Shallow landslides assessment in the wildland-urban interface: The Palomares Basin case study

Abstract

Shallow landslides are a recurrent and destructive hazard in the Metropolitan Area of Concepción, Chile, having caused at least 17 fatalities since 1980 along with substantial economic losses. The Palomares Basin, located on the urban fringe of Concepción, is characterized by a wildland-urban interface where urban development is surrounded by hills covered with plantations. This area is particularly vulnerable to both wildfire and landslide hazards, which are exacerbated by vegetation loss and increased soil erosion. This scenario highlights a pressing territorial issue: the impact of deforestation and the forestry activities on landslide occurrence within the wildland-urban interface.

The Palomares Basin exhibits regional conditions that are typical yet often overlooked: (1) deep, well-developed soils derived from highly weathered granitic regoliths, (2) steep slopes, (3) high-intensity rainfall, and (4) socio-economic factors such as: (a) the occupation of marginal hilltop areas due to socio-economic segregation, (b) inadequate land-use planning, and (c) human-induced factors that increase landslide susceptibility, particularly deforestation.

To assess the impact of deforestation on soil mechanics and root cohesion, and the subsequent increase in shallow landslide probability, we simulated pre- and post-deforestation scenarios using the STEP-TRAMM software. The input data included rainfall events from 2019; soil depth; elevation (using 0.2 m² LiDAR pixels for individual slopes and 5 m² pixels for the entire basin); soil texture; cohesion; friction angle; landslide inventory (derived from fieldwork and remote sensing); initial soil saturation; root cohesion; and vegetation cover status (pre- and post-deforestation, as detected through multispectral and high-resolution imagery analysis).

The software quantified the spatio-temporal probability and magnitude of landslides by accounting for changes in soil cohesion resulting from the loss of tree root structures. Simulations were validated against the landslide inventory, with a focus on comparing the effects of deforestation. Results indicated a significant increase in landslide probability under deforested conditions, with the post-deforestation scenario showing up to three times more landslides and five times more displaced material. At both the single-slope and basin scales, up to 70 % of simulated landslides coincided with the landslide inventory. At the single-slope scale, the best fit for spatio-temporal and volumetric landslide generation occurred with initial soil saturation levels between 0.5 and 0.8 and a soil cohesion of 1 kPa, with minimal temporal discrepancies between simulated and actual scenarios. At the basin scale, initial soil saturation levels of 0.5–0.7 and a soil cohesion of 3 kPa were most appropriate. Under similar precipitation events, deforestation was found to substantially increase landslide risk, exacerbating the vulnerability of populations due to a combination of inadequate urban planning, unsustainable forestry practices, and the inherent instability of developed slopes.