Soil water content measurements deliver reliable estimates of water fluxes: A comparative study in a beech and a spruce stand in the Tharandt forest (Saxony, Germany)

Previous studies have shown that soil water content can vary considerably within homogeneous sites. This small-scale variability of soil water is often neglected when studying water and carbon fluxes in forest ecosystems. In this paper, the small-scale variability of soil water was analyzed at two contrasting eddy-flux sites, a Norway spruce forest and a European beech forest. Simultaneous measurements of precipitation, eddy covariance, and sap flow, from soil water content readings, were used to answer the question of how representative soil water gain is during rainfall and evapotranspiration is during dry periods.Our study demonstrates that the spatial and temporal variability in soil water under spruce and beech was mainly due to the differences in soil properties and root intensity. This can be concluded from the fact that the pattern of soil moisture distribution and flow paths under the trees were generally stable throughout the season. As a tendency, areas with preferred accumulation of rainwater were mainly characterized by maximum soil water depletion. Therefore, the density of the installed water content sensors should correspond to the variability of soil properties as well as rooting intensity. Based on previous studies and our own results, it can be concluded that a horizontal and vertical distance between 10 and 30 cm is best suited for water content sensors to detect preferential flow paths and deliver reliable estimates of soil water balance.Despite the occurrence of preferential flow, we found that the soil water increase during rainstorm periods and the soil water depletion during dry periods can be estimated relatively well when the small-scale variability of soil properties is considered in the experimental setup. In general, the evaporation estimates based on eddy covariance, sap flow, and soil water balance were consistent. However, compared to the spruce site, at the beech site the gap between the evapotranspiration estimates based on eddy covariance and soil water balance were often relatively large. Differences in the spatial extent of these methods can only explain these discrepancies to a certain extent. We suggest that this might be mainly due to the lack of water content sensors in the immediate vicinity of the beech tree trunk. Thus, stemflow-induced wetting and subsequent drying around the trunk could not be monitored in our study. This may result in an underestimation of evaporation from the soil under beech using the soil water balance method compared to the eddy covariance method. Finally, soil water depletion under spruce led to a significant reduction of transpiration when the actual available plant soil capacity (AWC) was <40% of the potential AWC. In contrast to the spruce stand, a reduction of transpiration of beech due to water shortage was not observed.