Figure 10. The difference of GST (ΔGST) for each scenario relative to L4 (left) and L6 (right). The first row for the year without pumping while the second for the first year of pumping.
However, when GW pumping occurs, great difference of GST among scenarios are observed not only in the first year of simulation (Figures 10b and 10d) but also through the entire simulation period (Figure 9). For example, in Figure 9, the maximum increase of GST is about 4 K for L4 while it is less than 1 K for L14 in the first year of pumping (blue lines). The difference can also be observed for the following years of pumping (Figure 9). This indicates GW pumping significantly shortens the time scale required to reveal the difference of GST caused by the coupling depth in the integrated modeling. Therefore, with the increasing intensity of human activities nowadays, the effect of coupling depth on the heat components in land surface processes cannot be neglected even in a short-term simulation. Such shortened time scale also means weakened buffer capacity of the subsurface due to pumping. As discussed in section 3.2, the thermal conductivity, the specific heat capacity, and the volume to store the heat (e.g., the coupling depth) are all important to the buffer capacity of the subsurface (Cuesta-Valero et al., 2016). When simulating scenarios with pumping, the decrease of thermal conductivity and volumetric heat capacity both have negative effect on the buffer capacity, and thus the positive effects of the increased coupling depth become more prominent than that under natural conditions. The above explanations are also consistent with results from the second group of simulations with tenfold pumping rate that more significant effect of the coupling depth were shown immediately after the pumping occurs, if comparing results presented in Figures 5a and 5c.
  1. The root zones
The coupling depth between ParFlow and CLM determines the depth where the root zone is truncated. This in turn provides the distribution of root fraction in each layer which is important in the land surface water and energy processes. Hence, the root zone for each scenario is analyzed to help account for a portion of the different simulation results among scenarios in section 3.3. The root fraction in each layer (ri ) for the four land cover types (Figure 2b) is calculated by Eq. (3) from the source code of ParFlow.CLM and is shown in Figure 11,
(3)
where zh,i is the depth from the soil surface to the interface between layers i and i +1 (zh, 0 = 0, the soil surface),Nlevsoi is the number of soil layers for coupling, and ra and rbare plant-dependent root distribution parameters. Each land cover type in this study corresponds to one plant functional type withra and rb based on IGBP classification. In Figure 11, the root distribution for all plant types can be fully described with a coupling-layer number larger than 10 since the root fraction has exponentially decreased to zero. For L4, root extension in the vertical direction for all plant types are truncated. For L6, L7, and L8, slight truncation may be needed. For L11 and L14, the root zone should be fully described. Therefore, obvious difference for water and energy components is expected between L4 and other scenarios (Figure 10). However, it should be noted that deeper truncation does not mean better parameterization of the root-fraction distribution.