The improvement of large-scale-region landslide susceptibility mapping accuracy by transfer learning

Abstract

Machine-learning methodologies have increasingly been embraced in landslide susceptibility assessment. However, the considerable time and financial burdens of landslide inventories often result in persistent data scarcity, which frequently impedes the generation of accurate and informative landslide susceptibility maps. Addressing this challenge, this study compiled a nationwide dataset and developed a transfer learning-based model to evaluate landslide susceptibility in the Chongqing region specifically. Notably, the proposed model, calibrated with the warmup-cosine annealing (WCA) learning rate strategy, demonstrated remarkable predictive capabilities, particularly in scenarios marked by data limitations and when training data were normalized using parameters from the source region. This is evidenced by the area under the receiver operating characteristic curve (AUC) values, which exhibited significant improvements of 51.00%, 24.40% and 2.15%, respectively, compared to a deep learning model, in contexts where only 1%, 5% and 10% of data from the target region were used for retraining. Simultaneously, there were reductions in loss of 16.12%, 27.61% and 15.44%, respectively, in these instances.

Keywords

data-limited cases / transfer learning / landslide susceptibility / machine learning / normalization based on the parameters of the source domain

References

[1] StottP. How climate change affects extreme weather events [J]. Science, 2016, 352(6293): 1517-1518

[2] ZhangJ-Z, PhoonK K, ZhangD-M, et al. . Deep learning-based evaluation of factor of safety with confidence interval for tunnel deformation in spatially variable soil [J]. Journal of Rock Mechanics and Geotechnical Engineering, 2021, 13(6): 1358-1367

[3] WangZ-L, ShangY-Q, SunH-Y. Optimal location and effect judgment on drainage tunnels for landslide prevention [J]. Journal of Central South University, 2013, 20(7): 2041-2053

[4] SunS-R, LiuY, HaoS-F, et al. . Mechanical properties and microscopic failure process of exemplary argillaceous interlayer from East China [J]. Journal of Central South University, 2022, 29(6): 1973-1986

[5] ToniniM, PecoraroG, RomaillerK, et al. . Spatiotemporal cluster analysis of recent Italian landslides [J]. Georisk: Assessment and Management of Risk for Engineered Systems and Geohazards, 2022, 16(3): 536-554

[6] WangY, FangZ-C, WangM, et al. . Comparative study of landslide susceptibility mapping with different recurrent neural networks [J]. Computers & Geosciences, 2020, 138: 104445

[7] DuB-W, ZhaoZ-R, HuX, et al. . Landslide susceptibility prediction based on image semantic segmentation [J]. Computers & Geosciences, 2021, 155104860

[8] BrabbE. Innovative approaches to landslide hazard mapping [J]. Proc. 4th Int. Symp. Landslides, 1984

[9] GuzzettiF, ReichenbachP, ArdizzoneF, et al. . Estimating the quality of landslide susceptibility models [J]. Geomorphology, 2006, 81(1–2): 166-184

[10] CorominasJ, VanW C, FrattiniP, et al. . Recommendations for the quantitative analysis of landslide risk [J]. Bulletin of Engineering Geology and the Environment, 2014, 73(2): 209-263

[11] ZhangW-G, LiH-R, HanL, et al. . Slope stability prediction using ensemble learning techniques: A case study in Yunyang County, Chongqing, China [J]. Journal of Rock Mechanics and Geotechnical Engineering, 2022, 14(4): 1089-1099

[12] MaoY-M, ZhangM-S, WangG-L, et al. . Landslide hazards mapping using uncertain Naïve Bayesian classification method [J]. Journal of Central South University, 2015, 22(9): 3512-3520

[13] WangH-J, ZhangL-M, LuoH-Y, et al. . AI-powered landslide susceptibility assessment in Hong Kong [J]. Engineering Geology, 2021, 288106103

[14] ZhaoY, WangR, JiangY-J, et al. . GIS-based logistic regression for rainfall-induced landslide susceptibility mapping under different grid sizes in Yueqing, Southeastern China [J]. Engineering Geology, 2019, 259105147

[15] SunD-L, XuJ-H, WenH-J, et al. . Assessment of landslide susceptibility mapping based on Bayesian hyperparameter optimization: A comparison between logistic regression and random forest [J]. Engineering Geology, 2021, 281105972

[16] Goyes-PeñafielP, Hernandez-RojasA. Landslide susceptibility index based on the integration of logistic regression and weights of evidence: A case study in Popayan, Colombia [J]. Engineering Geology, 2021, 280105958

[17] HuangF-M, YinK-L, HuangJ-S, et al. . Landslide susceptibility mapping based on self-organizingmap network and extreme learning machine [J]. Engineering Geology, 2017, 22311-22

[18] AhmadH I, TayL T, Mohamad-SalehJ. Spatial landslide susceptibility modelling using metaheuristic-based machine learning algorithms [J]. Engineering with Computers, 2023, 39(1): 867-891

[19] ZhangJ-Z, HuangH-W, ZhangD-M, et al. . Experimental study of the coupling effect on segmental shield tunnel lining under surcharge loading and excavation unloading [J]. Tunnelling and Underground Space Technology, 2023, 140: 105199

[20] Van WestenC J, Van AschT W J, SoetersR. Landslide hazard and risk zonation—why is it still so difficult? [J]. Bulletin of Engineering Geology and the Environment, 2006, 65(2): 167-184

[21] ZhangJ-Z, HuangH-W, ZhangD-M, et al. . Simplified methods of tunnel structure deformation analysis considering spatial variability of soil properties [J]. Chine. J. Geotech. Eng., 2022, 44(1): 134-143(in Chinese)

[22] Pires De LimaR, MarfurtK. Convolutional neural network for remote-sensing scene classification: Transfer learning analysis [J]. Remote Sensing, 2019, 12(1): 86

[23] ZhangZ-M, TianR-R, SheronyR, et al. . Attention-based interrelation modeling for explainable automated driving [J]. IEEE Transactions on Intelligent Vehicles, 2023, 821564-1573

[24] FangM, YinJ, ZhuX-G. Transfer learning across networks for collective classification [C]. 2013 IEEE 13th International Conference on Data Mining, 2013, Dallas, TX, USA, IEEE: 161-170 December 7–10, 2013

[25] WeiF-M, ZhangJ-P, ChuY, et al. . FSFP: Transfer learning from long texts to the short [J]. Applied Mathematics & Information Sciences, 2014, 8(4): 2033-2040

[26] AiX, SunB-T, ChenX-Z. Construction of small sample seismic landslide susceptibility evaluation model based on Transfer Learning: A case study of Jiuzhaigou earthquake [J]. Bulletin of Engineering Geology and the Environment, 2022, 81(3): 116

[27] WangH-J, WangL, ZhangL-M. Transfer learning improves landslide susceptibility assessment [J]. Gondwana Research, 2023, 123: 238-254

[28] FuZ-Y, LiC-D, YaoW-M. Landslide susceptibility assessment through TrAdaBoost transfer learning models using two landslide inventories [J]. CATENA, 2023, 222106799

[29] WenB-P, WangS-J, WangE-Z, et al. . Characteristics of rapid giant landslides in China [J]. Landslides, 2004, 1(4): 247-261

[30] ZhouC, HuY-J, XiaoT, et al. . Analytical model for reinforcement effect and load transfer of pre-stressed anchor cable with bore deviation [J]. Construction and Building Materials, 2023, 379131219

[31] MathewJ, JhaV K, RawatG S. Landslide susceptibility zonation mapping and its validation in part of Garhwal Lesser Himalaya, India, using binary logistic regression analysis and receiver operating characteristic curve method [J]. Landslides, 2009, 6(1): 17-26

[32] YangC-M, ChangJ M, HungC Y, et al. . Life span of a landslide dam on mountain valley caught on seismic signals and its possible early warnings [J]. Landslides, 2022, 19(3): 637-646

[33] ZhangH-J, SongY-X, XuS-L, et al. . Combining a class-weighted algorithm and machine learning models in landslide susceptibility mapping: A case study of Wanzhou section of the Three Gorges Reservoir, China [J]. Computers & Geosciences, 2022, 158104966

[34] NianT-K, WuH, LiD-Y, et al. . Experimental investigation on the formation process of landslide dams and a criterion of river blockage [J]. Landslides, 2020, 17(11): 2547-2562

[35] YangB-B, YinK-L, LacasseS, et al. . Time series analysis and long short-term memory neural network to predict landslide displacement [J]. Landslides, 2019, 16(4): 677-694

[36] ZhangW-G, LiH-R, TangL-B, et al. . Displacement prediction of Jiuxianping landslide using gated recurrent unit (GRU) networks [J]. Acta Geotechnica, 2022, 17(4): 1367-1382

[37] ZhangW-G, WuC-Z, TangL-B, et al. . Efficient time-variant reliability analysis of Bazimen landslide in the Three Gorges Reservoir Area using XGBoost and LightGBM algorithms [J]. Gondwana Research, 2023, 12341-53

[38] WangL, WuC-Z, YangZ-Y, et al. . Deep learning methods for time-dependent reliability analysis of reservoir slopes in spatially variable soils [J]. Computers and Geotechnics, 2023, 159: 105413

[39] ShenS S P, TafollaN, SmithT M, et al. . Multivariate regression reconstruction and its sampling error for the quasiglobal annual precipitation from 1900 to 2011 [J]. Journal of the Atmospheric Sciences, 2014, 71(9): 3250-3268

[40] YanL, XuW-Y, WangH-L, et al. . Drainage controls on the Donglingxing landslide (China) induced by rainfall and fluctuation in reservoir water levels [J]. Landslides, 2019, 16(8): 1583-1593

[41] PanS J J, YangQ. A Survey on Transfer Learning [J]. IEEE Transactions on Knowledge and Data Engineering, 2010, 22(10): 1345-1359

[42] VrbančIčG, PodgorelecV. Transfer learning with adaptive fine-tuning [J]. IEEE Access, 2020, 8196197-196211

[43] ShaoS-Y, McaleerS, YanR-Q, et al. . Highly accurate machine fault diagnosis using deep transfer learning [J]. IEEE Transactions on Industrial Informatics, 2019, 15(4): 2446-2455

[44] GaoY-Q, MosalamK M. Deep transfer learning for image-based structural damage recognition [J]. Computer-Aided Civil and Infrastructure Engineering, 2018, 33(9): 748-768

[45] PathakY, ShuklaP K, TiwariA, et al. . Deep transfer learning based classification model for COVID-19 disease [J]. IRBM, 2022, 43(2): 87-92

[46] LianJ, DingX-M, ZhangL. Shaking table test on seismic response of an accumulation landslide reinforced by pile-plate retaining wall based on the time-frequency analysis method [J]. Journal of Central South University, 2023, 30(5): 1710-1721

[47] ReichenbachP, RossiM, MalamudB D, et al. . A review of statistically-based landslide susceptibility models [J]. Earth-Science Reviews, 2018, 180: 60-91

[48] NhuV H, ShirzadiA, ShahabiH, et al. . Shallow landslide susceptibility mapping: A comparison between logistic model tree, logistic regression, Naïve Bayes tree, artificial neural network, and support vector machine algorithms [J]. International Journal of Environmental Research and Public Health, 2020, 17(8): 2749

[49] ZhouX-Z, WenH-J, ZhangY-L, et al. . Landslide susceptibility mapping using hybrid random forest with GeoDetector and RFE for factor optimization [J]. Geoscience Frontiers, 2021, 125101211

[50] WongE, RiceL, KolterZ. Overfitting in adversarially robust deep learning [J]. International Conference on Machine Learning, 2020, 119: 8093-8104

[51] IoffeS, SzegedyC. Batch normalization: Accelerating deep network training by reducing internal covariate shift [J]. 32nd International Conference on Machine Learning, ICML 2015, 2015, 1448-456

[52] JiangJ, ZhengJ-T, YangY, et al. . CTCPPre: A prediction method for accepted pull requests in GitHub [J]. Journal of Central South University, 2020, 27(2): 449-468

[53] ZhangB-Y, LiM-Y, LiW-X, et al. . Machine learning strategies for lithostratigraphic classification based on geochemical sampling data: A case study in area of Chahanwusu River, Qinghai Province, China [J]. Journal of Central South University, 2021, 28(5): 1422-1447