Random Forests for Landslide Prediction in Tsengwen River Watershed, Central Taiwan

Youg-Sin Cheng; Teng-To Yu; Nguyen-Thanh Son

doi:10.3390/rs13020199

Random Forests for Landslide Prediction in Tsengwen River Watershed, Central Taiwan

Remote Sensing ◽

10.3390/rs13020199 ◽

2021 ◽

Vol 13 (2) ◽

pp. 199

Author(s):

Youg-Sin Cheng ◽

Teng-To Yu ◽

Nguyen-Thanh Son

Keyword(s):

Random Forests ◽

Influential Factors ◽

Landslide Prediction ◽

Study Region ◽

River Watershed ◽

Landslide Modeling ◽

Predictive Variables ◽

Landslide Hazards ◽

Machine Learning Approach ◽

Central Taiwan

Landslides have been identified as one of the costliest and deadliest natural disasters, causing tremendous damage to humans and societies. Information regarding the spatial extent of landslides is thus important to allow officials to devise successful strategies to mitigate landslide hazards. This study aims to develop a machine-learning approach for predicting landslide areas in the Tsengwen River Watershed (TRW), which is one of the most landslide-prone areas in Central Taiwan. Various spatial datasets were collected from 2009 to 2015 to derive 36 predictive variables used for landslide modeling with random forests (RF). The results of landslide prediction, compared with ground reference data, indicated an overall accuracy of 91.4% and Kappa coefficient of 0.83, respectively. The findings achieved from estimates of predictor importance also indicated to officials that the land-use/land-cover (LULC) type, distance to previous landslides, distance to roads, bank erosion, annual groundwater recharge, geological line density, aspect, and slope are the most influential factors that trigger landslides in the study region.

Download Full-text

Predictors of remission from body dysmorphic disorder after internet-delivered cognitive behavior therapy: a machine learning approach

10.31234/osf.io/eqcdx ◽

2019 ◽

Author(s):

Oskar Flygare ◽

Jesper Enander ◽

Erik Andersson ◽

Brjánn Ljótsson ◽

Volen Z Ivanov ◽

...

Keyword(s):

Machine Learning ◽

Logistic Regression ◽

Random Forests ◽

Clinical Utility ◽

Body Dysmorphic Disorder ◽

Prediction Models ◽

Behavioral Therapy ◽

Learning Approach ◽

Learning Approaches ◽

Machine Learning Approach

**Background:** Previous attempts to identify predictors of treatment outcomes in body dysmorphic disorder (BDD) have yielded inconsistent findings. One way to increase precision and clinical utility could be to use machine learning methods, which can incorporate multiple non-linear associations in prediction models. **Methods:** This study used a random forests machine learning approach to test if it is possible to reliably predict remission from BDD in a sample of 88 individuals that had received internet-delivered cognitive behavioral therapy for BDD. The random forest models were compared to traditional logistic regression analyses. **Results:** Random forests correctly identified 78% of participants as remitters or non-remitters at post-treatment. The accuracy of prediction was lower in subsequent follow-ups (68%, 66% and 61% correctly classified at 3-, 12- and 24-month follow-ups, respectively). Depressive symptoms, treatment credibility, working alliance, and initial severity of BDD were among the most important predictors at the beginning of treatment. By contrast, the logistic regression models did not identify consistent and strong predictors of remission from BDD. **Conclusions:** The results provide initial support for the clinical utility of machine learning approaches in the prediction of outcomes of patients with BDD. **Trial registration:** ClinicalTrials.gov ID: NCT02010619.

Download Full-text

The value of the post-captopril aldosterone/renin ratio for the diagnosis of primary aldosteronism and the influential factors: A meta-analysis

Journal of the Renin-Angiotensin-Aldosterone System ◽

10.1177/1470320320972032 ◽

2020 ◽

Vol 21 (4) ◽

pp. 147032032097203

Author(s):

Qiao Xiang ◽

Wen Wang ◽

Tao Chen ◽

Kai Yu ◽

Qianrui Li ◽

...

Keyword(s):

Sensitivity And Specificity ◽

Primary Aldosteronism ◽

Meta Analysis ◽

Challenge Test ◽

Influential Factors ◽

Time Interval ◽

Study Region ◽

Subgroup Analyses ◽

Study Population ◽

Meta Regression

Objective: The procedure for the captopril challenge test (CCT) in diagnosing primary aldosteronism (PA) is not standardized. We performed a meta-analysis to evaluate the controversial diagnostic value and influential factors of the post-captopril aldosterone/renin ratio (ARR). Methods: We searched literature in databases for eligible studies (until October 1, 2020). We extracted information regarding study and patient characteristics, CCT methods, outcome data. We pooled studies using the random-effect model. We performed meta-regression and six pre-specified subgroup analyses to explore heterogeneity. Results: Nineteen studies involving 4568 subjects were included. The pooled sensitivity and specificity were 0.825 (95% CI 0.804–0.844) and 0.919 (95% CI 0.908–0.928). The area under the summary receiver operating characteristic curve was 0.9487 (95% CI 0.9207–0.9767). Meta-regression revealed that heterogeneity might derive from time interval ( p = 0.0117) and study population ( p = 0.0033). Subgroup analyses showed significant differences between the subgroups stratified by the dose, posture, study region, time interval, cut-off value and study population for sensitivity and/or specificity ( p < 0.05). Conclusion: Post-captopril ARR is comparably valuable for diagnosing PA at cut-offs from 12.0 to 50.0. Conducting the CCT in the supine position with 25 mg of captopril may attain greater sensitivity. Conducting the CCT in the seated position with 50 mg of captopril may attain greater specificity. A 90-min time interval may perform best in both the sensitivity and specificity.

Download Full-text

Semi-continuous measurements of water-soluble inorganic ions over a high-altitude background site in central Taiwan

10.5194/egusphere-egu2020-12251 ◽

2020 ◽

Author(s):

Atinderpal Singh ◽

We-Ren Chen ◽

Chung-Te Lee

Keyword(s):

High Altitude ◽

Mass Concentration ◽

Sampling Site ◽

Inorganic Ions ◽

Water Soluble ◽

Study Region ◽

Backward Trajectories ◽

Continuous Measurements ◽

Water Soluble Inorganic Ions ◽

Central Taiwan

To better understand the abundance and sources of water-soluble inorganic ions (WSIIs), semi-continuous measurements of WSIIs were performed during autumn 2015 and spring 2016 at a high-altitude background station (2,862 m above sea level) on the summit of Mt. Lulin in central Taiwan. During autumn, the mass concentration of PM2.5, major WSIIs, and CO increased significantly from 12:00 to 18:00 hrs local standard time (LST), whereas the visibility and concentration of O3 decreased at the same time. The backward trajectories analyses showed that the sampling site was under the influence of lifted air masses by the upslope wind from 12:00 to 18:00 hrs. Thus the mountain-valley (M-V) circulation could be the major driving force for the observed aerosol diurnal patterns over the study region during autumn. In sharp contrast to autumn, five high aerosol loading events were observed during spring with each event lasting for a few days. These events were synchronized with the long-range transport of biomass burning (BB) smoke emissions from the Indochina region, as revealed from the fire count map and backward trajectories. The plumes appear to mask their characteristic diurnal features that are driven by the local M-V circulation. These plumes also affected the acidity of ambient aerosol. During BB events, aerosol was found to be relatively more alkaline in nature as revealed by higher molar ratio of [NH4+]calc/[NH4+]meas during BB events (0.88 &#177; 0.25) than that of the whole spring season (0.81 &#177; 0.33). The third BB event (BB3), March 29 to April 04, 2016, was the most prominent one among all BB events. During BB3, the mass concentration of PM2.5, NH4+, K+, NO3- and SO42- increased from 8.3 to 29, 0.01 to 2.0, 0.02 to 0.4, 0.01 to 1.6, and 0.4 to 4.1 &#956;g m-3, respectively as compared to before the event. A fog event (March 31; 0:00 to 10:00 LST) was also observed during the BB3 event that decreased the mass concentration of all the species significantly. It suggested that aerosol scavenging and cloud-active processing may occur in this fog event.

Download Full-text

Earthquake-triggered landslides in southwest China

Natural Hazards and Earth System Science ◽

10.5194/nhess-12-351-2012 ◽

2012 ◽

Vol 12 (2) ◽

pp. 351-363 ◽

Cited By ~ 21

Author(s):

X. L. Chen ◽

Q. Zhou ◽

H. Ran ◽

R. Dong

Keyword(s):

Southwest China ◽

Earthquake Magnitude ◽

The Tibetan Plateau ◽

Study Region ◽

Strong Earthquakes ◽

Landslide Hazards ◽

Geological Conditions ◽

Southeastern Margin ◽

Fault Belt ◽

The Relationship

Abstract. Southwest China is located in the southeastern margin of the Tibetan Plateau and it is a region of high seismic activity. Historically, strong earthquakes that occurred here usually generated lots of landslides and brought destructive damages. This paper introduces several earthquake-triggered landslide events in this region and describes their characteristics. Also, the historical data of earthquakes with a magnitude of 7.0 or greater, having occurred in this region, is collected and the relationship between the affected area of landslides and earthquake magnitude is analysed. Based on the study, it can be concluded that strong earthquakes, steep topography as well as fragile geological environment, are the main reasons responsible for serious landslides in southwest China. At the same time, it is found that the relationship between the area affected by landslides and the earthquake magnitude in this region are consistent with what has been obtained worldwide. Moreover, in this paper, it is seen that the size of the areas affected by landslides change enormously even under the same earthquake magnitude in the study region. While at the same tectonic place or fault belt, areas affected by landslides presented similar outline and size. This means that local geological conditions and historical earthquake background have an important influence on landslides distribution, and they should be considered when assessing earthquake-triggered landslide hazards at Grade 1 according to ISSMGE.

Download Full-text

Landslide susceptibility mapping using Forest by Penalizing Attributes (FPA) algorithm based machine learning approach

Vietnam Journal of Earth Sciences ◽

10.15625/0866-7187/42/3/15047 ◽

2020 ◽

Vol 42 (3) ◽

Cited By ~ 1

Author(s):

Tran Van Phong ◽

Hai-Bang Ly ◽

Phan Trong Trinh ◽

Indra Prakash

Keyword(s):

Machine Learning ◽

Landslide Susceptibility ◽

Susceptibility Mapping ◽

Landslide Susceptibility Mapping ◽

Susceptibility Map ◽

Conditioning Factors ◽

Environmental Conditioning ◽

Landslide Modeling ◽

Machine Learning Approach ◽

First Time

Landslide susceptibility mapping is a helpful tool for assessment and management of landslides of an area. In this study, we have applied first time Forest by Penalizing Attributes (FPA) algorithm-based Machine Learning (ML) approach for mapping of landslide susceptibility at Muong Lay district (Vietnam). For this aim, 217 historical landslides locations were identified and analyzed for the development of FPA model and generation of susceptibility map. Nine landslide topographical and geo-environmental conditioning factors (curvature, geology/lithology, aspect, distance from faults, rivers and roads, weathering crust, slope, and deep division) were utilized to construct the training and validating datasets for landslide modeling. Different quantitative statistical indices including Area Under the Receiver Operating Characteristic (ROC) curve (AUC) were used to evaluate the performance of the model. The results indicate that the predictive capability of the FPA is very good for landslide susceptibility mapping on both training (AUC = 0.935) and validating (AUC = 0.882) datasets. Thus, the novel FPA based ML model can be utilized for the development of accurate landslide susceptibility map of the study area and this approach can also be applied in other landslide prone areas.

Download Full-text

A Machine Learning Approach to Defect Parameters Extraction: Using Random Forests to Inverse the Shockley-Read-Hall Equation

2019 IEEE 46th Photovoltaic Specialists Conference (PVSC) ◽

10.1109/pvsc40753.2019.8980717 ◽

2019 ◽

Author(s):

Yoann Buratti ◽

Josef Dick ◽

Quoc Le Gia ◽

Ziv Hameiri

Keyword(s):

Machine Learning ◽

Random Forests ◽

Learning Approach ◽

Parameters Extraction ◽

Machine Learning Approach

Download Full-text

Assessment of atmospheric conditions over the Hong Thai Binh river watershed by means of dynamically downscaled ERA-20C reanalysis data

Journal of Water and Climate Change ◽

10.2166/wcc.2018.291 ◽

2018 ◽

Vol 11 (2) ◽

pp. 540-555 ◽

Cited By ~ 1

Author(s):

C. Ho ◽

A. Nguyen ◽

A. Ercan ◽

M. L. Kavvas ◽

V. Nguyen ◽

...

Keyword(s):

Temporal Resolution ◽

Wrf Model ◽

Reanalysis Data ◽

Observation Data ◽

Human Society ◽

Atmospheric Conditions ◽

Study Region ◽

River Watershed ◽

Environmentally Sustainable ◽

Atmospheric Data

Abstract Long-term, high spatial and temporal resolution of atmospheric data is crucial for the purpose of reducing the effects of hydro-meteorological risks on human society in an economically and environmentally sustainable manner. However, such information usually is limited in transboundary regions due to different governmental policies, and to conflicts in the sharing of data. In this study, high spatial and temporal resolution atmospheric data were reconstructed by means of the Weather Research and Forecasting Model-WRF with input provided from the global atmospheric reanalysis of the 20th century (ERA-20C) over the Hong-Thai Binh River watershed (H-TBRW). The WRF model was implemented over the physical boundaries of the study region based on ERA-20C reanalysis data and was configured based on existing ground observation data in Vietnam's territories, and the global Aphrodite precipitation data. With the validated WRF model for H-TBRW, the reconstructed atmospheric data were first reconstructed for 1950–2010, and then were evaluated by time series and spatial analysis methods. The results of this study suggest no significant trend in the annual accumulated precipitation depth, while there were upward trends in annual temperature at both the point and watershed scale. Furthermore, the results confirm that topographic conditions have significant effects on the climatic system such as on precipitation and temperature.

Download Full-text

Spatial pattern analysis of landslide using landscape metrics and logistic regression: a case study in Central Taiwan

Hydrology and Earth System Sciences Discussions ◽

10.5194/hessd-7-3423-2010 ◽

2010 ◽

Vol 7 (3) ◽

pp. 3423-3451 ◽

Cited By ~ 10

Author(s):

Y.-P. Lin ◽

H.-J. Chu ◽

C.-F. Wu

Keyword(s):

Logistic Regression ◽

Spatial Pattern Analysis ◽

Frequency Of Occurrence ◽

Explanatory Variables ◽

Landslide Hazards ◽

Richter Scale ◽

Multi Temporal ◽

Mean Size ◽

Central Taiwan

Abstract. The Chi-Chi Earthquake of September 1999 in Central Taiwan registered a moment magnitude MW of 7.6 on the Richter scale, causing widespread landslides. Subsequent typhoons associated with heavy rainfalls triggered the landslides. The study investigates multi-temporal landslide images from spatial analysis between 1996 and 2005 in the Chenyulan Watershed, Taiwan. Spatial patterns in various landslide frequencies were detected using landscapes metrics. The logistic regression results indicate that frequency of occurrence is an important factor in assessing landslide hazards. Low-occurrence landslides sprawl the catchment while the sustained (frequent) landslide areas cluster near the ridge as well as the stream course. From those results, we can infer that landslide area and mean size for each landslide correlates with the frequency of occurrence. Although negatively correlated with frequency in the low-occurrence landslide, the mean size of each landslide is positively related to frequency in the high-occurrence one. Moreover, this study determines the spatial susceptibilities in landslides by performing logistic regression analysis. Results of this study demonstrate that the factors such as elevation, slope, lithology, and vegetation cover are significant explanatory variables. In addition to the various frequencies, the relationships between driving factors and landslide susceptibility in the study area are quantified as well.

Download Full-text

Landslide Susceptibility Evaluation and Management Using Different Machine Learning Methods in The Gallicash River Watershed, Iran

Remote Sensing ◽

10.3390/rs12030475 ◽

2020 ◽

Vol 12 (3) ◽

pp. 475 ◽

Cited By ~ 19

Author(s):

Alireza Arabameri ◽

Sunil Saha ◽

Jagabandhu Roy ◽

Wei Chen ◽

Thomas Blaschke ◽

...

Keyword(s):

Machine Learning ◽

Landslide Susceptibility ◽

Learning Methods ◽

Validation Data ◽

Area Index ◽

Study Region ◽

River Watershed ◽

Machine Learning Methods ◽

Natural Break ◽

Very High

This analysis aims to generate landslide susceptibility maps (LSMs) using various machine learning methods, namely random forest (RF), alternative decision tree (ADTree) and Fisher’s Linear Discriminant Function (FLDA). The results of the FLDA, RF and ADTree models were compared with regard to their applicability for creating an LSM of the Gallicash river watershed in the northern part of Iran close to the Caspian Sea. A landslide inventory map was created using GPS points obtained in a field analysis, high-resolution satellite images, topographic maps and historical records. A total of 249 landslide sites have been identified to date and were used in this study to model and validate the LSMs of the study region. Of the 249 landslide locations, 70% were used as training data and 30% for the validation of the resulting LSMs. Sixteen factors related to topographical, hydrological, soil type, geological and environmental conditions were used and a multi-collinearity test of the landslide conditioning factors (LCFs) was performed. Using the natural break method (NBM) in a geographic information system (GIS), the LSMs generated by the RF, FLDA, and ADTree models were categorized into five classes, namely very low, low, medium, high and very high landslide susceptibility (LS) zones. The very high susceptibility zones cover 15.37% (ADTree), 16.10% (FLDA) and 11.36% (RF) of the total catchment area. The results of the different models (FLDA, RF, and ADTree) were explained and compared using the area under receiver operating characteristics (AUROC) curve, seed cell area index (SCAI), efficiency and true skill statistic (TSS). The accuracy of models was calculated considering both the training and validation data. The results revealed that the AUROC success rates are 0.89 (ADTree), 0.92 (FLDA) and 0.97 (RF) and predication rates are 0.82 (ADTree), 0.79 (FLDA) and 0.98 (RF), which justifies the approach and indicates a reasonably good landslide prediction. The results of the SCAI, efficiency and TSS methods showed that all models have an excellent modeling capability. In a comparison of the models, the RF model outperforms the boosted regression tree (BRT) and ADTree models. The results of the landslide susceptibility modeling could be useful for land-use planning and decision-makers, for managing and controlling the current and future landslides, as well as for the protection of society and the ecosystem.

Download Full-text