Frontiers in Water (Nov 2020)
Comparison of Soil Water Estimates From Cosmic-Ray Neutron and Capacity Sensors in a Semi-arid Pine Forest: Which Is Able to Better Assess the Role of Environmental Conditions and Thinning?
Abstract
Water scarcity in semi-arid regions is expected to increase under climate change, which will significantly affect forest ecosystems by increasing fire risk, diminishing productivity and water provisioning. Eco-hydrological forest management is conceived here as an adequate strategy to buffer climate change effects and increase forest resilience. Under this context, soil moisture is a key variable to quantify the impacts of eco-hydrological forest management on forest-water relations. Cosmic-ray neutron and capacitance probes are two different techniques for measuring soil moisture, which differ greatly in the spatial scale of the measurement support (i.e., few centimeters vs. several hectares). This study compares the capability of both methodologies in assessing soil water dynamics as a key variable that reflects the effects of forest management in a semi-arid environment. To this end, two experimental plots were established in Sierra Calderona in the province of Valencia in Spain in a post-fire regeneration Aleppo pine forest with high tree density. One plot was thinned (T) and the other remained as control (C). Nine capacitance probes and one Cosmic Ray Neutron Probe (CRNP) were installed in each plot. First, the CRNP was calibrated and validated, and subsequently, the performance of both techniques was analyzed by comparing soil moisture and its relationship with environmental variables and stand transpiration. The validation results confirmed the general reliability of CRNP to obtain soil moisture under semi-arid conditions, with a Kling-Gupta efficiency coefficient (KGE) between 0.75 and 0.84, although this performance decreased significantly when dealing with extreme soil moisture (KGE: −0.06–0.02). A significant effect of forest biomass and litter layer was also observed on CRNP-derived soil moisture, which produced an overestimation of soil moisture. The performance of both methodologies was analyzed by partial correlations between soil moisture and environmental variables and transpiration, as well as by applying Boosted Regression Trees to reproduce tree transpiration with each soil moisture measurement technique together with the environmental variables. Both methodologies were capable to reproduce tree transpiration affected by soil moisture, environmental variables and thinning, although CRNP always appeared as the most affected by atmospheric driving forces.
Keywords
