Surface temperature controls the pattern of post-earthquake landslide activity

Abstract

The patterns and controls of the transient enhanced landsliding that follows strong earthquakes remain elusive. Geostatistical models can provide clues on the underlying processes by identifying relationships with a number of physical variables. These models do not typically consider thermal information, even though temperature is known to affect the hydro-mechanical behavior of geomaterials, which, in turn, controls slope stability. Here, we develop a slope unit-based multitemporal susceptibility model for the epicentral region of the 2008 Wenchuan earthquake to explore how land surface temperature (LST) relates to landslide patterns over time. We find that LST can explain post-earthquake landsliding while it has no visible effect on the coseismic scene, which is dominated by the strong shaking. Specifically, as the landscape progressively recovers and landslide rates decay to pre-earthquake levels, a positive relationship between LST and landslide persistence emerges. This seems consistent with the action of healing processes, capable of restoring the thermal sensitivity of the slope material after the seismic disturbance. Although analyses in other contexts (not necessarily seismic) are warranted, we advocate for the inclusion of thermal information in geostatistical modeling as it can help form a more physically consistent picture of slope stability controls.

Figure 1

Active landslides in the WEER-MLI. (a) Coseismic scene and (b–d) three postseismic scenes showing the active landslides as polygons (data from Fan et al.). A slope unit subdivision is also displayed [software: ArcGIS 10.5, www.arcgis.com].

Figure 2

Model performance and susceptibility distribution. (a) ROC curves, (b) PR curves, and (c) probability density functions of landslide susceptibility for the coseismic (2008) and three postseismic (2011, 2013, 2015) scenes of the WEER-MLI [software: R 3.6.3, https://cloud.r-project.org/].

Figure 3

Variable effects of the model covariates over time. Effect of (a) mean peak ground acceleration, (b) mean slope, (c) maximum distance and (d) elongation of the SU, and (e) mean land surface temperature displayed as component smooth functions. The shaded areas are 95% confidence bands. The small vertical bars on the horizontal axes reflect the distribution of values across the SUs. The values in parentheses on the ordinates are the effective degrees of freedom, a proxy of the degree of non-linearity of the functions [software: R 3.6.3, https://cloud.r-project.org/].

Figure 4

Susceptibility maps across the years (a–d) and residual susceptibility maps (e–g). The box plot (h) tracks the change in susceptibility between scenes for each SU [software: ArcGIS 10.5, www.arcgis.com].

Figure 5

LST data for the WEER-MLI, averaged over the interval between successive image acquisitions. For the 2008 scene, one year of observations prior to the earthquake were considered; for the 2011 scene, data between the imaging date (April 2011) and the earthquake date in (12 May 2008) were taken, and so on. The figure also shows the average LST during eight years before the earthquake (2000–2008), for comparison [software: ArcGIS 10.5, www.arcgis.com].