Simulated or measured soil moisture: Which one is adding more value to regional landslide early warning?

Abstract:

Rainfall-triggered landslides are a serious risk to people and infrastructure in mountainous regions. Landslide early warning systems (LEWS) have demonstrated to be a valuable tool to inform decision makers about the imminent landslide danger and to move people or goods at risk to safety. Recent studies have shown an improvement of existing rainfall-based LEWS by including in-situ soil moisture measurements.

However, the use of sensor networks bears specific limitations (e.g. sensitivity to local conditions, data quality issues, costly installation and maintenance) that could be overcome by the application of numerical models. On the other hand, soil moisture models are restricted by assumptions and simplifications related to the soil hydraulic properties and the water transfer in the soil profile. Ultimately, the question arises how reliable and representative simulation-based early warnings are compared to using measurements.

To answer this, a one-dimensional heat and mass transfer model (CoupModel; Jansson, 2012) was applied at 35 sites in Switzerland to simulate soil moisture during the period of 2008 to 2019. A statistical framework (Wicki et al., 2020) was used to analyse and quantify the temporal soil moisture variation and to assess the forecast goodness for rainfall-triggered landslides by the comparison with a national landslide database. Finally, the forecast goodness was compared with applying the same statistical framework to measured soil moisture at the same sites and time period. Further, the use of a common parameter set allowed for the extension to an additional 120 sites where meteorological measurements are available to assess the effect of increasing the information density on the overall forecast skill.

First results indicate a similar forecast goodness if modelled soil moisture is used compared to measurement-based early warnings. For short distances between site and landslide location, the forecast goodness could be increased if more sites are included. Generally, this supports the use of models in LEWS to complement a monitoring network.

REFERENCES

Jansson, P-E (2012). CoupModel: Model Use, Calibration, and Validation. Transactions of the ASABE.

Wicki, A et al. (2020). Assessing the potential of soil moisture measurements for regional landslide early warning. Landslides.