Fire-induced changes in soil hydromechanical properties and implications for shallow landslide triggering in southeastern Brazil

Fire plays a significant role at the vegetation-soil-water interface, and its effects on soil properties and hydro-erosion dynamics have been widely investigated. However, fire effects on soil hydromechanical behavior and their relationship with increased susceptibility to shallow landslides remain poorly understood, particularly in tropical environments. In the mountainous region of Rio de Janeiro, Brazil, in the Nova Friburgo municipality, where an average of 316 fires occur per year, paleoenvironmental evidence indicates that shallow landslides preceded by wildfires occurred during the Holocene. To investigate fire-induced changes in soil hydromechanical properties, a field experiment applying controlled fire was conducted in 100 m² plots at the edge of a secondary Atlantic rainforest (RF) (October 2024) and in a homogenous grassland (GL) under antecedent grazing (September 2025). Flame temperature and soil temperatures were recorded at the surface, 5, and 10 cm depths. Adjacent unburned control plots were also established. Burned and unburned plots were outfitted with soil moisture sensors and soil suction sensors at different depths from surface to 150 cm. At burned RF, a tensiometer and additional suction sensors were also installed to improve measurement accuracy. In situ measurements of saturated hydraulic conductivity (Ksat) and soil water repellency were conducted at different instrumented depths. Laboratory analyses of texture, bulk density, Ksat, and soil water retention curves (SWRC) were conducted. To study the temporal changes of hydromechanical soil properties post-fire, measurements were investigated before burning and at successive post-fire intervals of 1 week, for selected parameters, 1 month, 6 months, and 1 year. Results from the flame temperature measurements during the experiment showed a non-uniform spatial distribution over time at both plots. While at RF burning experiment lasted for 2 hours with active flames, at GL the flames propagation lasted less than 10 minutes. Maximum soil temperatures at RF reached 361.5°C at the surface and 334.7°C at 5 cm depth, while at GL, surface temperature increased by approximately 50°C. At RF, in situ Ksat at 20 cm depth increased from 6.44 × 10^-5 m s^-1 before the fire to 1.51 × 10^-4 m s^-1 one week after burning, followed by a slow decrease over six months and one year, while no significant changes at greater depths (up to 150 cm) over the investigated time interval, or at any depth at GL were observed. Severe soil water repellency was detected at RF before burning, up to 10 cm depth, and progressively declined after fire, disappearing after six months, whereas no repellency was observed at GL. SWRC from both in situ and laboratory measurements indicated a reduction in volumetric water content at the saturation stage in the upper 20 cm at RF, with no changes at greater depths up to 150 cm and at all depths at GL. These findings will be used to improve the modelling of the post-fire hydromechanical soil behavior by integrating the in situ monitoring data and laboratory measurements, thereby enhancing the calibration of physically based models for shallow landslide susceptibility assessment.