(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .
Improved urban runoff prediction using high-resolution land-use, imperviousness, and stormwater infrastructure data applied to a process-based ecohydrological model [1]
['Jonathan Halama', 'U.S. Environmental Protection Agency', 'Corvallis', 'Or', 'United States Of America', 'Robert Mckane', 'Bradley Barnhart', 'Ncasi', 'Paul Pettus', 'Allen Brookes']
Date: 2023-12
VELMA is a spatially explicit ecohydrological model that includes multiple submodels representing integrated environmental components (hydrology and terrestrial biogeochemistry) to simulate the impacts of various stressors on dynamic responses of vegetation, soil, and water [21, 22]. VELMA parameters and watershed-scale simulation setup details are described in the 2014 VELMA 2.0 user manual [21]. Here we highlight key aspects of the model that were leveraged or enhanced to adapt VELMA for modeling urban watersheds. VELMA’s spatially explicit, gridded framework consists of a 2D surface layer used to model surface runoff on impervious to semi-pervious surfaces with four vertically delineated subsurface layers that allow water infiltration (Fig 2). The four soil layers represent vadose zone soils with user defined layer thicknesses. VELMA does not model water within deep (aquifer) layers. Listed in the following table are the soil parameters with the most influence on urban watershed modeling advancements (Table 1).
We applied VELMA to the Longfellow Creek watershed using a 5-m grid to simulate effects of fine-scale infrastructure features on urban runoff. Each 5-m cell location was assigned a land cover/vegetation type based on City of Seattle high-resolution GIS data [ 5 , 28 ]. Watershed vegetation cover types included trees, grass, and impervious surfaces such as roads, parking lots, and roofs. Plant biomass exists on all cells within the model, even if biomass is very low such as moss and small weeds. Growth on impervious surfaces occurs and if left undisturbed will reach unrealistically high levels. Therefore, all biomass growing on impervious surfaces were regularly suppressed through a biomass harvest/removal disturbance routine designed to maintain near bare surface conditions. This setup also left open the possibility of conducting future green roof simulations like the VELMA urban green roof applications described by Barnhart et al. [ 5 ].
Soil layers 1 through 4 represent the four subsurface soil matrix layers. To simulate subsurface hydrology, each soil layer has a set of parameters used to mimic the movement of water through the soil matrix ( Table 1 ). These parameters should all be considered and possibly calibrated when shifting a VELMA calibration from a nonurban to urban application. This was accomplished for Longfellow Creek watershed.
Urban watersheds contain non-natural components that impact hydrology as well. These include stormwater systems such as drains, piping, and outflows as well as other semi-natural features such as urban lakes and detention ponds. In addition, urban watersheds contain a higher percent of impervious surfaces that prevent water penetration into the soil matrix, route water, and shift evaporation rates. Modeling of these features found in urban environments were built into VELMA and are described below.
Evapotranspiration in VELMA is simulated via its Plant-Soil-Model (PSM) sub-model. PSM manages vegetation growth including the uptake of water and nutrients from the soil matrix, while the Water-Balance sub-model is responsible for routing and tracking water [ 21 ]. Briefly, plant roots take up water as a function of several parameters (amount of available water between field capacity and wilting point, plant root biomass, and others). Evapotranspiration (ET) is simulated as soil water taken up and transpired by plants to the atmosphere (“T” in ET), plus surface water evaporated (“E”) to the atmosphere. For the US Pacific Northwest region ET is approximately 20–30% of the total water balance [ 29 , 30 ].
At the start of each VELMA simulation day, a driver file adds precipitation (mm) to the surface layer of all receiving grid cells. For every cell within the simulated watershed, the daily amount of water transferred vertically from the surface to the uppermost (layer-1) soil layer depends upon the soil’s unsaturated storage capacity ( Fig 2 ). If the uppermost soil layer is saturated, water is transferred laterally along the surface layer to downslope cell neighbors according to elevation difference (i.e., per calculated gravitational flow paths, except when street curbs redirect water transversely across natural flow paths). Within the soil, water infiltrates vertically, layer to layer within a cell’s 4-layer soil column, again dependent on the receiving cell’s unsaturated storage capacity. Otherwise, water exceeding saturation capacity moves downslope laterally to cell neighbors. VELMA calculates daily changes in the degree of saturation within every soil layer within each watershed cell based on soil porosity, bulk density, layer thickness, total soil column depth, and calibrated parameters ( Fig 2 ). Finally, modeled surface and subsurface flow paths are used to categorize cells as “channel” or “non-channel”, such that daily surface and subsurface water transfers from non-channel to stream channel cells are reported as watershed runoff for that day.
The VELMA 2.0 model was designed to simulate natural watershed hydrology processes through two core sub-models, “Water-Balance” and “Plant-Soil-Model”. Water-Balance tracks the state of water volume in five “spatial pools” representing the surface to near surface terrain [ 21 ]. The first (top) pool represents the 2D surface layer, while the landscape’s soil matrix is comprised of the four subsurface 3D voxels (volume cells). The cell (grid) size of these pools can be set to essentially any resolution that meets a model user’s objectives and for which spatial data are available–most commonly 10-m and 30-m grids due to ease of access to publicly available landcover, soils, climate, and other spatial data at these scales. Due to the detail required to explicitly represent roads impact on hydrology within the Longfellow Creek watershed, the simulations presented here used a 5-m resolution.
2.2.4. Enhanced framework for urban modeling applications.
VELMA has reliably estimated hydrological and biogeochemical responses to change in natural systems but has lacked important stormwater infrastructure and other features to accurately simulate flashier patterns of runoff observed in urban watersheds. To improve the applicability of VELMA for urban watersheds, we enhanced the model in four ways:
Enhancement 1: Inclusion of partial to fully impervious surface.
Enhancement 2: Inclusion of piped network (e.g., MS4 and CSS stormwater systems).
Enhancement 3: Inclusion of curbed streets.
Enhancement 4: Impervious surface evaporation
VELMA’s Water-Balance sub-model carries out the model’s explicit cell-to-cell transfer of water, calculates the daily flow volumes, and summarizes the daily flow results as annual statistics. This core section of the model’s functionality was left intact to ensure the model still functioned well within natural landscapes. The water fate and transport enhancements to allow urban watershed applications was carried out through the addition of an impervious map and through the inclusion of a stormwater water disturbance feature, both described below.
VELMA Enhancement 1: Impervious surface. The VELMA Model possesses a surface representation for modeling precipitation, snow, air temperature, and surface organic material such as detritus, though a surface imperviousness representation did not exist. The surface layer was enhanced to enable users to specify surface layer perviousness per grid cell via a map, whereby a cell’s perviousness can be assigned as a floating-point map value between 1.0 (100% pervious) to 0.0 (100% impervious) (Fig 3).
This provided a mechanism to control water infiltration from the surface layer into the first soil layer. Any water not allowed to infiltrate remains on the surface and becomes excess water that will transfer to neighboring downslope cells.
An urban watershed setup can include an imperviousness surface with values ranging between 0.0 and 1.0. For the Longfellow Creek watershed setup all road and building cells were assigned a value of 0.0 (fully impervious) due to the 5-m resolution allowing homogeneous representation of those coverage types (Fig 3).
VELMA Enhancement 2: Curbed streets. Street systems within an urban landscape can include curbs or drainage ditches that route water along roadways to curb stormwater drain inlets or catchment basins. These engineered systems shift water away from natural (gravitational) flow pathways to instead transfer water either 1) unnaturally elsewhere within the watershed delineation (e.g., stream, detention pond, stormwater overflow vaults) shifting the hydrograph peak and timing, or 2) diverting water outside of the watershed delineation. When externally diverted water enters the stream system below the monitoring location or is routed outside the watershed delineation, that water is removed from the originating watershed’s hydrograph. To accurately represent water routing through both the natural landscape and engineered stormwater systems the water must be explicitly routed along the surface and through the watersheds infrastructure but executed within the constraints of the model’s framework. Curbed street network was representation within the Longfellow Creek watershed setup (Fig 4).
Within VELMA, water movement along curbs and within stormwater systems is processed by a cell-to-cell water transfer disturbance. If a cell-to-cell curb disturbance is in place, and water is present at that cell’s surface layer, the water transfer disturbance action will transfer water from cell-A to cell-B based on either the water percentage or water volume controlled by the disturbance action. The cell-to-cell transfer and water percentage or volume is all provided by the model user during the “Water-Disturbance” action setup.
The details of these water transfers are closely intertwined with stormwater system water transfers. The curb-to-curb, curb-to-inlet, and stormwater inlet-to-outlet are typically represented in one disturbance file and built into the same Water-Disturbance, therefore further details on both are described in the following section.
VELMA Enhancement 3: Piped network. Stormwater piping systems greatly influence how water routes through and out of a watershed. To enhance VELMA urban stormwater modeling capabilities, high-resolution GIS data have been added to represent spatially explicit stormwater inlets, pipe network, and outflows [26–28]. Within VELMA’s framework, stormwater movement is explicit regarding the starting inlet and ending outflow locations. VELMA is limited to only controlling the volume of water transferred within the pipe at a daily timestep, unlike SWMM that explicitly includes pipe diameter and length, stormwater vaults, and simulates water movement on an hourly timestep [31]. Within a VELMA simulation timestep, if a cell has been designated as a curb cell or stormwater inlet cell, any available surface water will be transferred from said cell to its corresponding transfer cell. This transfer of water does not explicitly account for stormwater carrying capacity due to limitations in pipe diameter or length that could result in the stormwater system being overwhelmed within less than a one-day timestep.
These urban water controlling features are provided to the simulation through a CSV data file dictating water and chemical transfers from cell-A to cell-B water due to engineered developments (e.g., stormwater piping, culverts, curbs, piped roof drainage) (Fig 5 and Table 2). Data detailed in this water disturbance CSV file dictates the i-index (1D unique value within a 2D grid) location of where water moves from, volume or percent of water transferred, and the cell location[s] water is transferred.
PPT PowerPoint slide
PNG larger image
TIFF original image Download: Fig 5. Water-disturbance explicit water transfer examples. VELMA is a fully gridded spatial framework, but for simplicity only small specific set of cells are represented in this map image overlay to convey the mechanism of curb and stormwater transfer actions occuring within VELMA during a simulation. Values 1 through 4 in this map image correspond to the first column of values in Table 2 [23].
https://doi.org/10.1371/journal.pwat.0000155.g005
A brief representation of transfers is presented with corresponding transfers described (Fig 5 and Table 2). Transfer numbers ‘1’ and ‘2’ denote curb-to-curb transfers occurring along the roadway to number ‘3’ Curb-to-Catchment transfer followed by the number ‘4’ Catchment-to-Outflow transfer.
For all water disturbances prescribed in a disturbance table, water transfers can only occur when water is present on the surface. For curb-to-curb water disturbance actions, these water transfers emulate the lateral movement of water downslope along the surface of a curbed street, usually to a stormwater inlet or drainage area.
Urban watersheds often possess stormwater systems that route water in directions contrary to natural flow paths (Fig 6). The VELMA disturbance module called “WaterDrainDisturbanceModel” was developed to quantitatively describe spatial and temporal transfers associated with this new water routing mechanism. This disturbance routine allows for all or a portion of any surface water at an assigned surface cell to be transferred to any other assigned surface cell.
Featured here are three uses of the “WaterDrainDisturbanceModel” to realistically represent the movement of water through the upper delineation of Longfellow Creek watershed to STA099.
Curb-to-curb disturbance: this disturbance moves surface water from one road cell to the next in a sequence representing lateral movement of water downhill along a curbed street.
Stormwater inlet-to-outlet disturbance: this disturbance transfers water from multiple street stormwater inlets to the stormwater outlet designated by the stormwater GIS pipe data.
Roof Drainage inlet-to-outlet disturbance: this disturbance transfers all water from roofs that have corresponding “lateral drainage” pipes out of the delineation.
Within these piped networks, the transfer of water for three different cases can be represented, where: 1) a singular input location transfers water to another singular outflow location (e.g., culverts, or tunneled streams), 2) multiple input locations transfer water to a singular outflow location (e.g., stormwater system inlets routing to one outflow location, inputs or drainage fields routed to a detention pond), and lastly 3) multiple input locations transfer water to multiple outflow locations (e.g., small to large stormwater system with multiple outflow locations, segmented stormwater system where water outflows into a stream, but then downstream water is piped again through a tunneled stream). All these installations alter the timing of water transporting through the landscape, whether through urban impervious areas or natural landscapes. The curb-to-curb action is crucial for accurately modeling water movement on impervious roadways; plus is vital for simulating urban contaminant transfers along the same roadways.
If the final curb-to-curb cell transfer in the sequence is to a drainage field or a bioswale, the water transferred will work vertically downward into the soil matrix. If the final cell transfer is curb-to-inlet, the transferred water will enter an explicit location within the stormwater system and transfer directly to an explicit outflow location, as designated by the stormwater GIS data.
VELMA Enhancement 4: Surface evaporation. Direct surface evaporation can be a considerable portion of a watershed’s total water budget. VELMA 2.0 modeled ET as the sum of evaporation of water on land surfaces plus plant transpiration, i.e., the process by which plants utilize soil water and evaporate it through pores in leaf surfaces. However, while VELMA 2.0 explicitly (mechanistically) simulated transpiration, evaporative losses from land surfaces are implicitly modeled based on watershed-scale water balance calculations. Owing to the importance of urban impervious surfaces evaporation, VELMA now explicitly models surface evaporative loss independent of plant evapotranspiration.
As landscapes transform from natural to urbanized, there is a large shift from vegetated soils to soils covered by impervious surfaces (e.g., concrete, asphalt, brick). This landscape change results in a new environment characteristic that possesses the process of direct evaporation from the surface to the atmosphere. To represent this new process the SurfaceEvaporation model was created and added to VELMA as a map-based component of a simulation. When SurfaceEvaporation is included in a simulation run, surface evaporation (mm/d) is calculated for each cell and tracked as a contributing component to total evapotranspiration at each simulated daily time step.
Surface evaporation is calculated using a US Class A Pan Evaporative approach derived by Linacre [32]. This approach requires dew point, an environmental metric often not measured at weather stations. To overcome that data limitation, Eccel’s observation that “when the daily minimum water vapor is saturated” the minimum air temperature “reaches dew point”, meaning that daily minimum temperature can be used as a reasonable surrogate for dew point [33].
Surface evaporation is calculated as: Eq 1 where EV is the surface evaporation (mm), T is temperature (°C), Z is elevation (m), I is irradiance (W/m2), F is change in air density by elevation (m), Ws is wind speed (m/sec), and Td is minimum temperature (°C), the surrogate for dew point. The variable F is calculated as: Eq 2 where Z is elevation (m).
Surface evaporation from two locations were compared to demonstrate the difference loss to the atmosphere due to the inclusion of the SurfaceEvaporation model (Fig 7). The evaporation at both the pervious and impervious cell locations start as similar quantities during the first three months of the year (winter season), though at low levels due to the colder weather and as jagged behavior for the impervious location. Surface evaporation increases as the air temperatures increase throughout the spring season. As rainfall events become less frequent and air temperatures increase into the summer season, the pervious cell undergoes less surface evaporation, while the impervious cells surface evaporation quantity increases to as high as 8 mm/d (Fig 7).
[END]
---
[1] Url:
https://journals.plos.org/water/article?id=10.1371/journal.pwat.0000155
Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.
via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/
