The microclimate in post-wildfire areas in northern New Mexico: modeling plant stress

Abstract

Forest canopies regulate understory microclimate, and their loss to high-severity fire can create conditions that exceed the physiological tolerances of juvenile trees of the same species. In the southwestern United States, post-fire tree establishment is often limited, resulting in persistent forest cover loss. We characterized and modeled postfire microclimatic variability to understand how spatial patterns of stress constrain seedling survival. We deployed 125 custom microclimate data loggers at each of three high-severity burn sites in northern New Mexico, recording hourly air temperature, relative humidity, and soil moisture over two growing seasons (2023–2024). Vapor pressure deficit (VPD) was derived from temperature and humidity. Using physiological literature, we then defined moderate, severe, and extreme stress thresholds for each variable. We integrated in situ observations with UAS-derived solar radiation, LiDAR-based topographic indices, and ERA-5 climate data in XGBoost models to generate sub-meter resolution maps of stress metrics, including total stress hours, stress days, and maximum consecutive stress duration (R² = 0.87–0.94). Extreme conditions were common, with temperatures reaching 59.99°C and VPD 17.12 kPa. Moderate stress thresholds were exceeded across 91.5% of site areas, with severe (84.9%) and extreme (58.6%) thresholds also widespread. However, stress magnitude and duration were highly heterogeneous at sub-meter scales. Atmospheric stress was primarily driven by solar radiation, whereas soil moisture stress was weakly correlated with atmospheric variables (r = 0.12–0.15) and instead controlled by profile curvature, catchment area, and relative slope position (61.5% of model influence), all derivable from 1 m LiDAR DEMs. Two sites separated by 150 km exhibited nearly identical stress distributions at similar elevations, suggesting that post-fire microclimate can be predicted from topography. This framework provides a scalable approach to identify low-stress microsites that may support tree regeneration following high-severity fire.