No abstract

Abstract. Large weighing lysimeters are currently the most precise method to directly measure all components of the terrestrial water balance in parallel via the built-in weighing system. As lysimeters are exposed to several external forces such as management practices or wind influencing the weighing data, the calculated fluxes of precipitation and evapotranspiration can be altered considerably without having applied appropriate corrections to the raw data. Therefore, adequate filtering schemes for obtaining most accurate estimates of the water balance components are required. In this study, we use data from the TERENO (TERrestrial ENvironmental Observatories) SoilCan research site in Bad Lauchstädt to develop a comprehensive filtering procedure for high-precision lysimeter data, which is designed to deal with various kinds of possible errors starting from the elimination of large disturbances in the raw data resulting e.g., from management practices all the way to the reduction of noise caused e.g., by moderate wind. Furthermore, we analyze the influence of averaging times and thresholds required by some of the filtering steps on the calculated water balance and investigate the ability of two adaptive filtering methods (the adaptive window and adaptive threshold filter (AWAT filter; Peters et al., 2014), and a new synchro filter applicable to the data from a set of several lysimeters) to further reduce the filtering error. Finally, we take advantage of the data sets of all 18 lysimeters running in parallel at the Bad Lauchstädt site to evaluate the performance and accuracy of the proposed filtering scheme. For the tested time interval of 2 months, we show that the estimation of the water balance with high temporal resolution and good accuracy is possible. The filtering code can be downloaded from the journal website as Supplement to this publication.

Hysteretic processes have been recognized for decades as an important characteristic of soil hydraulic behavior. Several studies confirmed that wetting and drying periods cannot be described by a simple functional relationship, and that some nonequilibrium of the water retention characteristics has to be taken into account. A large number of models describing the hysteresis of the soil water retention characteristic were successfully tested on soil cores under controlled laboratory conditions. However, its relevance under field conditions under natural forcings has rarely been investigated. In practice, the modeling of field soils usually neglects the hysteretic nature of soil hydraulic properties. In this study, long-term observations of water content and matric potential in lysimeters of the lysimeter network TERENO-SoilCan are presented, clearly demonstrating the hysteretic behavior of field soils. We propose a classification into three categories related to different time scales. Based on synthetic and long-term monitoring data, three different models of hysteresis were applied to data sets showing different degrees of hysteresis. We found no single model to be superior to the others. The model ranking depended on the degree of hysteresis. All models were able to reflect the general structure of hysteresis in most cases but failed to reproduce the detailed trajectories of state variables especially under highly transient conditions. As an important result we found that the temporal dynamics of wetting and drying significantly affects these trajectories which should be accounted for in future model concepts.