As discussed previously, a natural cone of depression occurred in the same area during the fall 2020 season, with low groundwater elevations in the same area occurring in the spring 2021 season, as shown by real-world groundwater monitoring data from the DWR . This finding is further supported by the parent model outputs of groundwater flux seen in Figure 14. The general trend of groundwater flow in the parent model is away from the model boundaries and towards the center of the model, where the cone of depression occurs; water always flows from areas of high hydraulic head to low hydraulic head. The cone of depression in our model is more exaggerated than the natural cone of depression that occurs in the same area. Therefore, we have identified it as a systematic error in this model. We want to emphasize that this underestimation in heads can create uncertainties for the fluxes in the child model; for the current state of the project, we were more interested in running the child model scenarios than prioritizing the calibration of the parent model. Flow magnitudes in Layer 1 are very high in the southwest corner of the parent model domain, indicating high magnitudes of flux away from the foothills of the Coast Ranges . The steep gradient and high-magnitude fluxes in that area could be a result of the topography of the ground surface, or a “pulling” of water towards the cone of depression due to high amounts of groundwater pumping in the central area of the parent model domain,blueberry cultivation where a lot of pumping for agriculture occurs.
Both are plausible explanations, and both may contribute to the presence of the cone of depression. Flow in Layer 6 illustrates the influence of pumping wells that extract groundwater from the shallow to middle layers of the model, with the cluster of concentrated head gradients displaying the “pull” of water from groundwater pumping wells. All layers indicate flow away from the Sacramento River near the child model domain area where the ag-MAR field sites are located. This is significant because in the area of interest, the Sacramento River is a losing stream, meaning it does not take water from the surrounding groundwater as it flows downstream. This finding suggests that although the ag-MAR sites are located close to the Sacramento River, most of the recharged groundwater is not at a great risk of being diverted to the Sacramento River as surface water flow. We were able to construct the parent model and analyze the results to understand the regional groundwater flow in the parent model domain. These regional flow patterns and magnitudes were extracted to serve as the child model’s boundary conditions that were informed by the natural geography and hydrology of the region. Essentially, the parent model results were a credibility check for the child model. Since we successfully replicated the parent model’s flow patterns in the child model, we gained confidence in the results of our three recharge scenarios that were used to test different parameters using the child model. Despite the cone of depression observed in the parent model, the flow directions do make sense, but the magnitude of fluxes might be skewed because of how exaggerated the cone of depression is.Each of the three scenarios were run using the child model, and the flow in the child model domain was a direct result of the flow that occurred in the same area of the parent model domain. As such, the results and flow directions observed from the child model were influenced by the results of the parent model. It is important to note that each of the child model scenarios was affected in the same way by the cone of depression that occurs in the parent model, which was merely a systematic error in the parent model. Analyzing the flow direction and magnitudes of flux in the child model results from Scenario 1 make it clear that groundwater flows southwest, away from the Sacramento River and towards the cone of depression in the parent model.
Although the cone of depression occurs naturally, as seen in real-world data, we should reiterate that the cone of depression in our parent model is exaggerated, which may have influenced the magnitude of the groundwater flux in our child model scenarios. Scenario 1 was the baseline model run with unaltered deep percolation data. The one to-one plot of observed and simulated equivalent heads and residual map for the child model indicated that the model performed very well in the baseline scenario model run, since the simulated values are close to the observed head values. In Scenario 1, we noticed that proximity to the river affected the variation in simulated head values. For example, the hydrograph and recharge bar chart for Well_34 showed very little variation in groundwater elevation since that well is located so close to the Sacramento River. The hydrograph and recharge bar chart for Well_2b showed more of a response to recharge rates. In the child model run for Scenario 2, deep percolation rates were increased by one order of magnitude to assess whether a significant hydrologic response would be observed within the same time frame . From the combined hydrographs and recharge bar charts for Scenarios 1 and 2, we saw a significant difference in the water table elevation between the two scenarios. The simulated heads in Scenario 2 were consistently higher than the simulated heads in Scenario 1 . It is clear that increasing the magnitude of recharge, as we did in Scenario 2, shows a significant hydrologic response through a raised water table when compared to the simulated heads plotted from Scenario 1. In the case of Well_2b in Scenario 2, we even noticed that the simulated heads exceeded the observed values, unlike the simulated heads from Scenario 1.
We also want to clarify that for Well_2b, the spike in simulated head would realistically flood that monitoring well by approximately 2 m. However, since Scenario 2 is purely hypothetical, we want to emphasize that our goal was to test an extreme increase in recharge rates in order to see a significant response. The water budget component graphs of both Scenario 1 and Scenario 2 exhibited the same pattern, but as expected, the rates in Scenario 2 are extremely scaled up when compared to Scenario 1. In Scenario 1, we observed that as recharge rates increased, water levels rose in the model. When recharge rates declined,plant pots with drainage water began to drain out from the model cells and flow out to storage. In Scenario 2, we also noticed the same pattern of increased outflows from the model cells to storage after recharge rates started to decline. It was difficult to compare the results of Scenario 3 to the results of Scenarios 1 and 2, since the time discretization of Scenario 3 was ten times larger, and only simulated heads were analyzed since there were no projected groundwater elevation data that could be used as input observed heads. We initially hoped to see at least a slight and gradual increase in storage inflows and groundwater levels over the course of the ten years for Scenario 3, but the water budget graph indicated that just simply replicating the 2019 recharge program year did not provide enough variability to show any significant improvement in groundwater storage or water levels over time. Perhaps gradually increasing recharge rates with each consecutive year would have shown an overall positive hydrologic response by the end of the ten years. Although we did not observe the response we had hoped to see for Scenario 3, we learned what does and does not work for the context of designing that hypothetical scenario. What may work in a future scenario might be gradually increasing recharge rates with each consecutive year, which might show a significant hydrologic response by the end of the ten years. We also believe that the plots of the hydrographs and recharge bar charts for Scenario 3 did not show any changes in groundwater levels over time because the amount of recharge was so little. And unfortunately, the sites that have been flooded are located very close to the Sacramento River. This is an issue in the context of this model because the river package acts as a boundary condition, which does not allow for much movement of water near that boundary. Theoretically, Scenario 3 showed results that should have been expected, since it was essentially a ten-year replication of Scenario 1, which had such small deep percolation rates from flooding that we saw very little hydrologic effects. In a future scenario, replicating the design of Scenario 2 for ten years, instead of Scenario 1, might show a significant difference.After comparing the results of the three scenarios, it is evident that increasing deep percolation rates yields the most significant hydrologic response. The results of Scenario 2 yielded the most positive response, increasing the simulated equivalent heads at each of the four groundwater monitoring wells. Global climate change has increased extreme hydrological events, such as long-term drought, extreme precipitation, and frequent wet-dry cycles . This can lead to greater uncertainty in agricultural production globally . Improving soil water retention capacity can increase the resilience of agroecosystems and the soil microbial communities on which they depend . As a by-product of biomass pyrolysis under oxygen limited conditions, bio-char soil amendments provide a potential soil carbon sequestration technology to help mitigate global climate change .
Adding bio-char can provide other agricultural benefits, such as reducing nutrient leaching and increasing soil cation exchange capacity .Because bio-char physical characteristics vary depending on feed stock and pyrolysis conditions , its capacity to modify soil water retention depends on the combination of bio-char and soil properties. Comparing bio-chars prepared from straw and pine wood at different temperatures, Burrell et al. found no consistent impact on plant available water in three agricultural soils. However, Hansen et al. found that two gasification bio-chars improved plant available water in two coarse-textured soils. Bio-char soil amendment can influence soil water retention properties by decreasing soil bulk density , increasing total soil pore volume and altering the pore-size distribution , increasing soil surface area, especially in coarse-textured soil , and increasing soil aggregation . However, many of these proposed mechanisms have not been validated based on direct evidence. Observation of soil moisture distribution in bio-char-amended soil at a finer resolution can provide direct information about potential mechanisms. The investigation of water movement and distribution in porous media using traditional methods, such as a pressure plate , is challenging since pressure plate can only measure water retention capacity when the internal moisture distribution has reached equilibrium. Neutron imaging technology, a non-destructive method, provides the possibility to observe moisture distribution in undisturbed porous media. Neutron imaging can measure the spatial and temporal moisture distributions with a high resolution and is sensitive to minute changes in soil volumetric water content. This tool enabled us to investigate bio-char’s potential impact on soil water retention and water movement between bio-char and supporting material in a defined system, e.g., organic matter free silica sand. An area that has not received much research attention is how bio-char aging after application to soil may lead to changes in its properties over time. Most studies of bio-char and soil water retention measure impacts in freshly amended systems; however, bio-char soil amendment is considered to be a long-term practice . Interactions between bio-char particles and soil components will gradually alter the bio-char surface, especially under field conditions . For example, fresh bio-char has a relatively high surface area associated with its internal micro- and macropores , however over time particles of soil organic matter fill bio-char pore space and decrease its specific surface area . Ren et al. found that bio-char surface area increased after aging for 0.5 year in an agricultural soil and decreased during the following 1.5 years. A three-month lab incubation experiment also showed application of in-situ aged bio-char had a greater impact on soil water holding capacity than did fresh bio-char . The results are inconsistent in part because they are conducted under different, sometimes artificial, conditions and do not reflect the realistic aging processes that occur in agricultural fields subjected to physical disturbance, UV exposure and wet-dry cycles. Thus, long-term field studies are needed to better understand impact of bio-char on soil water retention capacity in agricultural systems with different management practices.