|
|
P8 Urban Catchment Model
|
Introduction
P8 simulates the generation and transport of stormwater runoff pollutants in urban watersheds. Continuous water-balance and mass-balance calculations are performed on a user-defined drainage system consisting of the following elements:
Simulations are driven by hourly precipitation and daily air temperature time series. Runoff contributions from snowmelt are also simulated.
'P8' abbreviates "Program for Predicting Polluting Particle Passage Through Pits, Puddles, and Ponds", which more or less captures the basic features and functions of the model. It has been developed for use by engineers and planners in designing and evaluating runoff treatment schemes for existing or proposed urban developments. Design objectives are typically expressed in terms of percentage reduction in suspended solids or other water quality component.
Despite its limitations, P8 has been used by state and local regulatory agencies as a consistent framework for evaluating proposed developments. Depending on applications, other models could be either too simple (easily used, but ignoring important factors) or too complex (requiring considerable site-specific data and/or user expertise). P8 attempts to strike a balance to between those extremes.
Predicted water quality components include total suspended solids (sum of the individual particle fractions), total phosphorus, total Kjeldahl nitrogen, copper, lead, zinc, and total hydrocarbons. Simulated BMP types include detention ponds (wet, dry, extended), infiltration basins, swales, buffer strips, or other devices with user-specified stage/discharge curves and infiltration rates. A simple water budget algorithm can be used to estimate groundwater storage and stream base flow in watershed-scale applications.
Initial calibrations were based upon runoff quality and particle settling velocity data collected under the EPA's Nationwide Urban Runoff Program (Athayede et al., 1983). Calibrations to impervious area runoff parameters for Wisconsin watersheds have been subsequently developed.
Inputs are structured in terms which should be familiar to planners and engineers involved in hydrologic evaluation. Several tabular and graphic output formats are provided.
History &
Documentation
P8 was originally developed for the MSDOS operating system in 1990 ( Walker, 1990 ). The last MSDOS version (V2.4) was released in 1998. P8 was converted to the Windows operating system in 2006. The more recent Windows version is described here.
The conversion involved recoding from MSDOS Fortran to Visual Basic 2005. If not already installed on the user's computer, the Microsoft .NET framework (Version 2.0 or later) required to run the windows version is automatically downloaded from the Internet when P8 is installed. Microsoft Excel (Version 2003) is used for output and graphics and must be pre-installed on the user's computer.
The interface , input formats, and output formats have changed significantly relative to the MSDOS version. The underlying model structure and calibrations have not been changed. Future versions of P8 may incorporate updates of the calibrations and model features.
This document describes the windows application, but does not supercede all of the previous documentation. Detailed descriptions of the algorithms, calibration, and testing results are described in documentation for the DOS Version 1.0 ( p8v1doc.pdf) and the Version 2.0 update ( p8v2.pdf). Those reports and other current information on the model are posted at the P8 website ( wwalker.net/p8 ). To facilitate communications regarding updates etc, a page from the P8 website appears on the user's desktop each time the program is run.
Potential
Applications
Primary uses of the
model are based upon "relative" predictions, expressed in terms as percentage
removals:
These applications
are insensitive to errors associated with predicting untreated runoff water
quality and are therefore more accurate than predictions of concentrations or
loads. These results are sensitive, however, to the
particle properties , in particular the setting
velocities, which determine the removal rates at a given hydraulic load.
Default values for settling velocity and other particle characteristics are
provided. These are based upon data collected under the EPA's Nationwide
Urban Runoff Program (NURP, Athayde, 1983), as described in the
Version 1 documentation
.
Secondary uses of the model are based upon "absolute" predictions, expressed in terms of outflow, load, or concentration:
These types of applications are subject to greater error because of the high degree of site-to-site and storm-to-storm variability associated with urban runoff quality, as documented under the EPA's Nationwide Urban Runoff Program (Athayede et al., 1983). Local calibration may reduce absolute prediction error, but is rarely feasible. Differences in accuracy between relative and absolute predictions are not specific to P8 but characteristic of water quality models in general.
Cases
Each application or "Case" consists of a collection of watersheds and treatment devices, optionally linked in a one-dimensional branched network. Flow and particles (up to 5 classes, each with different settling velocities) are routed simultaneously from the watershed sources through the network of treatment devices.
Watersheds contain both pervious & impervious areas, defined by the following:
Runoff is driven by sum of rainfall & snowmelt. Particle buildup & washoff on impervious surfaces are modeled using equations derived largely from the EPA's Stormwater Management Model (SWMM). Runoff from pervious areas is predicted using the SCS Curve Number technique (USDA,SCS, 1964), as implemented in the GWLF model (Haith et al, 1992). Antecedent Moisture Conditions are adjusted based upon 5-day antecedent rainfall+snowmelt, air temperature, and season. Percolation is calculated by water balance. It can be ignored or routed to an aquifer device for simulation of baseflow. Watershed data can be defined using the P8 interface or imported from an external Excel file . The latter option can be used as an interface with GIS databases.
Devices are defined based upon factors
controlling hydraulic response and particle removal efficiency:
Specific inputs vary with
device types, which include:
Devices have up to three outlets (infiltration, normal outlet,
spillway/overflow), which are routed to other devices or out of the system,
typically to receiving waters. P8 keeps track of the overall
water and mass
balances of the entire device network, as well as those
of the individual devices.
Particle classes are defined based upon factors controlling watershed export & dynamics in treatment devices:
Default input values for particle classes are
provided, based upon calibration to "typical urban runoff" concentrations and
settling velocities measured under NURP (Athayede et al.,1983,1986; Driscoll,
1983).
Water Quality Components are defined based upon their weight distribution across particle classes, expressed in terms of milligrams per kilogram of particle mass. Default parameters for several water quality components are provided, based upon calibration to median, event-mean concentrations reported by NURP (Athayede et al., 1983).
Alternative particle/component calibrations are provided based upon the 50th and 90th percentiles of urban runoff concentrations measured under NURP. These can be used to generate "typical" and "worst-case" water quality predictions.
The following COMPONENTS are included in the initial calibrations:
If available, local or regional runoff quality data
may also be used for calibration.
Runoff simulations are driven by continuous hourly precipitation & daily air temperature time series. Runoff results from rainfall and snowmelt. A file of hourly precipitation values from the Providence Airport weather station is provided. A separate utility 'P8CONV.EXE' can be used to translate hourly precipitation data files obtained from the National Climatic Data Center and other sources.
Short storm sequences ("design storms") can be used in preliminary model runs. Experience with the model indicates that a reasonable approximation of long-term-average particle removal efficiency can be achieved by using a 1-inch, 24-hr, SCS Type II design storm with a 75-hour interval between storm midpoints.
Air temperatures are used in computing
evapotranspiration & snowfall/snowmelt. A file containing daily values from
Providence Airport (1969-1988) is provided.
Model Testing
The model has been tested against the following alternative models or
observed data sets (
Walker, 1990 ):
Model Limitations
Watershed simulations are limited by the following factors:
Device simulations are limited by the following factors:
General Limitations include:
* These limitations are associated more with measuring or specifying
input data than with the model itself.