Application of Hydrological Model Simulations in Landslide Predictions

Binru Zhao, Qiang Dai, Dawei Han, Jun Zhang, Lu Zhuo, Matteo Berti

Research output: Contribution to journalArticle (Academic Journal)peer-review

11 Downloads (Pure)

Abstract

The importance of soil moisture conditions in the initiation of landslides has been widely recognized. This study takes advantage of the distributed hydrological model to derive the soil wetness index. The derived soil wetness index is then used to determine soil wetness thresholds for landslide occurrences. In order to predict landslides based on alert zones, a zone threshold is introduced together with the soil wetness threshold to constitute the integrated threshold. We evaluate the prediction performance of the integrated thresholds with the use of skill scores and the receiver operating characteristic (ROC) curves. This study is carried out in a sub-region of the Emilia-Romagna region, Northern Italy. Results show that the derived soil wetness index could account for the hydrological process that is controlled by meteorological conditions and topographic properties. The proposed integrated threshold shows a better predictive capability than the rainfall threshold, demonstrating the effectiveness of applying the soil wetness index in landslide predictions. The optimal threshold is also determined by compromising the correct predictions and incorrect predictions, it is found that the optimal integrated threshold is more advantageous in reducing false alarms compared with the optimal rainfall threshold. This study highlights the potential of applying hydrological simulations in landslide prediction studies and provides a new way to make use of high-resolution data in zone-based landslide predictions.
Original languageEnglish
JournalLandslides
Early online date13 Dec 2019
DOIs
Publication statusE-pub ahead of print - 13 Dec 2019

Keywords

  • hydrological simulation
  • soil wetness index
  • landslide threshold

Fingerprint Dive into the research topics of 'Application of Hydrological Model Simulations in Landslide Predictions'. Together they form a unique fingerprint.

Cite this