Validation of landslide hazard models using a semantic engine on online news

The objective of this work is twofold: (i) automatically setting up a landslide inventory using a state-of-the art semantic engine based on data mining on online news and (ii) evaluating if the automatically generated inventory can be used to validate a regional scale landslide warning system based on rainfall-thresholds.The semantic engine scanned internet news in real time in a 50 months test period. At the end of the process, an inventory of approximately 900 landslides was automatically set up for the Tuscany region (23,000 km², Italy). Using a completely automated procedure, the inventory was compared with the outputs of the regional landslide early warning system and a good correspondence was found, e.g. 84% of the events reported in the news is correctly identified by the warning system.On the basis of the obtained results, we conclude that automatic validation of landslide models using geolocalized landslide events feedback is possible. The source of data for validation can be obtained directly from the Internet channel using an appropriate semantic engine dedicated to perform a monitoring of the Google News aggregator.Moreover, validation statistics can be used to evaluate the effectiveness of the predictive model and, if deemed necessary, an update of the rainfall thresholds could be performed to obtain an improvement of the forecasting effectiveness of the warning system.In the near future, the proposed procedure could operate in continuous time and could allow for a periodic update of landslide hazard models and landslide early warning systems with minimum or none human intervention.     

The Spatial Distribution and Attribute Parameter Statistics of Landslides Triggered by the May 12th, 2008, MW7.9 Wenchuan Earthquake

A complete landslide inventory and attribute database is the importantly fundamental for the study of the earthquake-induced landslide. Substantial landslides were triggered by the M_W7.9 Wenchuan earthquake on May 12^th, 2008. Google Earth images of pre- and post-earthquakes show that 52 194 co-seismic landslides were recognized and mapped, with a total landslides area of 1 021 km².Based on the statistics,we assigned all landslide parameters and established the co-seismic landslides database, which includes area, length, and width of landslides, elevation of the scarp top and foot edge, and the top and bottom elevations of each located slope. Finally, the spatial distribution and the above attribute parameters of landslides were analyzed. The results show that the spatial distribution of the co-seismic landslides is extremely uneven. The landslides that mainly occur in a rectangular area (a width of 30 km of the hanging wall of the Yingxiu-Beichuan fault and a length of 120 km between Yingxiu and Beichuan) are obviously controlled by surface rupture, terrain, and peak ground acceleration. Meanwhile, a large number of small landslides (individual landslide area less than 10 000 m²)contribute less to the total landslides area. The number of landslides larger than 10 000 m² accounts for 38.7% of the total number of co-seismic landslides, while the area of those landslides account for 88% of the total landslides area. The 52 194 co-seismic landslides are caused by bedrock collapse that usually consists of three parts:source area, transport area, and accumulation area. However, based on the area-volume power-law relationship, the resulting regional landslide volume may be much larger than the true landslide volume if the landslide volume is calculated using the influenced area from each landslide.     

Web data mining for automatic inventory of geohazards at national scale

In this study, we present a fully automated procedure to analyze online news using data mining techniques. It is then used to compile and continually update a geohazard database. The procedure is based on new technologies that publish news on the internet, i.e., the news is analyzed, georeferenced and attributed to a category of geohazards (the current categories are landslides, floods and earthquakes). A continuous flow of georeferenced events is established to populate and update the geodatabase automatically and in near-real time. We tested the procedure for 2 years at a national scale, creating a geodatabase containing more than 20,000 news items concerning geohazards that occurred in Italy. This procedure enables continuous feedback from events in the real world, such that information about geohazards can be fully exploited rapidly (compared to traditional techniques based on remote sensing, field surveys and historical inventories).