The majority of urban hydrological models focus primarily on differences in the rainfall-runoff ratio between pervious and impervious area. Impervious area (as I have mentioned in a previous post), is traditionally assumed to be the major causal source of changes in the hydrological cycle associated with urbanization. Infrastructure-centric urban hydrologic models assume that water that is infiltrated or evapotranspirated before reaching drains and pipes has “exited” the system. As more and more infiltration-based stormwater control measures are implemented however, this assumption needs to be more closely examined. Are urban soils and the urban subsurface truly effectively inexhaustible in capacity? This question is particularly important when we think about multiday rain events and changes in intensity of rainfall associated with climate change.
This is the reason why in my research I do not ignore subsurface dynamics in my hydrological modeling. By utilizing a coupled, fully three-dimensional surface-subsurface, land-atmosphere model, I can investigate the extent to which subsurface interactions may impact how we think about extensive infiltration practices, and how evapotranspiration (land-atmosphere flux) and infiltration (surface-subsurface flux) may be influenced by various controls in landscape, such as spatial configurations of tree cover, imperviousness, topography, wetness indices, and hydraulic conductivity and soil porosity.
I am finally finished running the simulations for all nine of the scenarios I am testing. For those unfamiliar with the ParFlow.CLM modeling system I am using, the outputs of this model are fully distributed in three-dimensional space, and the user can output 3-D pressure fields, and saturation, as well as specific components of the water balance, such as evapotranspiration and overland flow as part of the simulation.
A combination of post processing using ParFlow tools (deployed through commands using Tcl scripts), FORTRAN scripts for calculations on the entire domain over time, and visualization in VisIt allow me to really dig into sources of causality for responses that reflect the integration of heterogeneous conditions over the entire domain. For example, I noticed a large difference in peak overland flow between two of my scenarios, and could “search” my domain for major anomalies in response between the two:
Using “index select” in Visit, I could visually identify where positive pressure head was building up in this scenario compared to other scenarios. In this case, the main source of the the mitigated overland flow peak was due to pervious pavement located in ‘reverse crowned’ alleyways, so I made a cross section through the the alleyway to see how pressure head is propogating through the subsurface. In the above gif animation, you can see the positive pressure head (red) propagating down through the alley on the left hand side (to the left of the “burned in” pipe in the center of the slice). (Building footprints are visible at the surface in green because their imperviousness and higher manning’s coefficient direct overland flows and positive pressures away more quickly). The pervious pavement in the alley not only treated rain falling on the alleyway, but was also intercepting runoff from adjacent pervious and impervious areas. In fact, the build up of positive pressure in the layers underlying the alleyway even after the original peak suggest a prolonged subsurface flow interception of flows even after the initial peak response has passed. This observation is now leading me to think more about how to capture heterogeneity among different timescales of response and how this relates to spatial heterogeneity in urban watersheds.
In addition, such high resolution ecohydrological modeling (hydrological modeling that also accounts for the distribution of water in vegetation) will be increasingly important in quantifying other benefits of “green infrastructure” planning, such as urban microclimate control and urban heat island effect mitigation. The spatial placement and configuration of infiltration-based stormwater control measures, and thus spatially dependent changes to evapotranspiration rates are important to measures to quantify and have the potential to be highly variable even at relatively small spatial scales.