with different forest input data sources: field data with different levels of spatial detail and two methods based on airborne Lidar data. Three different forest stands are tested with several virtual terrain configurations. When rockfall energies are below 200 kJ, the forest protection effect is significant. For higher energies, it also exists but it is minor compared to the effects of topography and rock volume. For all forest input data sources, the estimated rockfall intensity is within ?13 and 16 % of the reference value, whereas the frequency is generally overestimated. Both Lidar methods yield a satisfactory forest protection effect evaluation, but single tree detection tends to underestimate it. Improvements are possible regarding the spatial heterogeneity of stem density and the diameter distribution by tree species.     

青海玉树断裂带地震落石的地震地质意义

2010年青海玉树Ms 7.1级大地震引发了一系列次生地质灾害,其中地震落石是除地震滑坡外沿断裂带及其邻侧最常见的现象。对玉树震区落石的调查发现,该区多处存在非常典型的多期地震落石分布现象,指示该区地震落石的发育与其他古地震现象类似,具有多期性和一定的原地复发性。实地调查表明,该区地震落石分布的主要特征为:多集中发育在活动断裂带附近的陡峭基岩斜坡下方,分布零散,且滚动较远,并常与古地震滑坡相伴生。初步获得的8个地震落石钙膜U系测年结果分布在距今6030±300a BP、4720±210a BP、3530±490~3560±280a BP、2010±160a BP、1090±70a BP、760±20a BP和230±20a BP年龄段,与该区古地震探槽和滑坡反映的地震事件比较吻合,进一步揭示玉树断裂带附近在全新世中晚期发生过多期可导致地表产生地震落石的事件。同时也说明,地震落石及其钙膜测年是特别值得进一步探索的潜在古地震研究方法或途径。     

A debris-flow alarm system for the Alpine Illgraben catchment: design and performance

We describe the development, implementation, and first analyses of the performance of a debris-flow warning system for the Illgraben catchment and debris fan area. The Illgraben catchment (9.5 km²), located in the Canton of Valais, Switzerland, in the Rhone River valley, is characterized by frequent and voluminous sediment transport and debris-flow activity, and is one of the most active debris-flow catchments in the Alps. It is the site of an instrumented debris-flow observation station in operation since the year 2000. The residents in Susten (municipality Leuk), tourists, and other land users, are exposed to a significant hazard. The warning system consists of four modules: community organizational planning (hazard awareness and preparedness), event detection and alerting, geomorphic catchment observation, and applied research to facilitate the development of an early warning system based on weather forecasting. The system presently provides automated alert signals near the active channel prior to (5–15 min) the arrival of a debris flow or flash flood at the uppermost frequently used channel crossing. It is intended to provide data to support decision-making for warning and evacuation, especially when unusually large debris flows are expected to leave the channel near populated areas. First-year results of the detection and alert module in comparison with the data from the independent debris-flow observation station are generally favorable. Twenty automated alerts (alarms) were issued, which triggered flashing lights and sirens at all major footpaths crossing the channel bed, for three debris flows and 16 flood flows. Only one false alarm was issued. The major difficulty we encountered is related to the variability and complexity of the events (e.g., events consisting of multiple surges) and can be largely solved by increasing the duration of the alarm. All of the alarms for hazardous events were produced by storms with a rainfall duration and intensity larger than the threshold for debris-flow activity that was defined in an earlier study, supporting our intention to investigate the use of rainfall forecasts to increase the time available for warning and implementation of active countermeasures.     

Rockfall risk management using a pre-failure deformation database

Terrestrial laser scanning (TLS) monitoring has been used to estimate the location, volume, and kinematics of a variety of small magnitude rockfalls before failure (1–1000 m³ range), and in some cases, potential failure time has been assessed through the application of inverse velocity methods. However, our current understanding of rock slope pre-failure behavior for this magnitude range and prediction ability is based on observations of a small number of failure case histories. In this study, a pre-failure deformation database was constructed for rockfall volumes exceeding 0.1 m³, observed over a 1252-day study interval at the Goldpan rock slope, British Columbia, Canada, in order to better understand the pre-failure behavior of rock slopes and provide an empirical means of estimating temporal failure ranges. Repeated TLS datasets were acquired at an average scanning interval of 2–3 months. A total of 90 rockfall events were recorded at this site, during this time period, of which 64 (71%) exhibited measurable deformation prior to failure. Classification of rockfalls by volume suggests that a scale dependency may exist, as deformation was detected for a greater proportion of rockfalls >?5 m³ (92%) than for smaller rockfalls in the range of 0.1–0.5 m³ (61%). A lower rate of pre-failure deformation detection was also reported for planar sliding failures as compared with wedge or toppling failures, suggesting that deformation was less easily detected for these failure types. This study proposes and implements a framework for rockfall assessment and forecasting that does not require continuous monitoring of deformation.     

Multiple rainfall event pollution transport by sustainable drainage systems: the fate of fine sediment pollution

A key design criterion of sustainable urban drainage systems is to mitigate urban stormwater pollution. Current research defines sustainable urban drainage systems (SuDS) pollutant treatment efficiency through the detention of total suspended solids, urban nutrients and heavy metal pollutants within the system during a design flow event, with research focusing on sand (>2 mm) sediment movement. The impact of multiple rainfall–runoff events on the fine sediment (<2 mm) treatment efficiency of SuDS is not yet well defined, and the temporal movement of detained sediment has not been investigated in detail. The field research presented in this paper addresses this research gap, monitoring ongoing fine sediment transport through a best-practice-designed SuDS network over 12 months through the use of a novel rare earth oxide trace methodology. Through time-stepped monitoring of the fine sediment pollution across three SuDS treatment trains (networks), the following key conclusions have been drawn. (1) That fine sediment becomes re-suspended and re-deposited within SuDS assets and the network as a result of ongoing multiple rainfall–runoff events. (2) That this re-suspension continues for over 52 weeks. (3) That by area, linear wetlands (within the monitored networks) outperform wetland and swale assets in multiple event fine sediment detention. And (4) that multiple event monitoring and analysis of fine sediment within a SuDS network highlights the under-performance of SuDS assets against current design event expectations.     

Evaluation of seismic events detection algorithms

Identification of seismic events from continuously recorded seismic data in real-time through a Digital Seismic Data Recording system is a difficult task. Despite the vast amount of research in this field, the signal processing and event parameters discrimination algorithms have not yet fully come of age. Presently, we have a wide spectrum of trigger algorithms, ranging from a very simple amplitude threshold type to the sophisticated ones based on pattern recognition approaches. Some of the other approaches use adaptive technique and neural network methods. Researchers are continuously making efforts for the development of algorithms using various techniques, which produce minimum false trigger. Some approaches have been reported which are accurate for detecting first phase of events and take minimum possible computational time. In this paper several approaches for detecting event signals in background noise are presented and their precision evaluation is discussed.     

Effects of sampling interval on the frequency - magnitude relationship of rockfalls detected from terrestrial laser scanning using semi-automated methods

Using change detection and semi-automated identification methods, it is possible to extract detailed rockfall information from terrestrial laser scanning data to build a database of events, which can be used in the development of the frequency-magnitude relationship for a slope. In this study, we have applied these methods to the White Canyon, a hazardous slope that presents rockfall hazards to the CN Rail line in British Columbia, to build a database of rockfalls including their locations, volumes, and block shapes. We identified over 1900 rockfall events during a 15-month period, ranging in volume from 0.01 to 45 m³. The frequency of these events changed throughout the year, with the highest periods of activity occurring over the winter months. We investigated how the sampling interval, or duration between scans, can affect how the rockfalls are identified, and therefore the frequency-magnitude relationship for the slope using datasets with fewer scans. We show that as the duration between scans becomes larger, fewer rockfalls are detected, as multiple events that have occurred in the same location cluster together into a single event. The results of this study can be used to assist the railways in planning the appropriate number and duration between future scans, in order to capture frequency-magnitude data for the slope with a desired level of detail.     

Calibration and validation of rockfall modelling at regional scale: application along a roadway in Mallorca (Spain) and organization of its management

The Tramuntana range, in the northwestern sector of the island of Mallorca (Balearic Islands, Spain), is frequently affected by rockfalls which have caused significant damage, mainly along the road network. In this work, we present the procedure we have applied to calibrate and validate rockfall modelling in this region, using 103 cases of the available detailed rockfall inventory (630 rockfall events collected since the eighteenth century). We have exploited STONE (Guzzetti et al. 2002), a GIS-based rockfall simulation software which computes 2D and 3D rockfall trajectories starting from a DTM and maps of the dynamic rolling friction coefficient and of the normal and tangential energy restitution coefficients. The appropriate identification of these parameters determines the accuracy of the simulation. To calibrate them, we have selected 40 rockfalls along the range which include a wide variety of outcropping lithologies. Coefficients values have been changed in numerous attempts in order to select those where the extent and shape of the simulation matched the field mapping. Best results were summarized with the average statistical values for each parameter and for each geotechnical unit, determining that mode values represent more precisely the data. Initially, for the validation stage, 10 well-known rockfalls exploited in the calibration phase have been selected. Confidence tests have been applied to their modelling results taking into account not only the success but also the mistakes. The best accuracy is obtained when the rockfall has a preferential trajectory and worse results when the rockfall follows two or more trajectories. Additionally, the greater the rockfall runout length, the less precise the simulation is. We have further validated the calibrated parameters along the Ma-road (111 km), the main transportation corridor in the range, using 63 rockfall events that occurred during the past 18 years along the road. Of the rockfalls where source areas were properly identified, 81.5 % are well represented by STONE modelling, as the travel paths and the depositional areas are successfully ascertained. Results of the analysis have been used by the Road Maintenance Service of Mallorca to assess hazard and risk posed by rockfall at regional scale to design the road management plan.     

Performance evaluation of the M8 algorithm to predict M7+ earthquakes in Turkey

The M8 algorithm is one of the most reliable intermediate-term middle-range earthquake prediction algorithms. The present study evaluates the ability of the M8 algorithm and its modified versions for predicting major events (M7+) in Turkey. Thirty different algorithms were developed by changing the radius of circle of investigation (CI) and the lower magnitude cutoff of the M8 algorithm. These modified algorithms were executed all over the territory of Turkey, and the results were evaluated using the error diagram. Each modified algorithm was executed for consecutive half-year intervals over a specified period of time. Subsequently, the seismic catalog was updated, and failures-to-predict ratio and the fraction of alarm were considered. Results showed that the location of areas of alarm change gradually over consecutive intervals, and no sudden changes can be observed. In addition, the annual changes of areas of alarm are not random and follow a pattern. This study also showed that the modified algorithm having a three to six annual average of events and a 393-km CI radius is an efficient algorithm for predicting the future seismic events in Turkey. This algorithm predicted six out of six target events, retrospectively, with a confidence level of 96.4 %. According to the obtained results, it will be possible to rely on this modified algorithm to predict near future earthquakes of Turkey. Furthermore, this study proves that it is possible to alter the M8 algorithm for being used in regional studies.     

Magnitude and frequency relations: are there geological constraints to the rockfall size?

There exists a transition between rockfalls, large rock mass failures, and rock avalanches. The magnitude and frequency relations (M/F) of the slope failure are increasingly used to assess the hazard level. The management of the rockfall risk requires the knowledge of the frequency of the events but also defining the worst case scenario, which is the one associated to the maximum expected (credible) rockfall event. The analysis of the volume distribution of the historical rockfall events in the slopes of the Solà d’Andorra during the last 50 years shows that they can be fitted to a power law. We argue that the extrapolation of the F-M relations far beyond the historical data is not appropriate in this case. Neither geomorphological evidences of past events nor the size of the potentially unstable rock masses identified in the slope support the occurrence of the large rockfall/rock avalanche volumes predicted by the power law. We have observed that the stability of the slope at the Solà is controlled by the presence of two sets of unfavorably dipping joints (F3, F5) that act as basal sliding planes of the detachable rock masses. The area of the basal sliding planes outcropping at the rockfall scars was measured with a terrestrial laser scanner. The distribution of the areas of the basal planes may be also fitted to a power law that shows a truncation for values bigger than 50 m² and a maximum exposed surface of 200 m². The analysis of the geological structure of the rock mass at the Solà d’Andorra makes us conclude that the size of the failures is controlled by the fracture pattern and that the maximum size of the failure is constrained. Two sets of steeply dipping faults (F1 and F7) interrupt the other joint sets and prevent the formation of continuous failure surfaces (F3 and F5). We conclude that due to the structural control, large slope failures in Andorra are not randomly distributed thus confirming the findings in other mountain ranges.     

Progressive failure leading to the 3 December 2013 rockfall at Puigcercós scarp (Catalonia,Spain)

A detailed analysis of the pre-failure behavior of the 3 December 2013 rockfall (1,012 m³) occurred on Puigcercós pilot study area (Catalonia, Spain) is presented. The exact date of failure was obtained based on a photographic monitoring performed every 4 h. The long-term monitoring (2,217 days) of the rock slope carried out by a Terrestrial LiDAR allowed the early detection of both pre-failure deformation and precursory rockfalls preceding the final failure. By means of the analysis of the pre-failure deformation, four different deformed areas were detected and the tertiary creep phase was observed in three of them. An attempt to predict the time to failure was performed using the Fukuzono’s (1985) method. Furthermore, the temporal evolution of the precursory rockfalls occurred in those four areas during the progressive failure showed a close resemblance with the exponential pattern of the cumulated displacements at tertiary creep stage. Finally, the study of the meteorological conditions did not show any single triggering factor associated with the final failure. Reversely, the increase in the occurrence of precursory rockfalls on several areas of the slope together with the observed acceleration on the deformation pattern reinforce the role of a progressive degradation of the stability conditions, which ultimately leaded to the 3 December rockfall event.     

融合区域潜在落石源区识别的峡谷区落石危险性评价

山西壶关太行山大峡谷景区为中国最美十大峡谷之一,但景区落石灾害频发,严重威胁景区安全运营。本文基于高精度地形信息与岩土体强度特性,采用坡度角分布方法开展区域尺度潜在落石源区识别,并引入岩体破坏敏感性指标定量描述潜在落石源区失稳概率。然后,利用经验模型Flow-R模拟落石运动扩散过程,获取落石的传播概率与能量分布情况。最后,提出落石危险性双因子评价模型实现落石危险性定量评估。获得主要结论如下:(1)研究区内潜在落石源区面积为25.7 km2(35.7%),主要以条带状分布于峡谷两侧陡壁。其中岩体破坏高敏感性区为3.3 km2。(2)研究区落石高危险区面积达3.22 km2,主要威胁景区内游客集散地与交通线路,尤其在S327荫林线红豆峡入口处落石危险性最高。(3)野外调查验证结果表明了应用坡度角分布方法识别潜在落石源区的高效性与准确性,提出的双因子评价模型可为峡谷区落石危险性评估提供快速解决方案。本文提出的“区域落石源区识别-源区失稳概率分析-落石危险性评估”的一整套技术方案能够为类似的高山峡谷区落石灾害早期识别及风险防控提供技术参考。     

SO2 degassing from Turrialba Volcano linked to seismic signatures during the period 2008–2012

In 1996, after 150 years of relative calm, Turrialba Volcano was reawakening. A visible plume and serious damage to surrounding vegetation due to acid rain are the most obvious signals. As part of the Network for Observation of Volcanic and Atmospheric Change project, four gas-monitoring stations were initially installed on the west flank of the volcano with the purpose of measuring sulphur dioxide emissions during this period of increased activity using the scanning-differential optical absorption spectroscopy technique. We present here the results of semicontinuous gas flux measurements over a period of 5 years (from 2008 to 2012), providing a novel data set that documents a relatively rapid increase in SO₂ fluxes from around 350 t day^?1 to around 4,000 t day^?1 leading up to an eruptive period, followed by a gradual return to the former baseline values. Gas flux data were also compared with seismic data for selected periods of interest, providing insights into the link between degassing and seismicity. The most important result from this comparison is the identification of an inflexion point in the gas emissions followed by a clearly increasing trend in seismic activity, distinguishable 6 months prior to a phreatic eruptive event that occurred on 5 January 2010. This signal can be interpreted as a possible indicator of future eruptive events. Monitoring of SO₂ thus complements seismic monitoring as a forecasting tool for eruptive events. Such monitoring is critical considering the proximity of Turrialba to the Central Valley, an area inhabited by more than 50 % of Costa Rica’s population.     

Summer rainstorm associated with a debris flow in the Amarilla gully affecting the international Agua Negra Pass (30°20′S), Argentina

The Central-West region of Argentina was seriously affected by a series of convective summer storms on January–February of 2013 generating many debris flows and rockfall in the Central Andes mountain regions. In particular, the unreported 8th February event caused the sad death of a 10-year-old child being completely ignored by society and local authorities. Despite this, meteorological conditions associated with this event and further episodes were rarely measured and determined mainly due to scarce meteorological stations in Andean mountain areas. In this paper, meteorological data from CMORPH algorithm and measurements of surrounding gauges were analyzed for estimating the triggering precipitation value of this event. As well, the particular debris flow channeled into the main branch of the Amarilla gully in the Agua Negra valley was geomorphologically described. The amount of precipitation associated with this debris flow was 5.5 and 13.2 mm accumulated previous to the event. This violent debris flow was generated in a talus zone in a periglacial environment located just below a covered rock glacier. However, the influence of the permafrost thawing in this process is not feasible. The altitude of the 0 °C isotherm was lower during the previous days of the event, and no monitoring on permafrost is available for this area. The volume of removed mass was estimated in 5 × 10⁴ m³, and the mean velocity was 35 km/h. Boulders of 4 m diameter were found in the source area, while the deposit is up to 75% sandy with clasts that hardly exceed 10 cm in the alluvial fan distal part. Herein the main objective is to advice about the probable catastrophic impact of similar events in the future. These findings could be useful for hazard remediation, mitigation, and prevention plans for the Agua Negra international pass under construction.     

Forecasting rock slope failure: how reliable and effective are warning systems?

Warning systems are increasingly applied to reduce damage caused by different magnitudes of rockslides and rockfalls. In an integrated risk-management approach, the optimal risk mitigation strategy is identified by comparing the achieved effectiveness and cost; estimating the reliability of the warning system is the basis for such considerations. Here, we calculate the reliability and effectiveness of the warning system installed in Preonzo prior to a major rockfall in May 2012. “Reliability” is defined as the ability of the warning system to forecast the hazard event and to prevent damage. To be cost-effective, the warning system should forecast an event with a limited number of false alarms to avoid unnecessary costs for intervention measures. The analysis shows that to be reliable, warning systems should be designed as fail-safe constructions. They should incorporate components with low failure probabilities, high redundancy, have low warning thresholds, and additional control systems. In addition, the experts operating the warning system should have limited risk tolerance. In an additional hypothetical probabilistic analysis, we investigate the effect of the risk attitude of the decision makers and of the number of sensors on the probability of detecting the event and initiating a timely evacuation, as well as on the related intervention cost. The analysis demonstrates that quantitative assessments can support the identification of optimal warning system designs and decision criteria.     

Nanoseismic monitoring of gravity-induced slope instabilities for the risk management of an aqueduct infrastructure in Central Apennines (Italy)

A monitoring system is operative in the Peschiera Springs slope (Central Apennines, Italy) to mitigate the landslide risk related to the hosted main drainage plant of Rome aqueducts by providing alert warning. Such a strategy allows to avoid out-of-service episodes so reducing extra-costs of water distribution management. The Peschiera Springs slope is involved in a rock mass creep characterized by an average steady strain rate of 1 mm year⁻¹ and responsible for several landforms including sinkholes, subvertical scarps and trenches. Moreover, an average aquifer discharge of 19 m³ s⁻¹ causes an intense limestone dissolution concentrated in correspondence with release bands and discontinuities that dislodge the jointed rock mass. Since 2008, an accelerometric network has been operating within the slope; about 1300 microseismic local events were recorded up to now, distinguished in failures and collapses. A control index, based on frequency of occurrence and cumulative energy of the recorded microseismic events was defined to provide three levels of alert. In 2013, a temporary nanoseismic Seismic Navigation System (SNS) array was installed inside a tunnel of the drainage plant to integrate the pre-existent seismic monitoring system. This array allowed to record 37 microseismic events, which locations are in good agreement with the evolutionary geological model of the ongoing gravitational slope deformation. In 2014, a permanent nanoseismic SNS array was installed in the plant and allowed to record several sequences of underground collapses including more than 500 events. The nanoseismic monitoring system is allowing to: (1) increase the detection level of the monitoring system; (2) locate hypocentres of the events; and (3) detect precursors of the strongest events.

     

Nanoseismic monitoring of gravity-induced slope instabilities for the risk management of an aqueduct infrastructure in Central Apennines (Italy)

A monitoring system is operative in the Peschiera Springs slope (Central Apennines, Italy) to mitigate the landslide risk related to the hosted main drainage plant of Rome aqueducts by providing alert warning. Such a strategy allows to avoid out-of-service episodes so reducing extra-costs of water distribution management. The Peschiera Springs slope is involved in a rock mass creep characterized by an average steady strain rate of 1 mm year^?1 and responsible for several landforms including sinkholes, subvertical scarps and trenches. Moreover, an average aquifer discharge of 19 m³ s^?1 causes an intense limestone dissolution concentrated in correspondence with release bands and discontinuities that dislodge the jointed rock mass. Since 2008, an accelerometric network has been operating within the slope; about 1300 microseismic local events were recorded up to now, distinguished in failures and collapses. A control index, based on frequency of occurrence and cumulative energy of the recorded microseismic events was defined to provide three levels of alert. In 2013, a temporary nanoseismic Seismic Navigation System (SNS) array was installed inside a tunnel of the drainage plant to integrate the pre-existent seismic monitoring system. This array allowed to record 37 microseismic events, which locations are in good agreement with the evolutionary geological model of the ongoing gravitational slope deformation. In 2014, a permanent nanoseismic SNS array was installed in the plant and allowed to record several sequences of underground collapses including more than 500 events. The nanoseismic monitoring system is allowing to: (1) increase the detection level of the monitoring system; (2) locate hypocentres of the events; and (3) detect precursors of the strongest events.     

Landslide susceptibility mapping at Vaz Watershed (Iran) using an artificial neural network model: a comparison between multilayer perceptron (MLP) and radial basic function (RBF) algorithms

Landslide susceptibility and hazard assessments are the most important steps in landslide risk mapping. The main objective of this study was to investigate and compare the results of two artificial neural network (ANN) algorithms, i.e., multilayer perceptron (MLP) and radial basic function (RBF) for spatial prediction of landslide susceptibility in Vaz Watershed, Iran. At first, landslide locations were identified by aerial photographs and field surveys, and a total of 136 landside locations were constructed from various sources. Then the landslide inventory map was randomly split into a training dataset 70 % (95 landslide locations) for training the ANN model and the remaining 30 % (41 landslides locations) was used for validation purpose. Nine landslide conditioning factors such as slope, slope aspect, altitude, land use, lithology, distance from rivers, distance from roads, distance from faults, and rainfall were constructed in geographical information system. In this study, both MLP and RBF algorithms were used in artificial neural network model. The results showed that MLP with Broyden–Fletcher–Goldfarb–Shanno learning algorithm is more efficient than RBF in landslide susceptibility mapping for the study area. Finally the landslide susceptibility maps were validated using the validation data (i.e., 30 % landslide location data that was not used during the model construction) using area under the curve (AUC) method. The success rate curve showed that the area under the curve for RBF and MLP was 0.9085 (90.85 %) and 0.9193 (91.93 %) accuracy, respectively. Similarly, the validation result showed that the area under the curve for MLP and RBF models were 0.881 (88.1 %) and 0.8724 (87.24 %), respectively. The results of this study showed that landslide susceptibility mapping in the Vaz Watershed of Iran using the ANN approach is viable and can be used for land use planning.