Development of a New Landslide Diagnosis and Prediction System in Korea - A Coupled WRF-Landslide Model
Abstract
The number of landslides in Korea has rapidly increased since 2000 due to regional warming, which has completely changed the summer rainfall pattern in Korea. This change is characterized by an increase in rainfall intensity that possesses a larger spatial variability, and therefore, makes mountain hazards, such as landslides and debris flows, more unpredictable. (1) A combined machine learning algorithm and land surface model-based flux estimation and (2) a coupled This study focuses on the latter modeling component, while the former flux estimation will be detailed in a separate presentation. The NCAM-WRF model uses the Korea Meteorological Administration global Unified Model as the initial and boundary conditions. Medium-range rainfall and soil moisture predictions are made at 810-m resolution, statistically corrected using a support vector machine (SVM), and then fed into the local landslide risk model as input data. The NCAM-WRF model prediction errors are obtained by comparing the daily average of the previous precipitation prediction with the observation. The SVM method is tested for one year, from August 2017 to July 2018, to calibrate the predicted time series data for each run. The results show that the SVM-based bias correction improves the NCAM-WRF rainfall prediction. The uncorrected rainfall root-mean-square error (RMSE) is 3.6 (bias = 0.3) for the site-averaged quantities, while the SVM-based bias correction reduces the RMSE to 2.1 (bias = 0.1). Here soil water content (SWC) changes from the unsaturated to saturated state are simulated in the coupled NCAM-WRF-landslide model as rainfall duration changes. A new landslide risk index (LRI) prediction is created based on the soil layer depth and SWC change, which can be spatially mapped across a study region and ranges from zero to one, with an increasing LRI value indicating a higher landslide risk. This presented system can complement existing landslide warning information in Korea to reduce extensive loss of life and property damage in landslide-prone areas, particularly in mountainous urban residential centers.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2018
- Bibcode:
- 2018AGUFMNH13C0712P
- Keywords:
-
- 4301 Atmospheric;
- NATURAL HAZARDSDE: 4302 Geological;
- NATURAL HAZARDSDE: 4313 Extreme events;
- NATURAL HAZARDSDE: 4333 Disaster risk analysis and assessment;
- NATURAL HAZARDS