Aerial Photogrammetry and Machine Learning Based Regional Landslide Susceptibility Assessment for an Earthquake Prone Area in Turkey

Abstract

Abstract. Landslide is a frequently observed natural phenomenon and a geohazard with destructive effects on economies, society and the environment. Production of up-to-date landslide susceptibility (LS) maps is an essential process for landslide hazard mitigation. Obtaining up-to-date and accurate data for the production of LS maps is also important and this task can be achieved by using aerial photogrammetric techniques, which can produce geospatial data with high resolution. The produced geospatial datasets can be integrated in data-driven methods for obtaining accurate LS maps. In the present study, LS map was produced by using data-driven machine learning (ML) methods, i.e. random forest (RF). An earthquake and landslide prone area from the south-eastern part of Turkey was selected as the study area. Topographical derivatives were extracted from digital surface models (DSMs) produced by using aerial photogrammetric datasets with 30 cm ground sampling distances. The lithological parameters were employed in the study together with an accurate landslide inventory, which were also delineated by using the high-resolution DSMs and orthophotos. The relationships between the landslide occurrence and the pre-defined conditioning factors were analyzed using the frequency ratio (FR) method. The results show that the RF method exhibits high prediction performance in the study area with an area under curve (AUC) value of 0.92.

A Comprehensive Assessment of XGBoost Algorithm for Landslide Susceptibility Mapping in the Upper Basin of Ataturk Dam, Turkey

“…It should also be noted that in the present study, the size of the study area (2718.7 km 2 ) was relatively large, which indicates that the results are promising for regional susceptibility assessment studies. On the other hand, when the LS prediction results from recent ML studies carried out in the East and Southeast Anatolia are compared; Karakas et al [20] obtained AUC values of 0.90 and 0.92 using the RF method in an area of 425 km 2 using very high resolution aerial photogrammetric datasets [20]. Sevgen et al [15] evaluated logistic regression, ANN and RF methods in the close vicinity of a dam reservoir again by using very high resolution aerial photogrammetric datasets; and obtained AUC values of 0.76, 0.84 and 0.95, respectively.…”

“…With the advancements in machine learning (ML) algorithms, geospatial technologies and computational power, production of LS maps have become less challenging. Among the common ML algorithms used for the LS mapping in the literature, the artificial neural networks (ANN), logistic regression (LR), deep learning methods, decision trees, random forest (RF), naïve Bayes tree, fuzzy logic and support vector machine (SVM) can be listed (e.g., see [12][13][14][15][16][17][18][19][20][21][22][23][24][25][26]). The ML techniques can increase the accuracy of the LS maps [23].…”

Frequency Ratio Assessment for Landslides Triggered by 6 February 2023 Kahramanmaras Turkiye Earthquakes Between Golbasi and Erkenek

“…A total of 1.013.390 pixels (9 features with 11.408 landslides and 1.001.982 non-landslide pixels in each with 25 m grid spacing) were analyzed to reveal the characteristics of the landslides (Figure 2). A total of 9 features were categorized depending on the literature to assess the FR (Karakas, 2021).…”

“…The production of high-quality hazard models requires complete and accurate landslide inventories. The Frequency Ratio (FR) model is widely utilized for assessing the correlations between spatial data and criteria by determining the likelihood of occurrence and non-occurrence for each category of conditioning factors (Dao et * Corresponding author al., 2020;Karakas et al, 2021;Pham et al, 2021). This analysis helps to identify the most significant factors contributing to the occurrence of the phenomenon and informs decisions related to risk management and mitigation strategies.…”

On the Effect of Dem Quality for Landslide Susceptibility Mapping

“…Thus, it is prone to landslides. The site was previously assessed for its landslide inventory (Karakas et al, 2021a) and the LSM production accuracy by using different machine learning (ML) methods (Karakas et al, 2021b(Karakas et al, , 2022. The location of the study area is illustrated in Figure 1.…”