Search

Functional structural root architecture models have evolved as tools for the design of improved agricultural management practices and for the selection of optimal root traits. In order to test their accuracy and reliability, we present the first benchmarking of root water uptake from soil using five well established functional structural root architecture models: DuMux, CPlantBox, RSWMS, OpenSimRoot and SRI. The benchmark scenarios include basic tests for water flow in soil and roots as well as advanced tests for the coupled soil root system. The reference solutions and the solutions of the different simulators are available through Jupyter Notebooks on a GitHub repository. All of the simulators were able to pass the basic tests and continued to perform well in the benchmarks for the coupled soil-plant system. For the advanced tests, we created an overview of the different ways of coupling the soil and the root domains as well as the different methods used to account for rhizosphere resistance to water flow. Although the methods used for coupling and modelling rhizosphere resistance were quite different, all simulators were in reasonably good agreement with the reference solution. During this benchmarking effort, individual simulators were able to learn about their strengths and challenges, while some were even able to improve their code. Some now include the benchmarks as standard tests within their codes. Additional model results may be added to the GitHub repository at any point in the future and will be automatically included in the comparison.

The link between soil water content (SWC) and evapotranspiration (ET) is of great importance in ecohydrology, agroecosystem management, and land-atmosphere interaction of earth-system analysis. There is still a need to understand the key processes linking SWC and ET in different vegetated ecosystems, especially at larger scales. Therefore, in this work, we used wavelet coherence analysis to explore the long-term relationship between SWC and ET among the predominant land use types (cropland, evergreen needleleaf forest, mixed forest, open shrubland, wooden tundra, grassland, and mixed tundra) in Europe during the last four decades (1980-2020). To this end, first a principal component analysis was performed among the SWC and ET data from GLDAS, GLEAM, and ERA5-land, and the first component was then used for further analyses when the target variable was in demand. Using the first component, then, we averaged SWC and ET data over the pixels covered by each land use type. Then, for each land-use types, we averaged the data daily for each decade to account for a representative decadal year of daily data. Then, wavelet coherence analysis was conducted between those averaged data of SWC and ET for each land-use type. The results showed a negative correlation between ET and SWC for all land use types, with ET lagging behind SWC, with an average phase shift value of 134 days for grassland (the minimum) and 168 days for mixed tundra (the maximum). Converting the phase shift values to a time lag [lag = n/2 - phase shift for the case phase shift < -n/4 where n represents the period (1/frequency) of the signals] shows (Fig. 1) that ET controls SWC with a lag of 15 days in mixed tundra (the minimum) and 48 days in grassland (the maximum). Moreover, we applied Mann-Kendall trend analysis test and found that the lag between SWC and ET decreases in mixed and wooden tundras with a slope value of -2.4 days/year, while it increases in cropland and grassland with a slope value of 1.6 days/year. Although there is a significant downward trend in evergreen needleleaf forests (with a slope value of -0.3 days/year) and an increasing trend in mixed forests and open shrublands (with a slope value of 0.4 days/year), the lower slope values in these land use types indicate that the change is slower compared to grasslands, croplands, and tundras. Figure 1- Temporal evolution of the lag between soil water content and evapotranspiration in the annual cycle in different land use types in Europe