w james paper C137

C154mg complete paper

| Comments? | Š1996,7 William James | updated 97/03/03 |

THE EVOLUTION OF CONTINUOUS STORMWATER MODELLING: OVERCOMING MODERN BARRIERS

Michael A. Gregory, P.Eng., Water Resources Engineer, Camp Dresser & McKee Inc., One Tampa City Center, Suite 1750, Tampa, FL 33602, USA.

William James, D.Sc., Ph.D., P.Eng., FASCE. Professor of Environmental and Water Resources Engineering, University of Guelph, Guelph, ON, Canada N1G 2W1.

Michael F. Schmidt, P.E., Principal Water Resources Engineer, and

Brett A. Cunningham, P.E., Senior Water Resources Engineer. Camp Dresser & McKee Inc., 6650 Southpoint Parkway, Suite 330, Jacksonville, FL 32216, USA.

Abstract

This paper reviews the evolution of long-term, continuous water resources and stormwater

modelling, described from the perspective of overcoming practical barriers. Traditional challenges

included lengthy computing times, complicated software, unavailable data, and large amounts of

data interpretation; challenges that were magnified by budget and schedule limitations. Many of

these have been overcome through recent technological advances in computing and information

science. Some obstacles remain, however, such as data management constraints. These constraints

are lessened by new graphical user-interfaces, which help to manipulate and transfer continuous

stormwater modelling data more efficiently.

Introduction

To simulate the response of stormwater runoff in urban and natural systems, engineers may use

either event or continuous hydrologic models. Event models simulate only discrete wet events,

and are often distinguished by output of a single runoff hydrograph computed in response to a

single rainfall event. Rainfall frequencies and design storm distributions are based on simple

characteristics of highly variable observed storm events, and the frequency of the computed

hydrograph is assumed to be the same as the rainfall. This approach ignores the dependence of

computed responses on input assumptions of antecedent hydrologic conditions and other factors,

and has been questioned (Litwin and Donigian, 1978; Adams and Howard, 1985; Huber et al.,

1986). Since the real inter-event dry periods are not generally related mathematically to the input

or design storm, the assumed start-up conditions are arbitrary (James and Robinson, 1986).

Examples of event-based models include the rational method, TR-20 from the United States (US)

Soil Conservation Service, and HEC-1 from the US Army Corps of Engineers.

Continuous stormwater models take into account a continuous water budget in the system, and

may be applied to meteorological datasets from weeks to several decades in duration. Since these

models are able to simulate hydrologic processes between rainfall events (for example,

evaporation and recovery of soilwater storage), the modeller does not have to make as many key

assumptions about start-up conditions for wet weather processes. Furthermore, the allocation of

runoff hydrograph frequencies is improved.

Examples of continuous models include the Storm Water Management Model (SWMM) and the

Hydrological Simulation Program - Fortran (HSP-F) from the US Environmental Protection

Agency; and the Storage, Treatment, Overflow, Runoff Model (STORM) from the US Army

Corps of Engineers. With these models, a dry weather timestep of one hour or more is typically

used because of the slow speed of inter-event processes; wet weather timesteps are chosen

according to the catchment response time and resolution of available data, generally five to fifteen

minutes.

Continuous simulation has been found useful for many design objectives (James and Robinson,

1986; Schmidt et al., 1993; CDM, 1995). Design objectives include:

Compute the number and duration of flow or volume exceedances for evaluation of erosion and

channel stability.

Evaluate water quality impacts on receiving waters.

Evaluate the impact of stormwater management alternatives.

Evaluate the performance of storage facilities.

Compute the number, duration and magnitude of diversions such as combined sewer overflows.

Long-term continuous modelling is most appropriate when addressing environmental issues, such

as the study of groundwater and surface water interactions (Schmidt et al., 1995a); for water

quality problems associated with nonpoint source pollution (Litwin and Donigian, 1978; Pitt,

1995); or for selecting antecedent moisture and baseflow conditions in flood routing studies

(Schmidt et al., 1995b). Since different rainfall events cause different pollutant washoff, transport,

and kinetic interactions, event-based methods are inappropriate for use with water quality

standards (Medina, 1987). Furthermore, whereas event-based methods have largely been

developed using data for rare storm events, it has been shown that water quality standards are

often violated during small, relatively common storm events (Pitt, 1986).

Event-based models have a place in engineering practice, however, many water resource,

erosional, and water quality effects require continuous simulation to fully comprehend the many

complex physical interactions within the system. Event-based methods are useful for teaching

fundamental hydrologic and hydraulic concepts. Yet in school, analysis and design economy is

often the focus of water resources engineering courses, and so very little coverage is given to

continuous modelling in textbooks. In practice, modellers must be aware of the dangers in

adopting short-term convenience and economy when prescribing solutions to long-term problems.

Solutions are best achieved by trying to understand the problem, not by simplifying the problem

solely in the interest of minimizing costs.

Traditional Barriers

Continuous stormwater modelling is not a new idea. Over thirty years ago, the Stanford

Watershed Model was developed and applied to large watersheds (Crawford and Linsley, 1966).

However, event-based models have continued their hold on the mainstream of water resources

engineering.

In practice, many barriers against adopting continuous modelling methods for analysis and design

of stormwater systems have been identified over the years. Technological advances in computing,

and information acquisition and exchange have helped to overcome many of these barriers.

Computing Power

Stormwater modelling has traditionally been concerned only with minimum, maximum, or average

water quantity or quality conditions during selected events. In the infancy of personal computers,

there were valid arguments about the feasibility of long-term continuous simulation, resulting

from deficiencies in available computer memory and processing speed. This argument has

deteriorated as high-power and less-expensive computers have overwhelmed the marketplace.

For example, test trials of 75-year simulations using SWMM (25 catchments, 25 pipes, 7

pollutants, 5 land uses, and an hourly timestep) took 7.5 hours on a 486DX-66 MHz computer,

compared to 67 hours on a 386DX-22 MHz computer (Kuch and James, 1993). Performance

results have improved further still, with 32-bit versions of SWMM run on computers with

Pentium processors.

Data Availability

Many opponents of continuous modelling have argued that there is a lack of input data for

calibrating and applying such models (Cao et al., 1994). New media, such as the Internet and CD-

ROM, have challenged conventional digital data media, offering open access to widely distributed

information systems and providing reliable, high capacity and high transfer speeds at low cost

(Gregory and James, 1995).

Long-term hydrologic datasets are available on CD-ROM from many public and private agencies.

In addition, a number of agencies offer World Wide Web home pages on the Internet from which

visitors can download datasets. The US Geological Survey and the US National Climatic Data

Center both offer data products on CD-ROM and via the Internet.

Modelling Expertise

It has been argued that continuous stormwater models are often too complicated, requiring

special expertise to operate (Marsalek and Sztruhar, 1994). Evolving computing environments

have led to improved user-interfaces, on-line technical assistance tools, and other software to

assist modellers (Gregory and James, 1995).

There is also a notion that calibration of continuous models is complicated, and fundamentally

different from the calibration of event models (Nix, 1994). However, it has been shown that, for

models structured in the manner of SWMM, the interaction of dry and wet processes may be

handled directly (James, 1993; Schmidt et al., 1995b).

Modern Barriers

Although powerful computers have helped to eliminate past arguments, new challenges have

arisen.

Model Complexity

Researchers are concerned about the complexity of continuous stormwater models like SWMM

or HSP-F. Models should be as complex as necessary to provide the answers needed to make

management decisions. Additionally, model complexity should depend on the nature of available

data and the specific algorithms used in the selected model. It has been suggested that, in certain

cases, some of the simpler approaches may be more appropriate to the level of available data for

many management decisions (Grayson et al., 1992).

As an indicator of the forces against continuous modelling, it has been proposed that simpler, less

data-intensive models provide as good or even better predictions than more physically-based

models (Jakeman and Hornberger, 1993). Yet, despite this advice to not make stormwater

modelling more complicated than it needs to be, Klemes (1986) warns us of the consequences of

ignoring the fundamentals, stating that ...the credibility of hydrologic models can rest only on at

least approximately correct rendering of the true dynamics of these processes.... Indeed, it may be

that our ethical responsibility requires that we seek a better understanding of hydrologic

complexity, and not ignore it by adopting simplistic methods (James, 1994).

Data Management

Long-term continuous stormwater modelling is not so much a matter of added complexity, rather,

it is more a matter of inconvenience for the modeller since more effort is required. With such

models, there is not much difference between event and long-term continuous modelling, except

that more processes must be calibrated, and more data must be managed. Focus therefore shifts

from model complexity to the provision of software utilities that assist with calibration and data

management.

While the old barriers are coming down, there appears to be one more hurdle in the evolution of

continuous modelling: the increasingly troublesome task of data management. Long-term

continuous modelling is impeded by the variety of scales and formats of modelling data, and the

task of organizing large volumes of time-series data (observations or computations that have been

made sequentially in time). Large sets of hydro-meteorological time-series data are required as

input, and large datasets are created as output.

Time-series data must be manipulated as contiguous blocks of related data, as opposed to being

addressed as individual data elements in related tables, as with commercial database systems. Two

time-series data management systems that have gained widespread use are ANNIE from the US

Geological Survey, and the Hydrologic Engineering Center Data Storage System (HECDSS)

from the US Army Corps of Engineers.

To satisfy the need for a data management system for continuous stormwater modelling data, a

graphical user-interface that automatically manages and processes long-term datasets was

developed (Gregory and James, 1995). Known as CASCCADE, the user-interface includes

several utilities:

File retrieval operations that let the user select hourly rainfall datafiles from an archival ANNIE

database. Files can be exported in a variety of formats for use by SWMM.

File archival operations that allow users to select various SWMM-formatted datafiles and

import them into an ANNIE database.

Data analysis operations that automatically process input files, launch the SWMM program, and

summarize statistical results.

Over 140 years of hourly rainfall data were used to test CASCCADE. It was found that between

150,000 and 200,000 years of hourly rainfall data from a single weather station could be stored on

one CD-ROM, or the same data could be transmitted via the Internet in only 12 hours during a

low-demand period (Gregory and James, 1995). In contrast, conventional data storage and

retrieval methods (including such tasks as submitting a formal request to the archiving agency,

and then waiting to receive the diskettes by mail) could take days or weeks.

This study helped reveal the practicality of long-term continuous modelling by overcoming data

management barriers. It also integrated a wide community of users by bringing together diverse

stormwater modelling applications and database systems. Further enhancements have been added,

resulting in the development of CASCADE2 (Wang, 1996). Together, CASCCADE and

CASCADE2 effectively link two popular continuous stormwater models, SWMM and HSP-F,

with two popular time-series data management systems, ANNIE and HECDSS.

By using standardized data formats, it was shown that the transfer of data between various

modelling applications was not constrained by the limitations imposed by the individual models. In

this way, model revisions and even new software could be accommodated. Such an approach

recognizes that technological advances will continue to evolve, and that modelling applications

must adapt accordingly.

Conclusions

Continuous simulation is a fundamental tool in the long-term assessment of urban and natural

water resource management systems. Continuous models have evolved in response to new and

emerging computer technologies. Unfortunately, event-based models continue to be used in the

same way, with the same data and the same assumptions as they were used long before computers

arrived. Event-based models remain popular even though they often ignore or greatly simplify

important hydrologic processes, and cannot be used to predict long-term environmental impacts

or interactions in a complex system.

The intent of this paper is not to deride event-based hydrology methods, but rather to address

practical barriers against the use of continuous hydrology models. Simple event-based methods

may be used for certain design objectives, but only with due regard to the uncertainty and

variability of physical processes. By applying continuous modelling methods, the modeller is able

to better recognize such issues.

Few barriers to long-term continuous modelling remain; therefore it is now more practical to use

continuous modelling methods when addressing the impact of urban systems on aquatic habitat.

References

Adams, B.J., and Howard, C.D.D. (1985). The pathology of design storms. University of

Toronto, Department of Civil Engineering, Publication No. 85-03, 32 pages.

Cao, C., Piga, E., and Saba, A. (1994). A continuous simulation approach to design storm

calibration. Water Science and Technology, Journal of the International Association on Water

Quality, Vol. 29(1-2), pages 11-20.

CDM (1995). Miami International Airport Terminal Area Water Quality Treatment System.

Preliminary Design Report for Dade County Aviation Department.

Crawford, N.H., and Linsley, R.K. (1966). Digital simulation in hydrology: Stanford watershed

model IV. Stanford University, Department of Civil Engineering, Technical Report No. 39, 210

pages.

Grayson, R.B., Moore, I.D., and McMahon, T.A. (1992). Physically based hydrologic modelling:

2. Is the concept realistic? Water Resources Research, Vol. 26(10), pages 2659-2666.

Gregory, M., and James, W. (1995). Management of time-series data for long-term, continuous

stormwater modelling. In Advances in Modelling the Management of Stormwater Impacts, W.

James (Editor), CHI Publications, ISBN 0 9697422-5-8, pages 115-151.

Huber, W.C., Cunningham, B.A., and Cavender, K.A. (1986). Continuous SWMM modelling for

selection of design events. In Urban Drainage Modelling, C. Maksimovic and M. Radojkovic

(Editors), Pergamon Press, ISBN 0-08-032558-0, pages 379-390.

Jakeman, A.J., and Hornberger, G.M. (1993). How much complexity is warranted in a rainfall-

runoff model? Water Resources Research, Vol. 29(8), pages 2637-2649.

James, W. (1993). Rules for responsible modeling. CHI Publications, Report R-184, 100 pages.

James, W. (1994). On reasons why traditional single-valued, single-event hydrology (typical

design storm methodology) has become simple-minded, dishonest, and unethical. In Urban

Hydrology and Hydraulics Workshop, US Army Corps of Engineers, Report No. SP-26, pages

169-181.

James, W., and Robinson, M.A. (1986). Continuous deterministic urban runoff modelling. In

Urban Drainage Modelling, C. Maksimovic and M. Radojkovic (Editors), Pergamon Press, ISBN

0-08-032558-0, pages 347-378.

Klemes, V. (1986). Dilettantism in hydrology: Transition or destiny? Water Resources Research,

Vol. 22(9), pages 177S-188S.

Kuch, A.W., and James, W. (1993). A sensitivity analysis tool for three generation continuous

urban stormwater quality management modelling. Proceedings of the Runoff Quantity and Quality

Model Group Conference (November, 1993), Reno, NV, American Society of Civil Engineers.

Litwin, Y.J., and Donigian Jr., A.S. (1978). Continuous simulation of nonpoint pollution. Journal

of the Water Pollution Control Federation, Vol. 50(10), pages 2348-2361.

Marsalek, J., and Sztruhar, D. (1994). Urban drainage: Review of contemporary approaches.

Water Science and Technology, Journal of the International Association on Water Quality, Vol.

29(1-2), pages 1-10.

Medina Jr., M.A. (1987). Water quality modelling and the regulation environment. In Pollution

Control Planning, W. James (Editor), Computational Hydraulics Inc., pages 176-191.

Nix, S. J. (1994). Urban stormwater modeling and simulation. Lewis Publishers, ISBN 0-87371-

527-6, 220 pages.

Pitt, R. (1986). Runoff controls in Wisconsin's priority watersheds. In Urban Runoff Quality:

Impact And Quality Enhancement Technology, B. Urbonas and L.A. Roesner (Editors), American

Society of Civil Engineers; ISBN 0-87262-577-X, pages 290-313.

Pitt, R. (1995). Effects of urban runoff on aquatic biota. In Handbook Of Ecotoxicology, D.J.

Hoffman et al. (Editors), CRC Press Inc., ISBN 0-87371-585-3, pages 609-630.

Schmidt, M.F., Quasebarth, T.F., and Johnson, T.J. (1993). An evaluation of detention pond

design using continuous simulation. Proceedings from the conference on Hydrology and

Hydrogeology of Urban and Urbanizing Areas (Boston, 1996); American Institute of Hydrology;

pages BMP1-BMP11.

Schmidt, M.F., Cunningham, B.A., and Mack, B.W. (1995a). The use of continuous SWMM for

water resources conservation planning. Proceedings of the 4th Biennial Stormwater Research

Conference (October, 1995), Clearwater, FL, Southwest Florida Water Management District.

Schmidt, M.F., Bergman, M.J., Smith, D.R., and Cunningham, B.A. (1995b). Calibration of

SWMM-EXTRAN using short-term continuous simulation. In Advances in Modelling the

Management of Stormwater Impacts, W. James (Editor), CHI Publications, ISBN 0 9697422-5-8,

pages 15-26.

Wang, Y. (1996). Integration of the US Army Corps of Engineers time-series data management

system with continuous SWMM modelling. M.Sc. Thesis, University of Guelph.