A method to derive spatially variable root cohesion maps from underground biomass maps for regional landslide susceptibility models

In physics-based models for regional landslide susceptibility, the slope stability is most sensitive to the changes of both soil and root cohesion, thus a stochastic approach for these variables is preferred. Root cohesion (RC) is frequently oversimplified and assumed constant and uniform across the landscape. However, the assumption of a constant RC may be inappropriate because root distribution varies spatially with forest characteristics and temporally with tree growth.

Root biomass (RB) is a parameter used to partly quantify the amount of carbon sequestered in forest soils and therefore some empirically based functions (Marklund, 1988) exist to estimate RB from forest characteristics (size and species) in Norway. In turn, RB can be correlated to the reinforcement provided by roots, and can be included in slope stability models as an additional RC term.

This study presents an expeditious estimation of spatially variable RC from RB maps. The resulting maps can be used as input to shallow landslide susceptibility models at regional scale.

The method uses the formula proposed by Hwang et al. (2015), which derives the basal root cohesion from the original W&W model as proportional to the root biomass per ground area, and the density of the roots. The assumptions of the method are the following:

• Basal RC is used to account for the role of forests on regional scale landslide susceptibility and hazard analysis.
• RC increase is directly proportional to RB.
• Only the roots with a diameter below 5 mm are concurring to the additional RC, thus only a percentage of the total root underground biomass as calculated for purposes of carbon sequestration is considered.

The method was applied for Norwegian forests consisting primarily of Pine (Pinus sylvestris), Spruce (Picea abies), and Birch (Betula pendula and pubescens) trees.

The RB maps were obtained by models produced by the Norwegian Institute of Bioeconomy Research (NIBIO), which use a mixture of airborne LIDAR and satellite imagery to characterise the above ground forest characteristics nationally (Astrup et al. 2019), then use the empirical functions to estimate RB. The surface resolution is 16 m².

The study area is located in Sunnfjørd, precisely in Jølster, where several rainfall-induced landslides occurred in July 2019, causing damages to local infrastructure.

Estimates of RC from RB are compared with empirical data from the literature for similar species in forests outside Norway, showing reasonable consistency in the values ranges obtained. Further validation is needed with empirical data from Norwegian forests.

This study provides a simple yet computationally efficient estimation of root cohesion from RB maps, which can be used to supply parameters for models accounting for the effect of vegetation on landslide susceptibility at regional scale. In the future, it will be necessary to develop more precise relationships of fine root biomass to above ground forest characteristics with respect to changing soil properties.

References:

Astrup et al. (2019). https://doi.org/10.1080/02827581.2019.1588989

Hwang et al. (2015). https://doi.org/10.1002/2014JG002824

Marklund, L.G. (1988) Biomassafunktioner för tall, gran och björk i Sverige = Biomass functions for pine, spruce and birch in Sweden. Umeå: Sveriges lantbruksuniversitet, Institutionen för skogstaxering.