# Hazard Assessment of Potential Large-Scale Landslides in the Watershed of the Chenyulan River

^{1}

^{2}

^{*}

## Abstract

**:**

## 1. Introduction

^{2}, of which some two-thirds consists of mountain slopes. Most of its geology belongs to the tertiary or quaternary strata, which are relatively new and quite fragile. The significant erosion of riverbeds caused by the steep slopes and swift currents increases the likelihood of slope disasters such as landslides, debris flows, and landslides. In addition, Taiwan has a subtropical marine climate, with an average annual rainfall of up to 2500 mm, mostly concentrated in summer and autumn. Its rapid industrialization and associated increase in population density have led to the development of hillsides, which—where not carefully planned and executed—has led to a decline in their stability. Moreover, the increasingly extreme hydrological conditions associated with climate change, when coupled with the above-mentioned topographical and human factors, mean that the scale of slope disasters is trending upwards. Road damage in mountainous areas inevitably results from heavy rain and typhoons, and comes at a high cost in life and property. In the years since the “921 Jiji earthquake” of 1999 loosened topsoil, the area of Taiwan deemed to be landslide-prone nearly doubled, from 8110 to 15,977 hectares; and worse, the areas actually affected by landslides increased from 2535 to 25,845 hectares [1]. Among the large-scale mass movement disasters that have been more prevalent in Taiwan since 1999, the most serious was the destruction of Xiaolin Village, following deep damage to Xiandu Mountain caused by Typhoon Morakot. Since that incident, scholars and government agencies have devoted increasing attention to large-scale landslide events, but such research is challenging because actual cases are relatively rare. Therefore, this study explores methods of modeling such events using satellite images taken before and after previous ones, to serve the wider aim of creating an early warning system that will reduce the impact of these disasters.

^{3}, or a depth of more than 10 m; and these same characteristics are likely to render them fast-moving [10,11]. It takes a lot of time and effort to investigate tragedies on sloping land the traditional way, which involves sending inspectors there on foot. Consequently, really steep terrain in many alpine places is not actually examined. Telemetry photos are of tremendous use in disaster prevention on sloping ground because they can serve as the foundation for further study and analysis across a wide area. As a result, these photos serve as the major source of data for the current work, and correlational analysis is utilized to examine the impact of different conditions on how landslides develop. It is intended that the findings will give government agencies entrusted with decreasing the losses from such calamities a quick and simple reference.

## 2. Materials and Methods

#### 2.1. Study Area

#### 2.2. Research-Case Selection

#### 2.3. Satellite Imaging by SPOT

#### 2.4. Screening of Terrestrial Factors

#### 2.5. Selection of Material Factors

#### 2.5.1. Geological Materials: Strata Types

#### 2.5.2. Geological Structure: Distance to Fault

#### 2.5.3. Vegetation Status: Pre-Event Average NDVI

#### 2.6. Inducing Factors

## 3. Results and Discussion

#### 3.1. Interpretation of Landslides in SPOT Images

#### 3.2. Statistical Analysis of Terrestrial Factors

#### 3.3. Inspection of the Rainfield Cutting Pearson

#### 3.4. Results of Discriminant Analysis

#### 3.5. Results of Logistic Regression Analysis

#### 3.6. Back-Propagation Neural Network Judgment Results

#### 3.6.1. Normalization

_{norm}= (X + a)/b

_{max}− 9 X

_{min})/8; b = (X

_{max}− X

_{min})/0.8; X is the actual value; X

_{max}is the maximum actual value; and X

_{min}is the minimum actual value.

#### 3.6.2. Back-Propagation Neural Network

#### 3.7. Receiver Operating Characteristic Curve

#### 3.8. Comparison of Three Analysis Modes

#### 3.9. Results of Cluster Analysis of Catchment Areas’ Landslide Potential

#### 3.9.1. Cluster A: Small Catchment Areas

#### 3.9.2. Cluster B: Large Catchment Areas

#### 3.9.3. Cluster C: Fragile Strata Catchment Areas

#### 3.9.4. Cluster D: Curvature Small Catchment Area

#### 3.9.5. Discuss of Cluster Analysis of Catchment Areas’ Landslide Potential

## 4. Conclusions

## Author Contributions

## Funding

## Institutional Review Board Statement

## Informed Consent Statement

## Data Availability Statement

## Conflicts of Interest

## References

- Large-Scale Landslide, Soil and Water Conservation Bureau, COA. Available online: https://246.swcb.gov.tw/Landslide/ (accessed on 18 July 2022).
- Varnes, D.J. Slope Movement Types and Processes. In Special Report 176: Landslides: Analysis and Control; Schuster, R.L., Krizek, R.J., Eds.; Transportation and Road Research Board, National Academy of Science: Washington, WA, USA, 1978; pp. 11–33. [Google Scholar]
- Santacana, N.; Baeza, B.; Corominas, J.; Paz, A.D.; Marturiá, A. GIS-Based Multivariate Statistical Analysis for Shallow Landslide Susceptibility Mapping in La Pobla de Lillet Area (Eastern Pyrenees, Spain). Nat. Hazards
**2020**, 30, 281–295. [Google Scholar] [CrossRef] - Liu, J.K.; Shih, T.Y. Collection and application of 3D space information with airborne lidar technology. J. Civ. Hydraul. Eng.
**2009**, 36, 52–63. [Google Scholar] - Renwick, W.; Brumbaugh, R.; Loeher, L. Landslide Morphology and Processes on Santa cruz Island, California. Geogr. Annaler. Ser. A Phys. Geogr.
**1982**, 64, 149–159. [Google Scholar] [CrossRef] - Fell, R.F.; Hungr, O.; Leroueil, S.; Riemer, W. Keynote Lecture—Geotechnical Engineering of The Stability Of Natural Slopes, And Cuts And Fills In Soil. In Proceedings of the International Conference on Geotechnical and Geological Engineering, Melbourne, Australia, 19–24 November 2000; Technomic Publishing Co.: Lancaster, PA, USA, 2000. Available online: https://www.researchgate.net/publication/267403410_Keynote_lecture-Geotechnical_engineering_of_the_stability_of_natural_slopes_and_cuts_and_fills_in_soil (accessed on 9 November 2022).
- Giannecchini, R. Relationship between rainfall and shallow landslides in the southern Apuan Alps (Italy). Nat. Hazards Earth System Sci.
**2006**, 6, 357–364. [Google Scholar] [CrossRef] - No. 4115; Public Works Research Institute Sediment Management Research Group Volcano/Debris Flow Team: Public Works Research Institute Material. PWRI: Tsukuba, Japan, 2008.
- Minami, N. Deep-seated Landslide and administrative measures, International Symposium on Slopeland Disaster Mitigation, SWCB et al., Taichung, Taiwan. 2010 International Landslide Disaster Technology Exchange Conference.
- Chigira, M. September 2005 rain- induced catastrophic rockslides on slopes affected by deep-seated gravitational deformations, Kyushu, southern Japan. Eng. Geol.
**2009**, 108, 1–15. [Google Scholar] [CrossRef] - Taiwan-Large-Scale Landslides, National Science and Technology Center for Disaster Reduction. Available online: https://tccip.ncdr.nat.gov.tw/ark_02_case_one.aspx?case_id=LS02 (accessed on 18 July 2022).
- Rainfall Classification by the Central Meteorological Bureau of the Ministry of Communications. 2020. Available online: https://www.cwb.gov.tw/V8/C/K/CommonFaq/rain_all.html (accessed on 15 July 2020).
- Huang, H.J. Talking about the landslide of the Forest. Taiwan For. J.
**2008**, 12, 54–64. [Google Scholar] - Lin, C.R. A Study on the Application of Potential Hazard Index to the Watershed Classification and Regionalization at Pingtung. Master’s Thesis, Department of Soil and Water Conservation, National Pintung University of Science and Technology, Pingtung, Taiwan, 2004. [Google Scholar]
- Wang, L.J.; Guo, M.; Sawada, K.; Lin, J.; Zhang, J. Landslide susceptibility mapping in Mizunami City, Japan: A comparison between logistic regression, bivariate statistical analysis and multivariate adaptive regression spline models. Catena
**2015**, 135, 271–282. [Google Scholar] [CrossRef] - Shou, K.J. Study on the Control Factors of Rainfall Induced Landslides in Central Taiwan. Master’s Thesis, Department of Civil Engineering, National Chung Hsing University, Taichung, Taiwan, 2007. [Google Scholar]
- Wu, B.Y. Landslides and Engineered Slopes. In The Past to the Future, Two Volumes + CD-ROM; Chapter: Analysis of control factors for landslides in Central Taiwan area; CRC Press: Boca Raton, FL, USA, 2006. [Google Scholar]
- Chen, H.H. Application of Satellite Imagery for Nan-Chin Road Potential Landsliding Analysis. Master’s Thesis, Department of Environmental Engineering and Science, Feng Chia University, Taichung, Taiwan, 2005. [Google Scholar]
- Wu, C.H.; Chen, S.C. A Landslide Potential Model for Ming-De Reservoir Watershed. J. Soil Water Conserv.
**2005**, 37, 155–168. [Google Scholar] - Jan, C.D. Rainfall Threshold Criterion for Debris-Flow Initiation, Soil and Water Conservation Bureau. Master’s Thesis, National Cheng Kung University, Tainan, Taiwan, 2002. [Google Scholar]
- Hong, C.Y. Time Series Analysis of Control Factors of Landslides in Central Taiwan. Master’s Thesis, Department of Civil Engineering, National Chung Hsing University, Taichung, Taiwan, 2010. [Google Scholar]
- Landslide Classification by Central Geological Survey, MOEA. Available online: https://www.moeacgs.gov.tw/eng/Faqs/faqs (accessed on 8 July 2020).
- Pratsinis, S.E.; Zeldin, M.D.; Ellis, E.C. Source Resolution of the Fine Carbonaceous Aerosol by Principal Componet-Stepwise Regression Analysis. Environ. Sci. Technol.
**1988**, 22, 212–216. [Google Scholar] [CrossRef] [PubMed] - Kaiser, H.F. The Application of Electronic Computers to Factor Analysis. Educ. Psychol. Meas.
**1960**, 20, 141–151. [Google Scholar] [CrossRef] - Hanley, J.A.; McNeil, B.J. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology
**1982**, 143, 29–36. [Google Scholar] [CrossRef] [PubMed][Green Version] - Wang, S.; Zhang, K.; Beek, L.P.H.; Tain, X.; Bogaard, T.S. Physically-based landslide prediction over a large region: Scaling low-resolution hydrological model results for high-resolution slope stability assessment. Environ. Model. Softw.
**2020**, 124, 104607. [Google Scholar] [CrossRef] - Bordoni, M.; Galanti, Y.; Bartelletti, C.; Persichillo, M.G.; Barsanti, M.; Giannecchini, R.; Avanzi, G.D.; Cevasco, A.; Brandolini, P.; Galve, J.P.; et al. The influence of the inventory on the determination of the rainfall-induced shallow landslides susceptibility using generalized additive models. CATENA
**2020**, 193, 104630. [Google Scholar] [CrossRef] - He, J.Y.; Qiu, H.J.; Qu, F.H.; Hu, S.; Yang, D.D.; Shen, D.D.; Zhang, Y.; Sun, H.S.; Cao, M.M. Prediction of spatiotemporal stability and rainfall threshold of shallow landslides using the TRIGRS and Scoops3D models. CATENA
**2021**, 197, 104999. [Google Scholar] [CrossRef]

**Figure 3.**Schematic of a receiver operating characteristic curve as indicative of classification accuracy.

**Figure 4.**Receiver operating characteristic curve, discriminant analysis, logistic regression analysis, and artificial neural network; AUC = area under curve.

**Table 1.**Rainfall classification by the Central Meteorological Bureau of the Ministry of Communications.

Type of Alert | Rainfall Classification | Definition |
---|---|---|

Heavy rain special | Heavy rain | Above 80 mm/24 h or 40 mm/1 h |

Torrential rain | Extremely heavy rain | Above 200 mm/24 h or 100 mm/3 h |

Torrential rain | Above 350 mm/24 h or 200 mm/3 h | |

Extremely torrential rain | Above 500 mm/24 h |

Parameter | Description of Calculation | Meaning (Where Not Self-Explanatory) |
---|---|---|

Total area of catchment | A = a × n A: grid area [${m}^{2}$];n: number of grids | |

Average elevation of catchment area | $\mathrm{H}=\frac{\sum \mathrm{h}}{\mathrm{n}}$ {h: elevation of the grid} | |

Average slope of catchment area | $\mathrm{S}=\frac{\sum \mathrm{slope}}{\mathrm{n}}$ {slope: slope value of the grid} | |

Total curvature | $\mathrm{Z}=\frac{\sum \mathrm{z}}{\mathrm{n}}$ {Z: total curvature of the grid} | The rate of change representing the slope or direction of slope |

Roughness | $r=\mathrm{max}\left({e}_{i}\right)-\mathrm{min}\left({e}_{i}\right)$ | The difference between the maximum and minimum of the surrounding elevation |

Terrain Ruggedness Index, TPI | $\mathrm{TPI}=\mathrm{e}-{{\displaystyle \sum}}_{i=1}^{8}\frac{{e}_{i}}{8}$ | The center-point elevation minus the average of the surrounding elevations |

Topographic Position Index, TRI | $\mathrm{TRI}={\displaystyle {\displaystyle \sum}_{i=1}^{8}}\frac{\left|\mathrm{e}-{e}_{i}\right|}{8}$ | The average of the difference between the elevation of the center point and the surrounding elevation |

Formation Name | Score | Landslide Ratio | Formation Name | Score | Landslide Ratio |
---|---|---|---|---|---|

Sanhsia group and its equivalents | 10 | 2.8% | Tatungshan formation | 1.91 | 0.28% |

Meta-conglomerate | 9.76 | 2.73% | Terrace deposits | 1.87 | 0.27% |

Lushan formation | 7.42 | 2% | Toukoshan formation and its equivalents | 1.8 | 0.25% |

Chinshui shale and its equivalents | 6.73 | 1.78% | Cholan formation and its equivalents | 1.79 | 0.25% |

Kankou formation | 5.81 | 1.5% | Hsichun formation and Neokao formation | 1.28 | 0.09% |

Juifang group and its equivalents | 3.81 | 0.87% | Alluvium | 1.03 | 0.01% |

Yehliu group and its equivalents | 3.69 | 0.84% | Lateritic terrace deposits | 1.03 | 0.01% |

Tananao schists | 2.21 | 0.38% | Basic rock | 1 | 0% |

Distance between Assessment Point and Fault Zone | Score | Distance between Assessment Point and Fault Zone | Score |
---|---|---|---|

<100 m | 10 | 600–700 m | 4 |

100–200 m | 9 | 700–800 m | 3 |

200–300 m | 8 | 800–900 m | 2 |

300–400 m | 7 | 900–1000 m | 1 |

400–500 m | 6 | >1000 m | 0 |

500–600 m | 5 |

Method | Correction Method | Start of Rain | End of Rain | Method |
---|---|---|---|---|

1 | No rain for previous 24 h | No rain for 24 consecutive hours | 1 | |

2 | 1 | No rain for previous 12 h | No rain for 12 consecutive hours | 2 |

3 | Hourly rainfall is greater than 4 mm | Rainfall is less than 4 mm for three consecutive hours | 3 | |

4 | Cumulative rainfall is at least 10 mm in the first 24 h | Cumulative rainfall is less than 10 mm for 24 h | 4 | |

5 | 3 | Hourly rainfall is greater than 4 mm | Rainfall is less than 4 mm for six consecutive hours | 5 |

6 | 4 | Cumulative rainfall is at least 10 mm in the first 12 h | Cumulative rainfall is less than 10 mm for 12 h | 6 |

Accuracy NDVI Difference | Landslide Accuracy | Non-Landslide Accuracy | Overall Accuracy |
---|---|---|---|

0.15 | 83.6% | 86.9% | 86.6% |

0.2 | 70.6% | 94.2% | 93.7% |

0.25 | 57.4% | 97.2% | 96.3% |

Accuracy NDVI Difference | Landslide Accuracy | Non-Landslide Accuracy | Overall Accuracy |
---|---|---|---|

0.15 | 72.7% | 86.6% | 86.3% |

0.2 | 61.5% | 95.3% | 95.2% |

0.25 | 32.6% | 95.5% | 95.5% |

Total Area of Catchment | Average Elevation of Catchment Area | Average Slope of Catchment Area | Total Curvature | Roughness | TPI | TRI | |
---|---|---|---|---|---|---|---|

Total area of catchment | 1.000 | 0.438 | 0.251 | 0.103 | 0.258 | −0.002 | 0.252 |

Average elevation of catchment area | 0.438 | 1.000 | 0.774 | 0.079 | 0.678 | −0.109 | 0.680 |

Average slope of catchment area | 0.251 | 0.774 | 1.000 | 0.109 | 0.985 | −0.105 | 0.988 |

Total curvature | 0.103 | 0.079 | 0.109 | 1.000 | 0.122 | 0.706 | 0.107 |

Roughness | 0.258 | 0.678 | 0.985 | 0.122 | 1.000 | −0.120 | 0.998 |

Topo-graphic Position Index, TPI | −0.002 | −0.109 | −0.105 | 0.706 | −0.120 | 1.000 | −0.140 |

Terrain Rugged-ness Index, TRI | 0.252 | 0.680 | 0.988 | 0.107 | 0.998 | −0.140 | 1.000 |

Ingredient | Initial Eigenvalues | Sum of Squares Loading Extraction | ||||
---|---|---|---|---|---|---|

Sum | Variation (%) | Accumulation (%) | Sum | Variation (%) | Accumulation (%) | |

1 | 3.686 | 52.661 | 52.661 | 3.686 | 52.661 | 52.661 |

2 | 1.614 | 23.055 | 75.716 | 1.614 | 23.055 | 75.716 |

3 | 0.951 | 13.587 | 89.303 | |||

4 | 0.372 | 5.312 | 94.614 | |||

5 | 0.359 | 5.126 | 99.740 | |||

6 | 0.016 | 0.233 | 99.973 | |||

7 | 0.002 | 0.027 | 100.000 |

Terrestrial Factor | Ingredients | |
---|---|---|

One | Two | |

Total area of catchment | 0.412 | 0.725 |

Average elevation of catchment area | 0.810 | −0.004 |

Average slope of catchment area | 0.967 | −0.008 |

Total curvature | 0.131 | 0.894 |

Roughness | 0.972 | −0.009 |

Topographic Position Index, TPI | −0.135 | 0.893 |

Terrain Ruggedness Index, TRI | 0.972 | −0.029 |

Factor | Maximum Daily Rain | Maximum Hourly Rainfall | Mean Hourly Rainfall | Maximum 24-h Rainfall | Cumulative Rainfall |
---|---|---|---|---|---|

Maximum daily rain | 1.000 | 0.287 | 0.425 | 0.905 | 0.726 |

Maximum hourly rainfall | 0.287 | 1.000 | 0.589 | 0.301 | 0.239 |

Mean hourly rainfall | 0.425 | 0.589 | 1.000 | 0.367 | 0.017 |

Maximum 24-h rainfall | 0.905 | 0.301 | 0.367 | 1.000 | 0.830 |

Cumulative rainfall | 0.726 | 0.239 | 0.017 | 0.830 | 1.000 |

Factor | 0 | 1 | Coefficient Vector (1–0) |
---|---|---|---|

Total area of catchment area | −0.001 | 0.010 | 0.01 |

Type of stratum | 0.895 | 0.865 | −0.03 |

Distance from fault | 0.857 | 0.855 | −0.002 |

Average slope of catchment area | 0.770 | 0.775 | 0.005 |

Total curvature | −14.663 | −15.045 | −0.382 |

Maximum daily rainfall | 0.024 | 0.031 | 0.008 |

Average NDVI before the event | 31.857 | 36.255 | 4.397 |

Hourly rainfall | 0.219 | 0.160 | −0.059 |

Constant | −33.138 | −38.222 | −5.084 |

Classification Result | Total | |||
---|---|---|---|---|

Category | 0 | 1 | ||

Training sample (1095 rows) | 0 | 891 (91.5%) | 83 (8.5%) | 974 (100%) |

1 | 29 (24%) | 92 (76%) | 121 (100%) | |

Correct discrimination rate = [(891 + 92)/(974 + 121)] × 100% = 89.8% | ||||

Classification result | Total | |||

Category | 0 | 1 | ||

Validation sample (219 rows) | 0 | 160 (88.9%) | 20 (11.1%) | 180 (100%) |

1 | 13 (33.3%) | 26 (66.7%) | 39 (100%) | |

Correct discrimination rate = [(160 + 26)/(180 + 39)] × 100% = 84.9% | ||||

Classification result | Total | |||

Category | 0 | 1 | ||

Overall sample (1314 rows) | 0 | 1051 (91.1%) | 103 (8.9%) | 1154 (100%) |

1 | 42 (26.3%) | 118 (73.7%) | 160 (100%) | |

Correct discrimination rate = [(1051 + 118)/(1154 + 160)] × 100% = 89% |

Code | Factor | Coefficient | Coefficient Value |
---|---|---|---|

${\mathrm{L}}_{1}$ | Total area of the catchment area | ${\mathrm{W}}_{1}$ | 0.007 |

${\mathrm{L}}_{2}$ | Type of stratum | ${\mathrm{W}}_{2}$ | −0.045 |

${\mathrm{L}}_{3}$ | Distance from the fault | ${\mathrm{W}}_{3}$ | 0.018 |

${\mathrm{L}}_{4}$ | Average slope of the catchment area | ${\mathrm{W}}_{4}$ | 0.043 |

${\mathrm{L}}_{5}$ | Maximum daily rainfall | ${\mathrm{W}}_{5}$ | 0.008 |

${\mathrm{L}}_{6}$ | Average NDVI before the event | ${\mathrm{W}}_{6}$ | 3.160 |

${\mathrm{L}}_{7}$ | Total curvature | ${\mathrm{W}}_{7}$ | −2.639 |

${\mathrm{L}}_{8}$ | Hourly rainfall | ${\mathrm{W}}_{8}$ | −0.116 |

Constant | −4.473 |

Classification Result | Total | |||
---|---|---|---|---|

Category | 0 | 1 | ||

Training sample (1095 rows) | 0 | 962 (98.8%) | 12 (1.2%) | 974 (100%) |

1 | 57 (47.1%) | 64 (52.9%) | 121 (100%) | |

Accurate discrimination rate = [(962 + 64)/(974 + 121)] × 100% = 93.7% | ||||

Validation sample (219 rows) | 0 | 173 (96.1%) | 7 (3.9%) | 180 (100%) |

1 | 26 (66.7%) | 13 (33.3%) | 39 (100%) | |

Accurate discrimination rate = [(173 + 13)/(180 + 39)] × 100% = 84.9% | ||||

Overall sample (1314 rows) | 0 | 1135 (98.4%) | 19 (1.6%) | 1154 (100%) |

1 | 83 (51.9%) | 77 (48.1%) | 160 (100%) | |

Accurate discrimination rate = [(1135 + 77)/(1154 + 160)] × 100% = 92.2% |

Training Sample (920 Rows) | Classification Result | Total | |||
---|---|---|---|---|---|

Category | 0 | 1 | |||

Original category | Number (920 rows) | 0 | 793 (97.5%) | 20 (2.5%) | 813 (100%) |

1 | 35 (32.7%) | 72 (67.3%) | 107 (100%) | ||

Accurate discrimination rate = [(962 + 64)/(974 + 121)] × 100% = 93.7% | |||||

Original category | Number (394 rows) | 0 | 327 (95.9%) | 14 (4.1%) | 341 (100%) |

1 | 14 (26.4%) | 39 (73.6%) | 53 (100%) | ||

Accurate discrimination rate = [(173 + 13)/(180 + 39)] × 100% = 84.9% | |||||

Original category | Number (1314 rows) | 0 | 1120 (97.1%) | 34 (2.9%) | 1154 (100%) |

1 | 49 (30.6%) | 111 (69.4%) | 160 (100%) | ||

Accurate discrimination rate = [(1120 + 111)/(1154 + 160)] × 100% = 93.7% |

Factor | Significance | Importance of Normalization |
---|---|---|

Total area of the catchment area | 0.247 | 100.0% |

Maximum daily rainfall | 0.206 | 83.2% |

Maximum hourly rainfall | 0.164 | 66.5% |

Maximum hourly rainfall | 0.134 | 54.3% |

Total curvature | 0.077 | 31.3% |

Type of stratum | 0.065 | 26.3% |

Average slope of the catchment area | 0.061 | 24.8% |

Distance from the fault | 0.045 | 18.2% |

**Table 18.**Area under curve discrimination-ability thresholds (Hanley and McNeil, 1982) [25].

Value of Area under the Receiver Operating Characteristic Curve (AUC) | Discrimination Ability |
---|---|

AUC ≧ 0.9 | Excellent discrimination ability |

0.7 ≦ AUC < 0.9 | Good discrimination ability |

0.5 ≦ AUC < 0.7 | Fair discrimination ability |

AUC = 0.5 | No ability to discriminate |

Observed Value Predictive Value | Unstable | Stable |
---|---|---|

Unstable | TP | FP |

true positive | false positive | |

Stable | FN | TN |

false negative | true negative |

Impact Factor | Cluster A | Cluster B | Cluster C | Cluster D | Sort by Size |
---|---|---|---|---|---|

Total catchment area (hectares) | 70.152 | 1067.59 | 680.815 | 329.178 | B > C > D > A |

Type of stratum | 7.444 | 6.220 | 8.902 | 7.395 | C > A > D > B |

Distance from fault | 2.249 | 0.800 | 0.091 | 1.433 | A > D > B > C |

Average slope of catchment area (degrees) | 33.086 | 38.294 | 36.486 | 35.002 | B > C > D > A |

Total curvature (degrees) | −0.0001 | 0.006 | 0.002 | −0.003 | B > C > A > D |

Items | Total Number | Number of Destructive Events | Destruction as a Percentage of the Cluster | Percentage of Total Damage |
---|---|---|---|---|

Cluster | ||||

Cluster A | 1038 | 42 | 4.1% | 26.3% |

Cluster B | 30 | 20 | 66.7% | 12.5% |

Cluster C | 66 | 34 | 51.5% | 21.3% |

Cluster D | 180 | 64 | 35.6% | 40% |

Cluster | A | B | C | D |
---|---|---|---|---|

Typhoon | (173 Areas) | (5 Areas) | (11 Areas) | (30 Areas) |

Typhoon Mindulle | 0 (0%) | 0 (0%) | 0 (0%) | 0 (0%) |

Typhoon Haitang | 18 (10.4%) | 5 (100%) | 11 (100%) | 26 (86.7%) |

Typhoon Kalmaegi | 0 (0%) | 0 (0%) | 0 (0%) | 0 (0%) |

Typhoon Sinlaku | 4 (2.3%) | 5 (100%) | 7 (63.6%) | 11 (36.7%) |

Typhoon Morakot | 8 (4.6%) | 5 (100%) | 8 (72.7%) | 13 (43.3%) |

Typhoon Saola | 12 (6.9%) | 5 (100%) | 8 (72.7%) | 14 (46.7%) |

Statistic | Number of Damaged Areas | Maximum Hourly Rainfall (mm/h) | Maximum Daily Rainfall (mm/day) |
---|---|---|---|

Typhoon | |||

Typhoon Mindulle | 0 | 48.4 | 281.6 |

Typhoon Haitang | 60 | 18.0 | 171.5 |

Typhoon Kalmaegi | 0 | 67.5 | 295.5 |

Typhoon Sinlaku | 27 | 27.6 | 299.5 |

Typhoon Morakot | 34 | 41.4 | 329.1 |

Typhoon Saola | 39 | 16.9 | 170.1 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Lin, J.-Y.; Chao, J.-C.; Wu, C.-L.
Hazard Assessment of Potential Large-Scale Landslides in the Watershed of the Chenyulan River. *Water* **2022**, *14*, 3692.
https://doi.org/10.3390/w14223692

**AMA Style**

Lin J-Y, Chao J-C, Wu C-L.
Hazard Assessment of Potential Large-Scale Landslides in the Watershed of the Chenyulan River. *Water*. 2022; 14(22):3692.
https://doi.org/10.3390/w14223692

**Chicago/Turabian Style**

Lin, Ji-Yuan, Jen-Chih Chao, and Cheng-Lin Wu.
2022. "Hazard Assessment of Potential Large-Scale Landslides in the Watershed of the Chenyulan River" *Water* 14, no. 22: 3692.
https://doi.org/10.3390/w14223692