Browsing by Person "Poltoradnev, Maksim"

Now showing 1 - 1 of 1

Measuring and modelling of soil water dynamics in two German landscapes
(2018) Poltoradnev, Maksim; Streck, Thilo
The soil water regime is focus of various disciplines including agricultural sciences, hydrology, weather forecast and climate modelling. As an inherent part of land surface exchange processes, the dynamics of soil water content (SWC) is simulated in distributed hydrological models and land surface models (LSM). The accuracy of the simulated SWC directly influences the simulation outcome and its performance. Biases in modelled temporal SWC dynamics and its spatial distribution lead to errors in evapotranspiration, runoff, cloud and precipitation simulations. The main objective of my thesis was to study the factors that control the SWC dynamics and its spatial variability. Long-term measurements from the soil moisture networks Kraichgau (KR) and Swabian Alb (SA) provided the data basis of this study. SWC was sensed based on the Time Domain Transmission (TDT) technique. In each region, 21 measuring locations were distributed across three spatial domains: an inner domain 3 km × 3 km (5 stations), a middle 9 km × 9 km (8 stations), and an outer domain 27 km × 27 km (8 stations). The sizes of the three domains correspond with typical grid sizes of coupled atmosphere-LSM models. All stations were mounted on cropped agricultural sites. Each station was equipped with a TDT sensor, installed 15 cm deep into the soil, a rain gauge and a remote transfer unit. After adjusting the sensor networks, an in-situ field calibration was performed to derive pedotransfer and site-specific calibrations for TDT soil moisture sensors. The chemical and physical analysis of soil samples collected at each station revealed that soil bulk density influences in both regions the TDT readings. Moreover, the pedotransfer calibrations included electrical conductivity in KR and silt fraction and organic nitrogen content on SA. These variables are relatively easy to measure. Accordingly, the pedotransfer calibrations derived in this study are a quick possibility to calibrate TDT sensors in areas with similar soil properties as in KR and SA. Nevertheless, the site-specific calibrations performed the best and were therefore used for further data analysis. In the second study, a three-year record of SWC and rainfall was evaluated. The response of the regional mean (theta) of SWC to a rain event was influenced by the seasonal water balance (SWB). In KR, the relation was more pronounced for positive SWB and less for neutral and negative SWB. On SA, where SWB was highly positive in all three years, the response of theta to rainfall was always strong. At the seasonal scale, the relationship between the spatial standard deviation of SWC (sigma) and theta was investigated through sigma-theta phase-space diagrams. The results show that with decreasing SWC sigma-theta data pairs are approaching sigma at the permanent wilting point (sigma-thetawp). With increasing SWC, in contrast, sigma-theta data pairs are moving towards sigma at saturation (sigma-thetas). These two points were termed anchor points. The sigma-theta relationships formed combinations of concave and convex hyperbolas reflecting the variability of soil texture and depending on sigma in relation to the anchor points. At the event scale, hysteresis in the sigma-theta was observed. Most sigma-theta clockwise hysteresis cases occurred at an intermediate and intermediate/wet state of SWC. Among the factors that trigger the initiation of a sigma-theta hysteretic loop, the present study revealed the following: rainstorms with spatially highly variable intensities (threshold rainfall intensity of 1.1 ± 0.6 mm and 2.9 ± 2.8 mm for KR and SA, respectively), preferential flow and, possibly, hysteresis in soil water retention curves. Based on these results, the following hypothesis was formulated: sigma-theta phase space diagrams are useful to test whether hydrological models or land surface models (LSMs) capture the realistic range of spatial soil water variability. The concept was tested with the Noah-MP LSM. Observations obtained from KR and SA soil moisture networks over a three-year period from 2010 to 2012 were used to build up the sigma-theta phase-space. The study included two different setups used to compute the hydraulic conductivity and the diffusivity: 1) the default setting: the Clapp and Hornberger approach, and 2) the van Genuchten-Mualem functions. The default model parameterization was stepwise substituted with site-specific rainfall, soil texture, leaf area index (LAI) and green vegetation fraction (GVF) data. The atmospheric forcing was obtained from eddy covariance stations located in the regions. Although the model matched observed temporal theta dynamics fairly well for the loess soils of KR, it performed poorly in the case of the shallow, clayey and stony soils of SA. The best match was achieved with the van Genuchten-Mualem functions and site-specific rainfall, soil texture, GVF and LAI. Nevertheless, the Noah-MP LSM failed to represent the spatial variability of SWC. In most cases, the simulated sigma-theta data points were located below the bottom edge of the envelope, which indicates that the model smooths spatial variability of soil moisture. This smoothing can be mainly attributed to missing topography and terrain information, inadequate representation of the spatial variability of soil texture and hydraulic parameters, and the model assumption of a uniform root distribution.