TF24 hydraulics: Investigate new multilayer model from DAVE

[dfalster]

The DAVE model team (based on LPJ GUESS) also shied away from using Richards Equation due to its complexities. The model in LPJ GUESS is a multilayer model based on the following

• Gerten, D., Lucht, W., Schaphoff, S., Cramer, W., Hickler, T., & Wagner, W. (2005). Hydrologic resilience of the terrestrial biosphere. Geophysical Research Letters, 32(21), 243. https://doi.org/10.1029/2005gl024247
• Zhou H et al. (2024) A comprehensive evaluation of hydrological processes in a second-generation dynamic vegetation model. Hydrological Processes 38: e15152. DOI: 10.1002/hyp.15152

Original Two layer model

The original model described in Gergen et al 2005, has two layers with following dynamics.

Note that percolation from each layer is dependent on its water content, but not on the difference in water content between layers. This is presumably what makes it different to Richards Equation.

Updated 15-layer model

A recent paper Zhou H et al. (2024) extended the model from 2 to 15 layers. Unfortunately they don't fully describe the equations, but here is text from the paper:

• [update to have] greater vertical resolution of the soil scheme. The model now has 15 soil layers with each layer being 0.1 m thick rather than the previous version with two layers of thickness of 0.5 and 1.0 m (Gerten et al., 2004).
• The precipitation input is interpolated daily as rain or snow depending on the temperature,. When the rain encounters the plants, interception loss occurs (Appendix Figure B2) and the remaining rain falls through vegetation as input to the soil layers.
• The two soil layers nearest the surface representing the top 0.2 m of soil are distinguished to calculate soil evaporation. Actual ET in every patch is calculated by adding up interception loss, transpiration for all PFTs, and evaporation from bare soil.
• The water content of soil layers is updated daily, considering the processes of interception, throughfall, snowpack change, root water uptake, evaporation, percolation, and runoff.
• To confirm that water balance is maintained in the model, we calculated whether the incoming water volume is equal to the outgoing water volume in each gridcell that is, whether precipitation and irrigation is equal to the sum of runoff, ET, soil moisture change, and snow change. The water balance check results are shown in Appendices B3 and B4.

The schematic representation of the water balance computed for each grid cell in LPJ-GUESS is shown in Appendix Figure B1:

Clare Stephens commented that she thought when rainfall is received, it is distributed amount among the layers. Need to confirm this.

Regarding plant water use:

• Uptake by plants is partitioned according to the PFT-specific fraction of roots situated in each layer. Trees and shrubs are usually assumed to have a larger proportion of their roots in the lower soil layers, while herbaceous plants are shallow-rooting. The PFT-specific root fractions for each 10 cm soil layer, parameterized following Jackson et al. (1996) are shown in Appendix Figure A4. This feature affects the relative performance of woody and herbaceous PFTs in dry environments differing in the seasonality of rainfall.

Figure A4 PFT-specific root fractions for each 10 cm soil layer, parameterized following Jackson et al. (1996).