Uncertainties of prediction accuracy in shallow landslide modeling

Sample size and raster resolution

Ataollah Shirzadi, Karim Solaimani, Mahmood Habibnejad Roshan, Ataollah Kavian, Kamran Chapi, Himan Shahabi, Saskia Keesstra, Baharin Bin Ahmad, Dieu Tien Bui*

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

13 Citations (Scopus)

Abstract

Understanding landslide characteristics such as their locations, dimensions, and spatial distribution is of highly importance in landslide modeling and prediction. The main objective of this study was to assess the effect of different sample sizes and raster resolutions in landslide susceptibility modeling and prediction accuracy of shallow landslides. In this regard, the Bijar region of the Kurdistan province (Iran) was selected as a case study. Accordingly, a total of 20 landslide conditioning factors were considered with six different raster resolutions (10 m, 15 m, 20 m, 30 m, 50 m, and 100 m) and four different sample sizes (60/40%, 70/30%, 80/20%, and 90/10%) were investigated. The merit of each conditioning factors was assessed using the Information Gain Ratio (IGR) technique, whereas Alternating decision tree (ADTree), which has been rarely explored for landslide modeling, was used for building models. Performance of the models was assessed using the area under the ROC curve (AUROC), sensitivity, specificity, accuracy, kappa and RMSE criteria. The results show that with increasing the number of training pixels in the modeling process, the accuracy is increased. Findings also indicate that for the sample sizes of 60/40% (AUROC = 0.800) and 70/30% (AUROC = 0.899), the highest prediction accuracy is derived with the raster resolution of 10 m. With the raster resolution of 20 m, the highest prediction accuracy for the sample size of 80/20% (AUROC = 0.871) and 90/10% (AUROC = 0.864). These outcomes provide a guideline for future research enabling researchers to select an optimal data resolution for landslide hazard modeling.

Original languageEnglish
Pages (from-to)172-188
JournalCatena
Volume178
DOIs
Publication statusPublished - 1 Jul 2019

Fingerprint

raster
landslide
prediction
modeling
conditioning
pixel
hazard
spatial distribution

Keywords

  • Alternating decision tree
  • GIS
  • Landslide susceptibility
  • Pixel and sample size
  • Uncertainty

Cite this

Shirzadi, A., Solaimani, K., Roshan, M. H., Kavian, A., Chapi, K., Shahabi, H., ... Bui, D. T. (2019). Uncertainties of prediction accuracy in shallow landslide modeling: Sample size and raster resolution. Catena, 178, 172-188. https://doi.org/10.1016/j.catena.2019.03.017
Shirzadi, Ataollah ; Solaimani, Karim ; Roshan, Mahmood Habibnejad ; Kavian, Ataollah ; Chapi, Kamran ; Shahabi, Himan ; Keesstra, Saskia ; Ahmad, Baharin Bin ; Bui, Dieu Tien. / Uncertainties of prediction accuracy in shallow landslide modeling : Sample size and raster resolution. In: Catena. 2019 ; Vol. 178. pp. 172-188.
@article{b92616a51064402387080b4a8e215c03,
title = "Uncertainties of prediction accuracy in shallow landslide modeling: Sample size and raster resolution",
abstract = "Understanding landslide characteristics such as their locations, dimensions, and spatial distribution is of highly importance in landslide modeling and prediction. The main objective of this study was to assess the effect of different sample sizes and raster resolutions in landslide susceptibility modeling and prediction accuracy of shallow landslides. In this regard, the Bijar region of the Kurdistan province (Iran) was selected as a case study. Accordingly, a total of 20 landslide conditioning factors were considered with six different raster resolutions (10 m, 15 m, 20 m, 30 m, 50 m, and 100 m) and four different sample sizes (60/40{\%}, 70/30{\%}, 80/20{\%}, and 90/10{\%}) were investigated. The merit of each conditioning factors was assessed using the Information Gain Ratio (IGR) technique, whereas Alternating decision tree (ADTree), which has been rarely explored for landslide modeling, was used for building models. Performance of the models was assessed using the area under the ROC curve (AUROC), sensitivity, specificity, accuracy, kappa and RMSE criteria. The results show that with increasing the number of training pixels in the modeling process, the accuracy is increased. Findings also indicate that for the sample sizes of 60/40{\%} (AUROC = 0.800) and 70/30{\%} (AUROC = 0.899), the highest prediction accuracy is derived with the raster resolution of 10 m. With the raster resolution of 20 m, the highest prediction accuracy for the sample size of 80/20{\%} (AUROC = 0.871) and 90/10{\%} (AUROC = 0.864). These outcomes provide a guideline for future research enabling researchers to select an optimal data resolution for landslide hazard modeling.",
keywords = "Alternating decision tree, GIS, Landslide susceptibility, Pixel and sample size, Uncertainty",
author = "Ataollah Shirzadi and Karim Solaimani and Roshan, {Mahmood Habibnejad} and Ataollah Kavian and Kamran Chapi and Himan Shahabi and Saskia Keesstra and Ahmad, {Baharin Bin} and Bui, {Dieu Tien}",
year = "2019",
month = "7",
day = "1",
doi = "10.1016/j.catena.2019.03.017",
language = "English",
volume = "178",
pages = "172--188",
journal = "Catena",
issn = "0341-8162",
publisher = "Elsevier",

}

Shirzadi, A, Solaimani, K, Roshan, MH, Kavian, A, Chapi, K, Shahabi, H, Keesstra, S, Ahmad, BB & Bui, DT 2019, 'Uncertainties of prediction accuracy in shallow landslide modeling: Sample size and raster resolution', Catena, vol. 178, pp. 172-188. https://doi.org/10.1016/j.catena.2019.03.017

Uncertainties of prediction accuracy in shallow landslide modeling : Sample size and raster resolution. / Shirzadi, Ataollah; Solaimani, Karim; Roshan, Mahmood Habibnejad; Kavian, Ataollah; Chapi, Kamran; Shahabi, Himan; Keesstra, Saskia; Ahmad, Baharin Bin; Bui, Dieu Tien.

In: Catena, Vol. 178, 01.07.2019, p. 172-188.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - Uncertainties of prediction accuracy in shallow landslide modeling

T2 - Sample size and raster resolution

AU - Shirzadi, Ataollah

AU - Solaimani, Karim

AU - Roshan, Mahmood Habibnejad

AU - Kavian, Ataollah

AU - Chapi, Kamran

AU - Shahabi, Himan

AU - Keesstra, Saskia

AU - Ahmad, Baharin Bin

AU - Bui, Dieu Tien

PY - 2019/7/1

Y1 - 2019/7/1

N2 - Understanding landslide characteristics such as their locations, dimensions, and spatial distribution is of highly importance in landslide modeling and prediction. The main objective of this study was to assess the effect of different sample sizes and raster resolutions in landslide susceptibility modeling and prediction accuracy of shallow landslides. In this regard, the Bijar region of the Kurdistan province (Iran) was selected as a case study. Accordingly, a total of 20 landslide conditioning factors were considered with six different raster resolutions (10 m, 15 m, 20 m, 30 m, 50 m, and 100 m) and four different sample sizes (60/40%, 70/30%, 80/20%, and 90/10%) were investigated. The merit of each conditioning factors was assessed using the Information Gain Ratio (IGR) technique, whereas Alternating decision tree (ADTree), which has been rarely explored for landslide modeling, was used for building models. Performance of the models was assessed using the area under the ROC curve (AUROC), sensitivity, specificity, accuracy, kappa and RMSE criteria. The results show that with increasing the number of training pixels in the modeling process, the accuracy is increased. Findings also indicate that for the sample sizes of 60/40% (AUROC = 0.800) and 70/30% (AUROC = 0.899), the highest prediction accuracy is derived with the raster resolution of 10 m. With the raster resolution of 20 m, the highest prediction accuracy for the sample size of 80/20% (AUROC = 0.871) and 90/10% (AUROC = 0.864). These outcomes provide a guideline for future research enabling researchers to select an optimal data resolution for landslide hazard modeling.

AB - Understanding landslide characteristics such as their locations, dimensions, and spatial distribution is of highly importance in landslide modeling and prediction. The main objective of this study was to assess the effect of different sample sizes and raster resolutions in landslide susceptibility modeling and prediction accuracy of shallow landslides. In this regard, the Bijar region of the Kurdistan province (Iran) was selected as a case study. Accordingly, a total of 20 landslide conditioning factors were considered with six different raster resolutions (10 m, 15 m, 20 m, 30 m, 50 m, and 100 m) and four different sample sizes (60/40%, 70/30%, 80/20%, and 90/10%) were investigated. The merit of each conditioning factors was assessed using the Information Gain Ratio (IGR) technique, whereas Alternating decision tree (ADTree), which has been rarely explored for landslide modeling, was used for building models. Performance of the models was assessed using the area under the ROC curve (AUROC), sensitivity, specificity, accuracy, kappa and RMSE criteria. The results show that with increasing the number of training pixels in the modeling process, the accuracy is increased. Findings also indicate that for the sample sizes of 60/40% (AUROC = 0.800) and 70/30% (AUROC = 0.899), the highest prediction accuracy is derived with the raster resolution of 10 m. With the raster resolution of 20 m, the highest prediction accuracy for the sample size of 80/20% (AUROC = 0.871) and 90/10% (AUROC = 0.864). These outcomes provide a guideline for future research enabling researchers to select an optimal data resolution for landslide hazard modeling.

KW - Alternating decision tree

KW - GIS

KW - Landslide susceptibility

KW - Pixel and sample size

KW - Uncertainty

U2 - 10.1016/j.catena.2019.03.017

DO - 10.1016/j.catena.2019.03.017

M3 - Article

VL - 178

SP - 172

EP - 188

JO - Catena

JF - Catena

SN - 0341-8162

ER -