A comparative machine learning approach to identify landslide triggering factors in northern Chilean Patagonia

Morales, Bastian; Lizama, Elizabet; Somos‐Valenzuela, Marcelo; Lillo‐Saavedra, Mario; Chen, Ningsheng; Fustos, Ivo

doi:10.1007/s10346-021-01675-9

Cited by 15 publications

(17 citation statements)

References 116 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The challenge many researchers face is selecting the most appropriate method. Thus, comparative analysis of the predictive performance of different machine-learning methods is a major topic in the landslide literature [115,116]. With a desire to produce a landslide susceptibility map with high prediction accuracy, we compared the predictive performance of four machine-learning methods.…”

Section: Discussionmentioning

confidence: 99%

A Robust Deep-Learning Model for Landslide Susceptibility Mapping: A Case Study of Kurdistan Province, Iran

Ghasemian

Shahabi

Shirzadi

et al. 2022

Sensors

View full text Add to dashboard Cite

We mapped landslide susceptibility in Kamyaran city of Kurdistan Province, Iran, using a robust deep-learning (DP) model based on a combination of extreme learning machine (ELM), deep belief network (DBN), back propagation (BP), and genetic algorithm (GA). A total of 118 landslide locations were recorded and divided in the training and testing datasets. We selected 25 conditioning factors, and of these, we specified the most important ones by an information gain ratio (IGR) technique. We assessed the performance of the DP model using statistical measures including sensitivity, specificity, accuracy, F1-measure, and area under-the-receiver operating characteristic curve (AUC). Three benchmark algorithms, i.e., support vector machine (SVM), REPTree, and NBTree, were used to check the applicability of the proposed model. The results by IGR concluded that of the 25 conditioning factors, only 16 factors were important for our modeling procedure, and of these, distance to road, road density, lithology and land use were the four most significant factors. Results based on the testing dataset revealed that the DP model had the highest accuracy (0.926) of the compared algorithms, followed by NBTree (0.917), REPTree (0.903), and SVM (0.894). The landslide susceptibility maps prepared from the DP model with AUC = 0.870 performed the best. We consider the DP model a suitable tool for landslide susceptibility mapping.

show abstract

Section: Discussionmentioning

confidence: 99%

A Robust Deep-Learning Model for Landslide Susceptibility Mapping: A Case Study of Kurdistan Province, Iran

Ghasemian

Shahabi

Shirzadi

et al. 2022

Sensors

View full text Add to dashboard Cite

show abstract

“…Instead, dominant forest disturbance comes from frequent windstorms (Parra et al, 2021), and less frequent earthquakes (Sepulveda et al, 2010) and volcanic eruptions (Mohr et al, 2017) the latter both are a result of active 105 subduction and intra-arc strike-slip motion along the Liquiñe-Ofqui Fault zone (Cembrano et al, 1996). All these disturbances have in common that they may trigger mostly shallow landslides (Korup et al, 2019;Morales et al, 2021). As landslides are largely, though not entirely, controlled by topography and thus independent from vegetation composition (Veblen and Alaback, 1996;DellaSala, 2011;Buma and Johnson, 2015;Parra et al, 2021), we assume them as largely constrained (and predictable) by topography which in turn allows us to explore the efficacy of disturbance-driven surface processes on carbon cycling without too 110 many assumptions that are often hard to justify.…”

Section: Pumalin Czo -Scope and Instrumentationmentioning

confidence: 99%

“…Following the approach by Dietze et al (2020) we automatically picked potential landslide events by a classic STA-LTA picker (short term window 0.5 s, long term window 300 s, on-ratio 3, off-ratio 1.1). For all detected events we calculated the spectrograms of all seismic stations and located the event using the signal migration technique (Burtin et 235 We emphasize that this particular hillslope event occurred during presumably 'dry' antecedent conditions as it happened during 240 the transition from Austral summer to Austral autumn (2022-03-22) and that landslide activity is concentrated during the rainy season (Morales et al, 2021). Our CZO may assist in improving rainfall duration-intensity thresholds (Guzzetti et al, 2008) , thus providingtogether with the wind dataa contribution for early warning, considering known limitations in the spatial variability of 245 storms (Fustos-Toribio et al, 2022).…”

Section: Carbon Mobilization By Landsliding 230mentioning

confidence: 99%

“…Landscape disturbances do not only modify the efficacy of erosion, as they may transiently shift systems from a quasi-steady into higher states (Vanacker et al, 2007), but may also likely cause a (temporal) switch from carbon sink to source functioning (e.g., Mohr et al, 2017). Tectonic disturbances, such as earthquakes (Sepúlveda et al, 2010) and explosive volcanic eruptions (Morales et al, 2021;Korup et al, 2019), are arguably the most powerful, yet low frequency erosion drivers in the Patagonian CTRs. 320…”

Section: Understand Landscape Disturbance Feedbacks In a Holistic Eco...mentioning

confidence: 99%

See 1 more Smart Citation

Ideas and Perspectives: Sensing Energy and Matter fluxes in a biota dominated Patagonian landscape through environmental seismology – Introducing the Pumalín Critical Zone Observatory

Mohr

Dietze

Tolorza

et al. 2023

Preprint

View full text Add to dashboard Cite

Abstract. The Coastal Temperate Rainforests (CTRs) of Chilean Patagonia are a valuable forest biome on Earth given its prominent role for biogeochemical cycling, ecological value and dynamic of surface processes. The Patagonian CTRs are amongst the most carbon rich biomes on Earth. Together with frequent landscape disturbances, these forests potentially allow for episodic, massive release of carbon into the atmosphere. We argue that despite their particular biogeographic, geochemical, and ecological role, the Patagonian CTRs in particular, and the global CTRs in general, are not adequately represented in the current catalogue listing critical zone observatories (CZO). Here, we present the Pumalin CZO as the first of its kind, located in the Pumalin National park in northern Chilean Patagonia. We have identified four core research themes for the Pumalin CZO around which our activities circle in an integrative, quantitative, generic approach using a range of emerging techniques. We aim to gain quantitative understanding of these topics: (1) carbon sink functioning, (2) biota-driven landscape evolution, (3) water, biogeological and energy fluxes, and (4) disturbance regime understanding. Our findings highlight the multitude of active functions that trees in particular, and forests in general, may have on the entire chain of carbon cycling. This highlights the importance of an integrated approach, i.e., ‘one physical system’, as proposed by Richter and Billings (2015), and accounting for the recent advances in pushing nature conservation along the Chilean coast.

show abstract

“…Reichenbach et al (2018) found that a total of 596 conditioning factors were considered in the existing work, with an average of nine condition factors in each model. In the existing research, the selection of condition factors is mostly determined by expert experience, which is very subjective (Bourenane et al, 2015;Morales et al, 2021;Zhao et al, 2021). The current research lacks a general framework to objectively select the condition factors.…”

Section: Introductionmentioning

confidence: 99%

Improving Geospatial Agreement by Hybrid Optimization in Logistic Regression-Based Landslide Susceptibility Modelling

Sun

Wen

et al. 2021

Front. Earth Sci.

View full text Add to dashboard Cite

This study aims to develop a logistic regression model of landslide susceptibility based on GeoDetector for dominant-factor screening and 10-fold cross validation for training sample optimization. First, Fengjie county, a typical mountainous area, was selected as the study area since it experienced 1,522 landslides from 2001 to 2016. Second, 22 factors were selected as the initial conditioning factors, and a geospatial database was established with a grid of 30 m precision. Factor detection of the geographic detector and the stepwise regression method included in logistic regression were used to screen out the dominant factors from the database. Then, based on the sample dataset with a 1:10 ratio of landslides and nonlandslides, 10-fold cross validation was used to select the optimized sample to train the logistic regression model of landslide susceptibility in the study area. Finally, the accuracy and efficiency of the two models before and after screening out the dominant factors were evaluated and compared. The results showed that the total accuracy of the two models was both more than 0.9, and the area under the curve value of the receiver operating characteristic curve was more than 0.8, indicating that the models before and after screening factor both had high reliability and good prediction ability. Besides, the screened factors had an active leading role in the geospatial distribution of the historical landslide, indicating that the screened dominant factors have individual rationality. Improving the geospatial agreement between landslide susceptibility and actual landslide-prone by the screening of dominant factors and the optimization of the training samples, a simple, efficient, and reliable logistic-regression–based landslide susceptibility model can be constructed.

show abstract

A comparative machine learning approach to identify landslide triggering factors in northern Chilean Patagonia

Cited by 15 publications

References 116 publications

A Robust Deep-Learning Model for Landslide Susceptibility Mapping: A Case Study of Kurdistan Province, Iran

A Robust Deep-Learning Model for Landslide Susceptibility Mapping: A Case Study of Kurdistan Province, Iran

Ideas and Perspectives: Sensing Energy and Matter fluxes in a biota dominated Patagonian landscape through environmental seismology – Introducing the Pumalín Critical Zone Observatory

Improving Geospatial Agreement by Hybrid Optimization in Logistic Regression-Based Landslide Susceptibility Modelling

Contact Info

Product

Resources

About