Significance of soil moisture variability in forest on Soil-Vegetation-Atmosphere Transfer model results

Understanding forest water cycles and the processes influencing them is critical for predicting how environmental changes may impact forest hydrology. Soil-Vegetation-Atmosphere Transfer (SVAT) models are essential tools for simulating and understanding water fluxes within forest ecosystems. However, the accuracy and reliability of these models are often limited by the quality and availability of input data, particularly soil hydraulic parameters. Consequently, assumptions made in the modeling process can lead to underestimations or overestimations of key water balance components, such as groundwater recharge. To improve the predictive accuracy of SVAT models, observed soil moisture data are commonly used to validate model parameterization, ensuring that simulated soil moisture levels match the observations. Nonetheless, determining which measured values or how many observations are adequate for model calibration poses a challenge due to the high spatial and vertical variability of soil moisture. This variability is driven by heterogeneity in soil properties and forest structure across the studied area.

For this reason, the study presented investigates the variability of soil moisture observations across two diverse forest environments differing primarily in soil matrix homogeneity and tree species composition. The influence of soil moisture variability on input parameter set adjustments required for effective SVAT model calibration was analyzed based on recorded soil moisture by a sensor network installed at both sites. Specifically, the study examined whether significant modifications to the parameter set are necessary to match simulated and observed soil moisture from identified soil moisture clusters at the sites.

The results confirmed that soil moisture variability was greater at the site with a more heterogeneous soil matrix, both spatially and with depth. At such locations, significant adjustments to input parameters are needed to match simulated and observed soil moisture, substantially affecting the simulation of individual water balance components. Using a mean soil moisture value for model validation proved inadequate for capturing the full range of variability at such locations. Conversely, at the site with a more homogenous soil matrix, soil moisture variability was low and adjustments to input parameters were negligible. Here, using the mean soil moisture value for model calibration is sufficient to represent the water balance accurately. The results highlight the importance of considering soil moisture variability when calibrating SVAT models, particularly in heterogeneous environments.