Assessment of alternative forest road routes and landslide susceptibility mapping using machine learning

ABSTRACT

Background:

Forest roads are among the most basic infrastructure used for forestry activities and services. To facilitate the increased amount of biomass harvesting adequately, the existing road network may require modifications to allow forest transportation within harvesting units that are not yet accessed by the roads. The construction of a forest road can trigger landslides, so the necessary constraints should be considered when the road is being planned to preclude such problems. Landslide Susceptibility Mapping (LSM) has become an integral part of the growing process of machine learning (ML), providing a more effective platform for practitioners, planners, and decision-makers. This study aims to reveal the most suitable alternative routes for a forest road, especially in areas susceptible to landslides, and to provide an effective tool for decision-makers.

Results:

For this purpose, two models were developed through ML: Logistic Regression (LR) and Random Forest (RF). Elevation, slope, aspect, curvature, Topographic Wetness Index (TWI), Stream Power Index (SPI), distance from the fault, the road, and the stream, and lithology were considered as the main landslide susceptibility factors in these models. The best model was obtained by the RF approach with an Area Under ROC Curve (AUC) value of 81.9%, while the LR model was 78.2%. LSM data was used as a base, and alternative routes were obtained through CostPath analysis.

Conclusion:

It has been shown that the ML methods used in this study can positively contribute to decision-making by providing more effective LSM calculations in studies to determine alternative routes in a forest road network.

Keywords:
Roads routing problem; random forest; logistic regression; R software

HIGHLIGHTS

Route determination that can be passed with the least damage in landslide sensitive areas Innovative approach to computer aided forest road routing.

Planning of an environmentally friendly alternative forest road.

INTRODUCTION

Landslides, which can occur almost anywhere in the world, are mass movements of land ranging from small areas to ones of considerable scale that threaten people and the environment and cause various degrees of loss and damage ( Glade and Crozier, 2005). Landslides also have negative short and long-term economic consequences for those affected and may incur heavy costs ( Klose et al., 2015). Many factors affect the formation of landslides. They include elevation ( Sarma et al., 2020), slope (degree) ( Saha and Saha, 2020), aspect ( Lee and Min, 2001), curvature ( Pourghasemi et al., 2018), the TWI ( Nhu et al., 2020), the SPI ( Gholami et al., 2019), distance from the fault ( Shirzadi et al., 2017), distance from the road ( Sun et al. 2020), distance from the stream ( Wubalem and Meten, 2020), and lithology ( Rosi et al., 2018). However, areas susceptible to landslides can be effectively classified in advance using GIS techniques. Therefore, having identified these susceptible areas, more detailed planning may be conducted, and measures taken to prevent negative outcomes. ( Raja et al., 2017; Bugday and Akay, 2019).

In Turkey, forests are in mountainous, sloping regions with higher rainfall levels than the surrounding landscape and the removal of vegetation for road construction increases the risk of further landslides. With an increasing demand for wood as a raw material, new methods of construction are required. This is especially true in areas susceptible to landslides, where greater detail should be included in planning new forest roads to prevent damage and loss from landslides. Road construction also has a direct effect on the landslide since it results in an increase in all elements of the risk equation (Risk = Hazard × Exposure × Vulnerability) ( Lummen and Yamada, 2014), while landslides damage various infrastructure facilities, such as roads and buildings ( Cascini et al., 2013). It is vital, therefore, that Remote Sensing (RS) techniques and GIS are employed to minimize these damaging and negative effects or to avoid these risky areas while providing positive alternatives.

Landslide Susceptibility Mapping (LSM) modeling studies have grown in number and location throughout the world during the past 20 years, with the success rates of the models generated at between 65% and 98%. ( Kavzoglu et al., 2019). Approaches to LSM modeling vary widely, and some of the most common approaches are highlighted in this study: AHP ( Kayastha et al., 2013; Roccati et al., 2021; Grozavu and Patriche, 2021), ANFIS ( Paryani et al., 2020; Chen et al., 2021), ANN ( Chen et al., 2017), PSO-ANN ( Moayedi et al., 2019), Weighting Factor ( Yalcin, 2008; Hussain et al., 2021), Bayesian ( Sun et al., 2021; Lee et al., 2020), Deep Learning ( Dao et al., 2020; Ngo et al., 2021), Frequency Ratio ( Senanayake et al., 2020; Berhane et al., 2020), Fuzzy Logic ( Tsangaratos et al., 2018; Razifard et al., 2019), Logistic Regression ( Schlögel et al., 2018; Chen et al., 2019), Machine Learning ( Ghorbanzadeh et al., 2019, Kavzoglu et al., 2019, Mohammady et al., 2021), M-AHP ( Nefeslioglu et al., 2012; Bugday and Akay, 2019), Multilayer Perceptron Neural Network ( Li et al., 2019; Hong et al., 2020), SWARA ( Dehnavi et al., 2015; Pourghasemi et al., 2019).

The main aim of the study was to reveal the most suitable LSMs to use when planning road construction projects in forests located in steep and sloping areas and to create a support system to determine the limitations of alternative routes. To this end, two different models were created through LR and RF modeling. Elevation, slope (degree), aspect, curvature, TWI, SPI, distance from the fault, the road, the stream, and lithology, were used in the modeling. Using data obtained from the models, suitable routes were created automatically by computer with the help of least CostPath analysis using ArcGIS software. The results were compared with those obtained from a traditional approach to route planning.

MATERIAL AND METHODS

Study Area

The study area is located in the central border of Zonguldak province in the north of Turkey and in Devrek Forest District ( Figure 1). It is also located in an area where forest areas are widespread and landslides are experienced in Karadere locality. The study area is 29095 ha and located between the latitude of 41°16' 22'' and 41°20' 24'' and longitude of 31°47' 40'', 31°56' 18''. In the study area, there are pure and mixed stands of beech, hornbeam, sessile oak, black pine forests, which are generally in maturity for harvesting. These forests are managed according to the principles of Ecosystem-Based Functional Planning (ETFOP) according to their various functions ( Zengin et al., 2011). The existing forest roads in the study area are low-standard B-type roads (6m road width and 4m road surface) which are defined according to the geometric classification of the General Directorate of Forestry (GDF).

Figure 1.
Location of landslides in the study area.

Landslide Susceptibility Factors

The landslide susceptibility factors evaluated in this study include elevation, slope (degree), aspect, curvature, Topographic Wetness Index (TWI), Stream Power Index (SPI), distance from the fault, the road, and the stream, and lithology. Elevation is an effective factor in forest road planning because it increases both landslide sensitivity and cost in the performance of road construction ( Akay, 2006). As elevation also means an increase in the distance to settlements, it additionally reduces the costs of periodic maintenance works ( Bugday and Akay, 2019). Aspect is one of the factors that affects soil properties and the growing environment. Aspect was studied in eight different directions ( Lee and Talib, 2005). The slope is one of the most effective factors in landslide formation ( Ma et al., 2020). It is also an important factor as it has a direct effect on the construction costs for forest roads ( Akay et al., 2008). In this study, International Union of Forest Research Organizations (IUFRO) slope classes were expressed in five different degree classes, 0–5.71, 5.71–13.80, 13.80–21.88, 21.88–31.99, and >32. Curvature is among the factors that affect both the direction and severity of the landslide ( Ohlmacher, 2007). The TWI is widely used to determine the location and size of areas saturated with water at the topographic level ( Zhang et al., 2020). SPI is defined as the ability of flowing water to erode topography, considering the assumption that the flow is proportional to the specific basin area ( Sameen et al., 2020). Distance from the fault is one of the factors that is widely used in landslide susceptibility studies and it plays an important role in triggering landslides ( Demir, 2019). In this study, it was analyzed in five zones including 0.5 km, 1 km, 2 km, 5 km, and 10 km. Another important factor in triggering the landslide is the distance from the road ( Sur et al., 2021). In this study, it was evaluated in six zones with intervals of 100 m, 250 m, 500 m, 1,000 m, 1,500 m, and 2,000 m. Distance from the stream is commonly used in studies where proximity relationship is important ( Wang et al., 2017). In this study, stream distances were considered in four zones including 0.5 km, 1 km, 2 km, and 5 km. Lithology affects the cost of construction of forest roads because it reveals characteristics of the bedrock ( Tang et al., 2021). In this study, lithology was evaluated in six different groups ( Figure 2).

Figure 2.
SM factors in forested area; (a) elevation, (b) slope (degree), (c) aspect, (d) curvature, (e) Topographic Wetness Index (TWI), (f) Stream Power Index (SPI), (g) distance from the fault, (h) distance from the road, (i) distance from the stream, and ( j) lithology.

The Digital Elevation Model (DEM) was obtained free of charge from ASTER GDEM, published on the web, and elevation, aspect, slope (degree), curvature, TWI, and SPI factors were generated using ArcGIS 10.3 TM software. Distance from the road was obtained by using forest subdistrict databases where the study area was located. The field data of lithology, distance from the fault and stream, and landslides that had occurred in the past, were obtained from the General Directorate of Mineral Research and Explorations (GDMRE) ( Duman et al., 2011).

Generating LSM

In this study, the LSM tool pack developed by Sahin et al. (2020), working with R integration in an ArcGIS environment, was used to develop LSM. According to this tool pack, with LR and RF modeling, effective and more accurate LSM estimates can be made by using the above factors. Out of 108 landslides, 80% (86 landslides) of the data was used for training purposes and 20% (22 landslides) for testing.

ArcGIS 10.3 software was used to evaluate the factors with LR and RF methods. To validate the models obtained by these approaches, information on landslide events that had occurred in the past was tested. Models developed according to LR and RF were tested with Receiver Operating Characteristic (ROC) analysis and AUC value. The AUC score was classified as 0.9–1.0 (excellent), 0.8–0.9 (very good), 0.7–0.8 (good), 0.6–0.7 (moderate), and 0.5–0.6 (weak) ( Swets, 1988). The model outputs were recorded as raster images.

Determination of Forest Road Routes

ArcGIS-CostPath analysis was used in automatically determining alternative roads’ routes, which was the final stage of the study. This analysis included a methodology that allowed for an objective comparison of alternative scenarios for weighting factors (i.e. slope length, elevation, slope, positive cardinal points, production amount, etc.) that determine the location of a route ( ESRI, 2016). The methodology used three scenarios to identify alternative routes that were compared to determine the effectiveness and susceptibility of this approach.

First, the route study started with the determination of the two points that were outside the existing roads that needed to be connected to each other and the alternative routes to be planned. Route limitation was made by the positioning of the starting and destination points. Planning for the road was based on the slope criteria, which were obtained by traditional methods. In the next stage, the route was recalculated with the help of CostPath analysis, with consideration given to the data on landslide sensitivity that was obtained through the LR and RF methods. The workflow for this study is summarized in Figure 3.

Figure 3.
Flowchart of the study

RESULTS

Digital Maps of Landslide Susceptibility Factors

Study area elevation values varied between 100 m and 1,020 m, with an average elevation of 500 m ( Figure 3a). The dominant aspect of the study area was to the south. The average slope of the area was 16.1° and in the working area it was determined as 52.2°. The length of the roads in the study area was calculated as 514 km in a total area of 29,094.6 ha.

LSM modeling and validation

In this study, models with ten factors were developed according to LR and RF modeling. The ROC, which highlights the model’s performance in LSM studies, and the AUC, which expresses the area under this graph, were used for the validation. First, the importance of each factor was analyzed by the application of widely used statistical methods, chi-square, information gain, and random forest importance. Table 1 lists the results from high to low significance levels. As the table shows, each method produced different feature weights. In the rankings, it was observed that the first four factors were in the same order in this study, and there were differences in the rankings of other factors according to the statistical approach.

Table 1.
Feature importance’s of the feature ranking algorithms

No	Factors	Chi-Squared	Factors	Information Gain	Factors	Random Forest Importance
1	Aspect	0,4521	Aspect	0,3882	Aspect	303,7328
2	Lithology	0,3385	Lithology	0,1246	Lithology	82,0112
3	Elevation	0,2444	Elevation	0,0983	Elevation	74,0761
4	Slope	0,1987	Slope	0,0217	Slope	58,8738
5	Dis.stream	0,1459	Dis.stream	0,0108	Dis.stream	32,6708
6	SPI	0,1307	SPI	0,0097	Dis.fault	29,2779
7	TWI	0,1140	TWI	0,0085	TWI	25,5365
8	Dis.road	0,1041	Dis.road	0,0077	SPI	23,3148
9	Dis.fault	0,0961	Dis.fault	0,0071	Dis.road	21,5252
10	Curvature	0,0739	Curvature	0,0055	Curvature	16,5537

The next step was to evaluate the effects of these features on the performance of the prediction model. For this purpose, factors were included in the iterative estimation process in the LSM tool pack by placing them in ascending order according to their estimated importance. Estimates were made for a data set with an increasing number of factors at each iteration. From these estimates, the best subset, which included factors that provided high or equal estimation, was selected and the best models were determined. Various binary statistical tests (Wilcoxon signed-rank test, F-Test, Kolmogorov Smirnov test, and One Sample T-Test) are presented in the LSM tool pack ( Table 2). In this study, the scenario using chi-square as the feature sorting algorithm and the F-Test to analyze the differences in the performance of the prediction algorithm in the best subset selection is discussed. All possible scenarios, which consider factor selection and statistical approaches, are shown in Table 2.

Table 2.
Best feature subset size by Chi-Square, Information Gain, and Random Forest Importance

Feature ranking method	Case no	Statistical test used for subset selection	Model No: The best subset size selected by performance of LR	Features in the best subset
Chi-Square	Case 1	F-test	Model 9	Aspect, Lithology, Elevation, Slope, Dis.stream, SPI, TWI, Dis.road, Dis.fault
	Case 2	Kolmogorov Smirnov test	Model 8	Aspect, Lithology, Elevation, Slope, Dis.stream, SPI, TWI, Dis.road
	Case 3	One Sample T-Test	Model 5	Aspect, Lithology, Elevation, Slope, Dis.stream
	Case 4	Wilcoxon signed-rank test	Model 9	Aspect, Lithology, Elevation, Slope, Dis.stream, SPI, TWI, Dis.road, Dis.fault
Information Gain	Case 5	F-test	Model 9	Aspect, Lithology, Elevation, Slope, Dis.stream, SPI, TWI, Dis.road, Dis.fault
	Case 6	Kolmogorov Smirnov test	Model 9	Aspect, Lithology, Elevation, Slope, Dis.stream, SPI, TWI, Dis.road, Dis.fault
	Case 7	One Sample T-Test	Model 8	Aspect, Lithology, Elevation, Slope, Dis.stream, SPI, TWI, Dis.road
	Case 8	Wilcoxon signed-rank test	Model 9	Aspect, Lithology, Elevation, Slope, Dis.stream, SPI, TWI, Dis.road, Dis.fault
RF-Importance	Case 9	F-test	Model 9	Aspect, Lithology, Elevation, Slope, Dis.stream, SPI, TWI, Dis.road, Dis.fault
	Case 10	Kolmogorov Smirnov test	Model 9	Aspect, Lithology, Elevation, Slope, Dis.stream, SPI, TWI, Dis.road, Dis.fault
	Case 11	One Sample T-Test	Model 5	Aspect, Lithology, Elevation, Slope, Dis.stream, SPI
	Case 12	Wilcoxon signed-rank test	Model 8	Aspect, Lithology, Elevation, Slope, Dis.stream, SPI, TWI, Dis.road

Logistic regression analysis

LR is a widely used modeling approach in studies on landslide susceptibility. The most successful models out of a total of ten factors using this approach are shown in Table 2. The AUC value (91.2172) of the Case 1 model was selected as the most successful according to the LR approach; estimate, standard error, z-value, and Pr values of the factors are given in Figure 4. This image presents a positive correlation between aspect, curvature, slope, SPI, lithology factors, elevation, TWI, and distance from the fault, the stream, and the road to the landslide formation. In addition, when the p-values were examined in terms of statistical significance in the study area, it was calculated that the factors of aspect, elevation, slope, lithology, distance from the fault, and distance from the stream were more important.

Figure 4.
LR model AUC score and statistics of factors

Random Forest analysis

RF is one of the most commonly used modeling approaches in studies on landslide susceptibility. Again, ten factors were used, and the most successful model obtained (Case 1) and the order of importance of the factors are shown in Figure 5.

Figure 5.
RF model AUC score and statistics of factors

According to the results obtained here, the ranking based on the LR approach differed from the RF method. Ranking according to the RF approach was aspect, lithology, elevation, slope, distance from the stream, distance from the fault, TWI, SPI, distance from the road, and curvature. According to this rating and the IncNodePrutiy measure, it was calculated that the importance levels of the factors, aspect, lithology, elevation, and slope, were quite high when compared with other factors.

Performance Comparison of LR and RF modeling approaches

At this stage, the best performing LSM model (Case 1, Model 9) was compared according to LR and RF. The performance results are presented as a graphic and a table in Figure 6. Considering the results, it was determined that the best model was 9 in both approaches and it had good to very good model success with a RF-AUC score of 0.81 and a LR-AUC score of 0.78.

Figure 6.
Models performance results of LR and RF approaches

Alternative forest road route detection

The study area generally consisted of beech stands that were at the cutting age. This area is among several locations that will not be opened for production in the near future. Instead, it is planned to develop the stands to their optimum through continuing forest maintenance with the aim of producing quality timber product for the market. Results from the LR and RF approaches in the study showed that landslide susceptibility was quite high in the northern and eastern parts of the study area ( Figures 7 and 8). Road construction was required in the study area. In this study, CostPath analysis was conducted using ArcGIS software. As a result of the planning, it was discovered that the route developed by traditional methods would pass through the middle of the area with high landslide susceptibility. In the analysis of route determination using LR and RF LSMs produced in this study, it was determined that alternative routes should pass through areas with very low susceptibility to landslides.

Figure 7.
LSM generated from the LR method and CostPath road route

Figure 8.
LSM generated from the RF method and CostPath road route

DISCUSSION

Forest road planning studies in Turkey have accelerated since the 1960s ( Erdaş, 1997). With the widespread use of developing technologies, the determination of road routes and the construction works have been modernized. In addition to these developments, there has been a significant increase in software and the number of expert users of this technology. As in other sectors, GIS software is used for multi-criteria path planning and alternative routes in the forestry sector. The software contributes greatly to more effective decision-making in the planning process. It is especially important in providing advance information when determining which areas are susceptible to landslides and for providing a basis for studies to be carried out in those areas. The GIS system, therefore, is a convenient tool for decision-makers and planners in providing this transfer of information. In this study, LSM was developed according to two different approaches, LR and RF. The achievements of models in national and international literature vary according to approaches between approximately 65% and 98% AUC ( Kavzoglu et al., 2019). The main factors affecting the success rates are the sensitivity and quality of the data, the size of the area studied, and the advantages of the approach used. In similar studies, Roccati et al. (2021) found the AUC value to be 73% according to nine factors; Grozavu and Patriche (2021) reported 75% according to three factors; Mohammady et al. (2021) found 73.8% according to 12 factors; and Kavzoglu et al. (2019) reached 96% according to eight factors.

The determination of alternative routes is generally searched to provide the location of the line required to optimally connect the starting and ending points ( Bast et al., 2016). This study aimed to reveal alternative routes in a forest area that required location of a new forest road network. For this purpose, three different forest road routes (based on traditional method, LR, and RF methods) were introduced ( Figure 7 and 8). As in this study, CostPath analysis is widely used in determining alternative routes. The difference in this study, however, is that the analysis of determining alternative routes was applied to the forest road. In a related study, Kadi et al. (2019) used AHP as a multi-criteria decision method and the route planning was made in MATLAB software. Picchio et al. (2018) implemented alternative road planning method, similar to present study, was used, but the landslide criteria were not considered by them. In another related study, Bugday and Akay (2019) evaluated landslide criteria in forest road planning, but alternative routes were not systematically searched.

CONCLUSION

Results from the model showed a susceptibility to landslide throughout the study area. Therefore, the introduction of alternative routes in such sensitive areas acquires greater importance. For effective planning, it is essential that the elements and stages of a plan are evaluated in greater detail to provide clearer information to the decision-maker. Considering the landslide criteria will provide a goal-oriented environment where costs can be calculated more consistently depending on the purpose of the road and its geometric features. The traditional length of the route determined by this study was approximately 2,500 m; while it was approximately 3,080 m according to the LR method, and 3,650 m according to RF. Depending on the purpose of the road, LR or RF routes may be preferred. Unplanned and inappropriately implemented forest roads may cause environmental damages, thus, it is important to minimize the possible damages through conducting detailed planning and proper implementation in the field. In addition, it is considered that multi-criteria planning with the help of GIS will be beneficial in the short and long term, especially in areas with landslide susceptibility, so that the service expected from forest roads can be ensured without interruption throughout the year.

ACKNOWLEDGMENTS

We would like to thank the General Directorate of Forestry in Turkey and General Directorate of Mineral Research and Explorations in Turkey for providing data.

REFERENCES

• ABDI, E., MAJNOUNIAN, B., DARVISHSEFAT, A., MASHAYEKHI, Z., SESSIONS, J. A GIS-MCE based model for forest road planning. Journal of Forest Science, 55(4), 171-176, 2009.
• AKAY, A. E. Minimizing total costs of forest roads with computer-aided design model. Sadhana, 31(5), 621-633, 2006.
• AKAY, A.E., ERDAS, O., REIS, M. YUKSEL, A. Estimating Sediment Yield from a Forest Road Network by Using a Sediment Prediction Model and GIS Techniques. Building and Environment. 43, 687–695, 2008
• BAST, H., DELLING, D., GOLDBERG, A., MÜLLER-HANNEMANN, M., PAJOR, T., SANDERS, P., WAGNER, D., WERNECK, R. F. Route planning in transportation networks. In Algorithm engineering (pp. 19-80). Springer, Cham., 2016.
• BERHANE, G., KEBEDE, M., ALFARAH, N., HAGOS, E., GRUM, B., GIDAY, A., ABERA, T. Landslide susceptibility zonation mapping using GIS-based frequency ratio model with multi-class spatial data-sets in the Adwa-Adigrat mountain chains, northern Ethiopia. Journal of African Earth Sciences, 164, 103795, 2020.
• BUGDAY, E., AKAY, A. E. Evaluation of forest road network planning in landslide sensitive areas by GIS-based multi-criteria decision making approaches in Ihsangazi watershed, Northern Turkey. Šumarski list, 143(7-8), 325-336, 2019.
• CASCINI, L., PEDUTO, D., PISCIOTTA, G., ARENA, L., FERLISI, S., FORNARO, G. The combination of DInSAR and facility damage data for the updatingof slow-moving landslide inventory maps at medium scale. Natural hazards and earth system sciences, 13(6), 1527-1549, 2013.
• CHEN, W., CHEN, X., PENG, J., PANAHI, M., LEE, S. Landslide susceptibilitymodeling based on ANFIS with teaching-learning-based optimization and Satin bowerbird optimizer. Geoscience Frontiers, 12(1), 93-107, 2021.
• CHEN, W., POURGHASEMI, H. R., KORNEJADY, A., ZHANG, N. Landslide spatial modeling: Introducing new ensembles of ANN, MaxEnt, and SVM machine learning techniques. Geoderma, 305, 314-327, 2017.
• CHEN, W., ZHAO, X., SHAHABI, H., SHIRZADI, A., KHOSRAVI, K., CHAI, H., ZHANG, S., ZHANG, L., MA, J., CHEN, Y., WANG, X., AHMAD, B. B., LI, R. Spatial prediction of landslide susceptibility by combining evidential belief function, logistic regression and logistic model tree. Geocarto International, 34(11), 1177-1201, 2019.
• DEHNAVI, A., AGHDAM, I. N., PRADHAN, B., VARZANDEH, M. H. M. A new hybrid model using step-wise weight assessment ratio analysis (SWARA) technique and adaptive neuro-fuzzy inference system (ANFIS) for regional landslide hazard assessment in Iran. Catena, 135, 122-148, 2015.
• DEMIR, G. GIS-based landslide susceptibility mapping for a part of the North Anatolian Fault Zone between Reşadiye and Koyulhisar (Turkey). Catena, 183, 104211, 2019.
• DAO, D.V., JAAFARI, A., BAYAT, M., MAFI-GHOLAMI, D., QI, C., MOAYEDI, H., PHONG, T.V., LY, H., LE, T., TRINH, P.T., LUU, C., QUOC, N.K., THANH, B.N., PHAM, B.T. A spatially explicit deep learning neural network model for the prediction of landslide susceptibility. Catena, 188, 104451.ISSN 0341-8162, https://doi.org/10.1016/j.catena.2019.104451, 2020.
  » https://doi.org/10.1016/j.catena.2019.104451
• DUMAN, T.Y., T. ÇAN, Ö. EMRE. 1/1.500.000 Türkiye Heyelan Envanteri Haritası, Maden Tetkik ve Arama Genel Müdürlüğü Özel Yayınlar Serisi -27, Ankara, Türkiye. ISBN:978-605-4075-85-3, 2011.
• ERDAŞ, O. Orman Yolları–Cilt I. KTÜ Orman Fakültesi Yayınları, 187, 25.4, 1997.
• ESRI. 2016. Available at: https://desktop.arcgis.com/en/arcmap/10.3/tools/spatial-analyst-toolbox/cost-path.htm Accessed in: 16.11.2020.
  » https://desktop.arcgis.com/en/arcmap/10.3/tools/spatial-analyst-toolbox/cost-path.htm
• GHOLAMI, M., GHACHKANLU, E. N., KHOSRAVI, K., PIRASTEH, S. Landslide prediction capability by comparison of frequency ratio, fuzzy gamma and landslide index method. Journal of Earth System Science, 128(2), 1-22, 2019.
• GHORBANZADEH, O., BLASCHKE, T., GHOLAMNIA, K., MEENA, S. R., TIEDE, D., ARYAL, J. Evaluation of different machine learning methods and deeplearning convolutional neural networks for landslide detection. Remote Sensing, 11(2), 196, 2019.
• GLADE, T., CROZIER, M. J. Landslide hazard and risk: concluding comment and perspectives. Landslide hazard and risk. Wiley, Chichester, 767-774, 2005.
• GROZAVU, A., PATRICHE, C. V. Mapping landslide susceptibility at national scale by spatial multi-criteria evaluation. Geomatics, Natural Hazards and Risk, 12(1), 1127-1152, 2021.
• HONG, H., TSANGARATOS, P., ILIA, I., LOUPASAKIS, C., WANG, Y. Introducing a novel multi-layer perceptron network based on stochastic gradient descent optimized by a meta-heuristic algorithm for landslide susceptibility mapping. Science of the total environment, 742, 140549, 2020.
• HUSSAIN, M. L., SHAFIQUE, M., BACHA, A. S. Landslide inventory and susceptibility assessment using multiple statistical approaches along the Karakoram highway. northern Pakistan. Journal of Mountain Science, 18(3), 2021.
• KADI, F., YILDIRIM, F., SARALIOGLU, E. Risk analysis of forest roads using landslide susceptibility maps and generation of the optimum forest road route: a case study in Macka, Turkey. Geocarto International, 1-18, 2019.
• KAVZOGLU, T., COLKESEN, I., SAHIN, E. K. Machine Learning Techniques in Landslide Susceptibility Mapping: A Survey and a Case Study. In Landslides: Theory, Practice and Modelling (pp. 283-301). Springer, Cham., 2019.
• KAYASTHA, P., DHITAL, M. R., DE SMEDT, F. Application of the analytical hierarchy process (AHP) for landslide susceptibility mapping: A case studyfrom the Tinau watershed, west Nepal. Computers & Geosciences, 52, 398408, 2013.
• KLOSE, M., DAMM, B., TERHORST, B. Landslide cost modeling for transportation infrastructures: a methodological approach. Landslides, 12(2), 321-334, 2015.
• LEE, S., MIN, K. Statistical analysis of landslide susceptibility at Yongin, Korea. Environmental geology, 40(9), 1095-1113, 2001.
• LEE, S., TALIB, J. A. Probabilistic landslide susceptibility and factor effect analysis. Environmental Geology, 47(7), 982-990, 2005.
• LEE, S., LEE, M. J., JUNG, H. S., LEE, S. Landslide susceptibility mapping using naïve bayes and bayesian network models in Umyeonsan, Korea. Geocarto international, 35(15), 1665-1679, 2020.
• LI, D., HUANG, F., YAN, L., CAO, Z., CHEN, J., YE, Z. Landslide susceptibility prediction using particle-swarm-optimized multilayer perceptron: Comparisons with multilayer-perceptron-only, bp neural network, and information value models. Applied Sciences, 9(18), 3664, 2019.
• LUMMEN, N. S., YAMADA, F. Implementation of an integrated vulnerability and risk assessment model. Natural hazards, 73(2), 1085-1117, 2014.
• MA, S., QIU, H., HU, S., PEI, Y., YANG, W., YANG, D., CAO, M. Quantitative assessment of landslide susceptibility on the Loess Plateau in China. Physical Geography, 41(6), 489-516, 2020.
• MOAYEDI, H., MEHRABI, M., MOSALLANEZHAD, M., RASHID, A. S. A., PRADHAN, B. Modification of landslide susceptibility mapping usingoptimized PSO-ANN technique. Engineering with Computers, 35(3), 967984, 2019.
• MOHAMMADY, M., POURGHASEMI, H. R., AMIRI, M., TIEFENBACHER, J. P. Spatial modeling of susceptibility to subsidence using machine learning techniques. Stochastic Environmental Research and Risk Assessment, 1-12, 2021.
• NAJAFI, A., SOBHANI, H., SAEED, A., MAKHDOM, M., MOHAJER, M. M. Planning and assessment of alternative forest road and skidding networks. Croatian Journal of Forest Engineering: Journal for Theory and Application of Forestry Engineering, 29(1), 63-73, 2008.
• NEFESLIOGLU, H.A., SAN, T., GOKCEOGLU, C., DUMAN, T.Y. An assessment on the use of Terra ASTER L3A data in landslide susceptibility mapping. Int. J. Appl. Earth Obs. Geoinf. 14 (1), 40–60, 2012.
• NGO, P. T. T., PANAHI, M., KHOSRAVI, K., GHORBANZADEH, O., KARIMINEJAD, N., CERDA, A., LEE, S. Evaluation of deep learning algorithms for national scale landslide susceptibility mapping of Iran. Geoscience Frontiers, 12(2),505-519, 2021.
• NHU, V. H., HOANG, N. D., NGUYEN, H., NGO, P. T. T., BUI, T. T., HOA, P. V., HOA, SAMUI, P., BUI, D. T. Effectiveness assessment of Keras based deeplearning with different robust optimization algorithms for shallow landslide susceptibility mapping at tropical area. Catena, 188, 104458, 2020.
• OHLMACHER, G. C. Plan curvature and landslide probability in regions dominated by earth flows and earth slides. Engineering Geology, 91(2-4),117-134 , 2007.
• PARYANI, S., NESHAT, A., JAVADI, S., PRADHAN, B. Comparative performance of new hybrid ANFIS models in landslide susceptibility mapping. Natural Hazards, 103, 1961-1988, 2020.
• PICCHIO, R., PIGNATTI, G., MARCHI, E., LATTERINI, F., BENANCHI, M., FODERI, C., VENAZZI, R., VERANI, S. The application of two approaches using GIS technology implementation in forest road network planning in an Italian mountain setting. Forests, 9(5), 277, 2018.
• POURGHASEMI, H. R., GAYEN, A., PANAHI, M., REZAIE, F., BLASCHKE, T. Multi-hazard probability assessment and mapping in Iran. Science of the total environment, 692, 556-571, 2019.
• POURGHASEMI, H. R., GAYEN, A., PARK, S., LEE, C. W., LEE, S. Assessment of landslide-prone areas and their zonation using logistic regression, logitboost, and naïvebayes machine-learning algorithms. Sustainability, 10(10), 3697, 2018.
• POURGHASEMI, H. R., YANSARI, Z. T., PANAGOS, P., PRADHAN, B. Analysis and evaluation of landslide susceptibility: a review on articles published during 2005–2016 (periods of 2005–2012 and 2013–2016). Arabian Journal of Geosciences, 11(9), 193, 2018.
• RAJA, N. B., ÇIÇEK, I., TÜRKOĞLU, N., AYDIN, O., KAWASAKI, A. Landslide susceptibility mapping of the Sera River Basin using logistic regression model. Natural Hazards, 85(3), 1323-1346, 2017.
• RAZIFARD, M., SHOAEI, G., ZARE, M. Application of fuzzy logic in the preparation of hazard maps of landslides triggered by the twin Ahar-Varzeghan earthquakes (2012). Bulletin of Engineering Geology and the Environment, 78(1), 223-245, 2019.
• ROCCATI, A., PALIAGA, G., LUINO, F., FACCINI, F., TURCONI, L. GIS-Based Landslide Susceptibility Mapping for Land Use Planning and Risk Assessment. Land, 10(2), 162, 2021.
• ROSI, A., TOFANI, V., TANTERI, L., STEFANELLI, C. T., AGOSTINI, A., CATANI, F., CASAGLI, N. The new landslide inventory of Tuscany (Italy) updated with PS-InSAR: geomorphological features and landslide distribution. Landslides, 15(1), 5-19, 2018.
• SAHA, A., SAHA, S. Comparing the efficiency of weight of evidence, support vector machine and their ensemble approaches in landslide susceptibility modelling: A study on Kurseong region of Darjeeling Himalaya, India. Remote Sensing Applications: Society and Environment, 19, 100323, 2020.
• SAHIN, E. K., COLKESEN, I., ACMALI, S. S., AKGUN, A., AYDINOGLU, A. C. Developing comprehensive geocomputation tools for landslide susceptibility mapping: LSM tool pack. Computers & Geosciences, 144, 104592, 2020.
• SAMEEN, M. I., PRADHAN, B., LEE, S. Application of convolutional neural networks featuring Bayesian optimization for landslide susceptibilityassessment. Catena, 186, 104249, 2020.
• SARMA, C. P., DEY, A., KRISHNA, A. M. Influence of digital elevation models on the simulation of rainfall-induced landslides in the hillslopes of Guwahati, India. Engineering Geology, 268, 105523, 2020.
• SCHLÖGEL, R., MARCHESINI, I., ALVIOLI, M., REICHENBACH, P., ROSSI, M., MALET, J. P. Optimizing landslide susceptibility zonation: Effects of DEM spatial resolution and slope unit delineation on logistic regression models. Geomorphology, 301, 10-20, 2018.
• SENANAYAKE, S., PRADHAN, B., HUETE, A., BRENNAN, J. Assessing soil erosion hazards using land-use change and landslide frequency ratio method: A case study of Sabaragamuwa province, Sri Lanka. Remote Sensing, 12(9), 1483, 2020.
• SHIRZADI, A., BUI, D. T., PHAM, B. T., SOLAIMANI, K., CHAPI, K., KAVIAN, A., SHAHABI H., REVHAUG, I. Shallow landslide susceptibilityassessment using anovel hybrid intelligence approach. Environmental Earth Sciences, 76(2), 60. doi.org/10.1007/s12665-016-6374-y, 2017.
  » https://doi.org/10.1007/s12665-016-6374-y
• SUN, D., WEN, H., WANG, D., XU, J. A random forest model of landslide susceptibility mapping based on hyperparameter optimization using Bayes algorithm. Geomorphology, 362, 107201, 2020.
• SUN, D., XU, J., WEN, H., WANG, D. Assessment of landslide susceptibility mapping based on Bayesian hyperparameter optimization: A comparison between logistic regression and random forest. Engineering Geology, 281, 105972, 2021.
• SUR, U., SINGH, P., RAI, P. K., THAKUR, J. K. Landslide probability mapping by considering fuzzy numerical risk factor (FNRF) and landscape change for road corridor of Uttarakhand, India. Environment, Development and Sustainability, 1-29, 2021.
• SWETS, J. A. Measuring the accuracy of diagnostic systems. Science, 240(4857), 1285-1293, 1988.
• TANG, R. X., YAN, E., WEN, T., YIN, X. M., TANG, W. Comparison of Logistic Regression, Information Value, and Comprehensive Evaluating Model for Landslide Susceptibility Mapping. Sustainability, 13(7), 3803, 2021.
• TSANGARATOS, P., LOUPASAKIS, C., NIKOLAKOPOULOS, K., ANGELITSA, V., ILIA, I. Developing a landslide susceptibility map based on remote sensing, fuzzy logic and expert knowledge of the Island of Lefkada, Greece. Environmental Earth Sciences, 77(10), 1-23, 2018.
• WANG, F., XU, P., WANG, C., WANG, N., JIANG, N. Application of a GIS-based slope unit method for landslide susceptibility mapping along the LongziRiver, Southeastern Tibetan Plateau, China. ISPRS International Journal of Geo-Information, 6(6), 172, 2017.
• WUBALEM, A., METEN, M. Landslide susceptibility mapping usinginformation value and logistic regression models in Goncha Siso Eneses area, northwestern Ethiopia. SN Applied Sciences, 2(5), 1-19, 2020.
• YALCIN, A. GIS-based landslide susceptibility mapping using analytical hierarchy process and bivariate statistics in Ardesen (Turkey): Comparisons of results and confirmations. Catena, 72(1), 1–12. https://doi.org/10.1016/j.catena.2007.01.003, 2008.
  » https://doi.org/10.1016/j.catena.2007.01.003
• ZENGIN, H., BOZALI, N., ASAN, Ü., DESTAN, S., DEĞIRMENCI, A. S. Ekosistem Tabanlı Fonksiyonel Planlamada Tamsayılı Programlama ile Optimizasyon. KSU J. Engineering Sci., Special Issue, p183-190, 2011.
• ZHANG, Y. X., LAN, H. X., LI, L. P., WU, Y. M., CHEN, J. H., TIAN, N. M. Optimizing the frequency ratio method for landslide susceptibility assessment: A case study of the Caiyuan Basin in the southeast mountainous area of China. Journal of Mountain Science, 17(2), 340-357, 2020.

Publication Dates

• Publication in this collection
  16 Dec 2022
• Date of issue
  2022

History

• Received
  08 Dec 2021
• Accepted
  08 July 2022