Big slopes, little data: data-driven nowcasting of deep-seated landslide deformation

Landslides are a major geohazard in hilly and mountainous environments. We focus on slow-moving, deep-seated landslides that are characterized by gradual, non-catastrophic deformations of millimeters to decimeters per year and cause extensive economic damage. To assess their potential impact and for the design of mitigation solutions, a detailed understanding of the slope processes is desired. Moreover, where landslide hazard mitigation is impossible, early warning systems are a valuable alternative to reduce landslide risk.
Recent studies have demonstrated the effective application of machine learning for deformation forecasting to specific cases of slow-moving, non-catastrophic, deep-seated landslides. Machine learning, combined with satellite remote sensing products offers new opportunities for both local and regional monitoring of areas with unstable slopes and associated processes without costly and logistically challenging inspection of the landslide. To test to what extent data-driven machine learning techniques and remote sensing observations can be used for landslide deformation forecasting, we developed a machine learning based nowcasting model on the multi-sensor monitored, deep-seated Vögelsberg landslide, near Innsbruck, Tyrol, Austria. Our goal was to link the landslide deformation pattern to the conditions on the slope, and to produce a four-day, short-term forecast, a nowcast, of deformation accelerations.
Changes in hillslope hydrology shift the balance between the shear strength of the soil and the shear (sliding) force applied by the gravitational forces acting on the landmass. Therefore, precipitation, snowmelt, soil moisture, evaporation, and air temperature were identified as hydro-meteorological variables with high potential for forecasting deformation dynamics. Time series of those variables were obtained from remote sensing sources where possible, and otherwise from reanalysis sources as surrogate for data that is likely to be available in the near future. Deformation, the result of slope instability, was monitored daily by a local, automated total station.
Interferometric Synthetic Aperture Radar (InSAR) has shown to be a valuable resource of deformation information from space. However, due to the complex interaction with topography in mountainous environments, its potential is often questioned. We showed that 91% of the world’s slopes are observable by InSAR, given the presence of a coherent scatterer, i.e. a natural or man-made object that exhibits consistent radar reflection over time. A global map is provided to indicate the sensitivity of InSAR to assess downslope deformation on any particular slope. To quickly assess the presence of coherent scatterers, before further investigation, we developed an application in Google Earth Engine to estimate the presence and location of coherent scatterers on a slope. However, the current accuracy and temporal resolution of Sentinel-1 SAR acquisitions proved insufficient to identify the acceleration phases at Vögelsberg.
The five years of daily deformation and hydro-meteorological observations at the Vögelsberg landslide is quite limited for a machine learning model. Therefore, a nowcasting model of low complexity was required. To limit the number of parameters to be optimized, the model was designed to mimic a bucket model, a simple hydrological model. A shallow neural network based on long short-term memory, was implemented in TensorFlow, as custom sequence of existing building blocks. Furthermore, a traditional neural network and recurrent neural network were tested for comparison. Thanks to the limited complexity of the model, the major contributors could be determined by trial-and-error of nearly 150 000 model variations.
Models including soil moisture information are more likely to generate high quality nowcasts, followed by models based solely on precipitation or snowmelt. Although none of the shallow neural network configurations produced a convincing nowcast deformation, they provide important context for future attempts. The machine learning model was poorly constrained as only five years of observations were available in combination with the four acceleration events that occurred in these five years. Furthermore, standard error metrics, like mean squared error, are unsuitable for model optimization for landslide nowcasting.
We showed that landslide deformation nowcasting is not a straightforward application of machine learning. The complexity of the machine learning model formulation at the Vögelsberg illustrates the necessity of expert judgement in the design and evaluation of a data-driven nowcast of slowly deforming slopes. Furthermore, to prepare for unexpected modelling developments, a high level of project level data organisation is recommended. There is a long road ahead for the large scale implementation of machine learning in landslide nowcasting and Early Warning Systems. However, a future, successful nowcasting system will require a simple, robust model and frequent, high quality and event-rich data to train upon.