C137 complete paper |
| | Comments? | ©1996,7 William James | updated 97/03/03 | |
ON REASONS WHY TRADITIONAL SINGLE-VALUED, SINGLE-EVENT HYDROLOGY (TYPICAL DESIGN STORM METHODOLOGY) HAS BECOME SIMPLE-MINDED, DISHONEST AND UNETHICAL William James, P.Eng., FASCE1 ABSTRACT The position taken in this paper is decidedly anthropo-centri-fugal, to coin a new word; it avoids placing society at the center of concern. Proceeding from a discussion of human populution, it describes some dimensions of ethical design for sustainable ecosystems. Arguments are presented that event modeling, and its associated design methodology, at best contributes to the destruction of aquatic ecosystems: the principal argument in favour of design storm methods is design economy, but cheap stormwater drainage design is an avoidance of consideration of the inevitable long-term ecological impacts of that design, tantamount to a deliberate decision to remain ignorant of the impacts of urban drainage system design. Eco-sensitive design on the other hand demands the adoption of continuous modeling. Even though computer hardware, software and expertise is now more than capable of supporting long-term continuous modeling, current engineering design manuals do not support this position. Never-the-less, professional engineering associations all over North America have recently adopted new ethical codes that require ecosystem-sensitive design. To effect timely adoption of eco-ethical design, all who suffer from impaired aquatic ecosystems, including engineers constrained to practice conventional hydrology exclusively, should seek or demand a comprehensive paradigm shift in stormwater management, away from traditional design storm methods, and towards the adoption of fuzzy, continuous models. Originally presented with a large number of visual illustrations, not reproduced here, the style of this paper is somewhat colloquial. INTRODUCTION Efforts of the writer's research group are directed towards the enhancement of modern continuous computational methods, such as the US EPA's Stormwater Management Model (SWMM, Huber and Dickinson, 1988), enhancements which have been seriously under-funded in the last ten years, a situation that continues. Nevertheless, the responsible and proper management of surface water pollution in North America and indeed, globally, is fundamentally important to all living things, and thus it is appropriate for us all to actively participate in the enhancement and wide adoption of methodology such as SWMM, especially together with the US Army's Hydrological Engineering Center, an office more renowned for software that is practical than eco-sensitive. Our work is graduate student research, funded basically by the Canadian national Natural Sciences and Engineering Research Council, and from sources where-ever there is a common interest. One example of our disparate activities, is our newsletter SWMM News & Notes (for which, incidentally, we are always soliciting contributions). We also sponsor an annual conference in the Toronto area on modeling the management of the impacts of urban stormwater, alternating six-monthly with a conference somewhere in the United States. Annually, the Canadian proceedings are published as a peer-refereed, archival book in the style of a hard-cover textbook, with extensive bibliographic tools. As well, we support three Internet list servers, including SWMM-USERS[@UOGUELPH.CA], which is moderately active, and has an extremely high signal-to-noise ratio. Before continuing the logical thrust of this paper, readers need to have the usage of certain terms clarified. Modeling has the simple purpose, in this discussion, of evaluating the likely performance of potential arrays of various best management practices (BMPs) in an impacted watershed, and determining the "best" array, according to some presumed, prevailing values. If we had prior knowledge of all such performance, sufficient to determine reasonably precise optimal BMP capacities, then modeling would of course be a waste of time. The corollary to this is that, if we had available prior field observations of the performance of all arrays of all types of BMPs for all aquatic systems and physiographies, again we would not need to model. Because these two conditions are infeasible, given the inherent complexity, we may conclude that modeling will remain necessary for the forseeable future. Processes underlying the BMP performance are imperfectly known, even for the as-is system, the best-known of the arrays to be modeled, and thus the computed model output is uncertain. The to-be (proposed interventions, the so-called what-if? scenarios) and as-was systems (pre-development, especially if pre-forest-clearance or pre-agricultural) are even more uncertain. This is especially true of surface water pollution processes. Field data monitoring is required to validate the models; if the model were certain for all arrays then monitoring would of course be a waste of time. Because even the best available codes are uncertain models of proposed BMP arrays, field data monitoring will also remain necessary for the forseeable future. In surface water quality modeling, models that disregard all dry-weather processes are referred to as event models, while models that include code for processes that are active during dry weather, such as pollutant build-up, evapo-transpiration, storage depletion, recovery of loss rates, and so on, are termed continuous models. Continuous models also usually include processes associated with winter seasons. Event modeling suffers the limitation that every run is governed by arbitrary assumptions of start-up conditions, which are themselves seldom subject to careful modeling scrutiny, such as sensitivity-, calibration, and error analysis (SCEA). Event models are obviously only run for short durations. To the extent that the effect of these initial conditions persists through the model run, the computed results may be unreliable. There is a great deal of hydrology literature asserting that these start-up effects are indeed important (James and Robinson, 1986a, provide a review). Of course, the fundamental premise of event hydrology is that the design storm probability is exactly equal to that of the associated flood event. However, studies have shown that this is not true. Klemes (1987) eloquently condemns event modelers who claim to be able to compute risk and reliability, referring to the association of a fixed probability to a design event as wishful thinking. Event modeling evolved in bygone times before computing, and it is simply no longer appropriate to adopt such simplistic methodology (James and Shivalingaiah, 1986). Modeling should always be continuous, rather than event-oriented, for all design inference. However, short runs for both dry and wet events, and events that are combinations thereof, are recommended for SCEA. In these cases there is no start-up error, because the initial state is given by the observed record. James (1993) argues that 75-year continuous modeling has now become feasible, indeed desirable, in order to address concerns of sustainability. In a landmark case, the Supreme Court of Canada upheld a decision in favour of downstream riparian land owners suffering fluviological impacts resulting from the urbanization of a large city. Arguments that helped to convince the judges were based on 47-year continuous modeling, whereas the losing side based their analyses on event hydrology and that for very few events, only one point on one event being used for calibration (James, in press). Event hydrology cannot be used to evaluate fluvial morphology downstream (James and Robinson, 1986b). Readers should be aware that this case is precedent-setting in the US as well. "POPULUTION" ISSUES In this section, we attempt to build a case against anthropocentrism in engineering design, based on empirical evidence that engineering design very often leads to landscape interventions that increase the human population locally, and that the process seems to have no imposed limits. Writing for the moment not as a technical expert, but as a full-time resident drinker of water downstream of major US conurbations, I am perhaps qualified to challenge some uncharted premises of municipal-engineering political-correctness: viz. that further growth and development is necessarily and unconditionally good. I can and do question whether we bipeds are not already too numerous in North America. Overall, the population density in the US part of North America is probably about an order of magnitude greater than that of the Canadian part. Significant problems arise when populations of 25 thousand (say) to 25 million people congregate in urban areas (Mexico City's population will soon be larger than all of Canada, and comparable to California). Modelers of best management practices for the management of the impacts of surface water pollution must, at some point, probably sooner rather than later, ask a leading question: are we doing enough in North America about containing the numbers of large mammals, especially humans, that directly cause a deterioration in the natural environment? When will there be development enough? Is a population of 7 million sufficient in SE Michigan, or Southern Ontario? Why not 70 or 700 million? Or is 700 thousand more reasonable? Why not 70 thousand for that matter? These seem like questions that are too big for engineers to answer (even, one suspects, army engineers), but we are involved because we write the computer codes that relate pollution to population activities, and that lead to development controls. What would be the basis for the answer to these questions of carrying-capacity vs. sustainability, other than the so-called sustainable quality-of-life? I have not yet met any individual who feels that local surface water quality, and concomitantly her/his quality of life, would be better if the local region developed from housing and employing, say, 7 thousand, to, say, 7 million. In fact the opposite seems clearly to be the human experience. This truth is evidently not universally acknowledged by the councils of our elected representatives, however, who very often make indirect use of the results of our water-environment models to encourage further development, and thus population growth. In this particular respect our models are not functioning: as they do not explicitly examine carrying capacity, they do not test all reasonable alternative management strategies. It is the large numbers of us people, especially the very large numbers in our North American cities, that destroy natural habitats. Pet and farm animal populations are also causing serious problems at several locations in our part of the globe. Processes now incorporated in our stormwater and water quality models do not make the connection (between population carrying capacity or land-use controls on the one hand, and aquatic ecosystem changes or pollution on the other) sufficiently pointedly; modelers need to be able to convince decision-makers that negative growth is a valid strategy that ought to be routinely examined. COMING TO TERMS WITH SUSTAINABILITY ISSUES In this section we attempt to develop the argument that continuous modeling is ecosystem-friendly, and that certain concerns can be met, especially if the simulation period is very long, say 75 years or so. The biggest problem in North America, as I have ventured to suggest, is loss of natural aquatic habitat. In my judgment, in the water environment, this is determined by three derivative environmental factors, above all others:
Such sustainability and eco-restoration factors are first and foremost computable by continuous modeling, or period-of-record modeling. Lack of progress in continuous modeling methodology has been truly amazing. Having attended these and related conferences and workshops, between five and ten per year for the past 30 years, I have heard and seen copious event modeling. It is as though event modeling still makes sense, or as though event modeling is easier than continuous modeling, or cheaper. It can be demonstrated that none of these assertions is true anymore. Norm Crawford published his dissertation on the Stanford Watershed Model in 1962, which is more than a generation ago, a long time given modern science and technology. Read my lips: that's 32 years! In this time, continuous modeling has again and again proven effective, reliable and cost-effective. Backing up a bit, we should define more carefully the term model (here a deterministic surface water quality model). For our purposes, a model of a drainage system is taken to be the combination, such as a SWMM application, of both (a) a program, together with (b) its input datafiles. Thus a model may have a useful life extending over decades. In this sense, SWMM is a misnomer. It is important to come to terms with the fact that both the program and the model evolve over the years. Thus models may become very complex, gradually integrating encyclopedic knowledge of component processes as it becomes available, applying it to vast databases as the databases build over time, examining potentially thousands of arrays of best management practices (BMPs) as they are put forward from time-to-time, re-running each array for say 75 years of hydro-meteorologic and physical topo-hydrographic data. A further rule may now be posed: A model can and should be tested for SCEA and the uncertainty always reported as a part of the computed output. Indeed not to do so, is likely a dereliction in design engineering ethics. In other words, it is simply dishonest not to report the uncertainty associated with any computed response. By comparison, the very concept of event modeling has become obsolete. Originally proposed in the late 19th century, and developed in the first half of the 20th century, long before automatic computers, the need for such simplistic logic has long gone. Even the use of the term implies that the (only) event is a wet event; everything else is an non-event, and nothing happens between rainfalls. Yet many processes that are active during dry periods are extremely important: recovery of infiltration capacity, recovery of storage, build-up of pollutants, to name a few. Some models have been developed recently that set up a series of alternating wet and dry events, as a kind of compromise with tradition. But this also leads to complications: how can the fuzzy periods between wet and dry events be accommodated by such continuous-event-hydrology? For significant periods of natural time there is very light precipitation and/or saturated humidity levels; during these periods it is not clear whether pollutants are building-up or washing-off. Process disaggregation into an artificial wet/dry dichotomy is unnecessary today, given modern computing. Of course, continuous modeling brings difficulties, primarily associated with data management. Fortunately interfaces with GIS systems, and with time series management systems such as HEC-DSS and ANNIE, are being developed in the stormwater management modeling arena. The integration of ANNIE with SWMM, making the interaction between SWMM and HSPF transparent, is long overdue. This is not a trivial challenge. Model developers must resist the pursuit of the simplistic. It is not true that simplistic models are easy to understand or apply: subsuming complexity into a simple coefficient is often difficult to explain. Too many of our colleagues still seem to be teaching, particularly in our civil engineering schools, that simplistic representation is itself somehow a worthy pursuit. It is difficult to convince practicing engineers that rolling infinite complexity into a simple fudge-factor is sometimes anti-intellectual. On the other hand, end-users should understand that only complex modeling, along with sensitivity analysis, parameter estimation, error analysis, and field data collection programs, can help us determine which processes are important, but only if all potentially relevant processes are represented in our models. Surely this is the correct intellectual approach. We must fight the notion, spawned from our BC (before computing) heritage, that the most simple representation possible is intellectually desirable. I remind you that this is 1994 in the United States (California at that) supposedly one of the most highly educated spots on earth - we ought not to be supporting the blinkered pursuit of the simplistic per se. My research group has decided to take sustainability, or, more accurately, modeling for sustainability, seriously. We believe that long-term modeling, of the order of 50 or 100 years, say three generations, is very important. We believe that this approach is timely, and credible, and in this we find support in publications by several modern writers, who contend that society understands and is concerned about changes over three generations. Families have an in-built memory of this length, because grandparents talk to grandchildren. Moreover, we believe that component model processes such as sedimentation and fluviology, and perhaps habitat destruction, can be computed over simulated times of (say) 75 years. One-hour time-steps for moderate problems (40 sub-catchments and conveyances, 15 pollutants) can readily be computed for such durations in an eight-hour working day on inexpensive personal micro-computers. Certainly three generation modeling (3GM) with SWMM is readily available, although we still need quality code for a six-minutely rainfall generator. It is not unusual in North America to find rain records of 50 years and longer, although the available time step is rather coarse. A suitable FORTRAN program for rainfall rates at fine time steps has been developed at Monash University in Australia. It seems likely that 3GM will at least start to make a significant difference to the way developers do business in this arena, and thus to our environments and ecosystems. SOME ECO-ETHICAL CONCERNS In this section we proceed to some new dimensions of eco-sensitive design, not yet associated with continuous modeling, but impossible to reconcile with event methodology. Apart from human population control, or reduction, there can be no activity more urgent to the future of the world than to reverse the loss of natural habitat (Kennedy, 1993). Habitat loss is the result of anthropogenic activity in watersheds, such as urbanization. Lazaro (1990) describes the physiology, anatomy and the relentless morphology of cities, with their consequent degradation of aquatic environments. By urbanization, we mean the concentration of people into urban settlements, and the change in land-use first from indigenous, original forest or prairie, to rural or agricultural, then to urban commercial and industrial. All land-use changes affect the hydrology of an area, but urbanization is by far the most forceful (Leopold, 1968). There is no doubt that the contemporary population intensification into urban areas will continue through the next few generations, and that the associated hydrological problems of aquatic habitat destruction will become increasingly more acute. There also can be no doubt that conventional design-storm, or event-hydrology methodology, focuses on extreme simplification, precisely to reduce the cost of the design phase of landscape interventions in, or changes to, watersheds. That is why simplistic methods have been strongly supported by some pro-development forces, such as land developers, construction industries, government departments of agriculture, and consulting engineers who profit from such design studies and construction. They advocate extremely fast, short-cut and cheap design, design that is not concerned with, but may even denigrate, the study of long-term ecosystem impacts. Design costs, whether cheap or expensive, are eventually passed on to homeowners, of course. Recent consumer surveys, however, evidently suggest that homeowners in North America are prepared to share the higher cost of eco-sensitive design. Long-term continuous water quality modeling on the other hand leads naturally to consideration of impacts on aquatic ecosystems. Necessary information is brought into focus in the foreground. As an example, salmonid (cold water fish) habitat requires seasonal and growth-phase upper and lower limits on flow depth and flow velocity, a continuous low upper limit on turbidity (for sight feeders), a mean summer low upper limit on mean water temperature, very small deposition of suspended material in spawning areas and seasons, and all this in addition to the established limits on transported chemicals. Event hydrology does not provide such information, whereas output from continuous SWMM, for example, begs to be fed into the In-stream Flow Incremental Method (IFIM) for evaluating fish habitat (Navarro et al., 1994). Features of an unpolluted river are: (a) ecosystem diversity, (b) flow and water quality stability, and (c) self-purification. In a few minutes it is possible to collect two dozen visible plant and animal species, and 100 microorganisms. Proponents of ecological sustainability regard nature in this complexity not just as a set of limiting concentrations, but as a better model for the design of housing, townships, neighbourhoods and regional economies. Sustainability depends upon replicating the structure and function of natural systems with their far-reaching inter-connectedness. Orr (1992) suggests a number of concerns for design that is sensitive to sustainable eco-systems:
(The emphases are those of the present writer; not all design precepts were cited). Wet-event-hydrologists (such as HEC-1 users) will raise objections to the above position. And while they would agree that traditional design has led to numerous minor eco-disasters, such as the loss of cold-water fisheries locally, it is true that these have been more the result of ignorance than of deliberate eco-vandalism. Conventional, narrow design does not care about downstream ecosystems; the case was neatly posited by Field and Lager almost two decades ago (1975): ....simply stated, the problem is as follows. When a city takes a bath, what do you do with the dirty water? [This suggests that there is no alternative but to degrade aquatic and riparian ecosystems downstream of urban creeks.] Serious questions should be asked not only about the quality of the designs proposed, but also about the quality of the design study itself. By now we all know that simple engineering solutions have very often led to ecosystem nightmares. Wedell Berry (cited by Orr, 1992) states that a bad solution is bad because it acts destructively upon the larger patterns [of nature] in which it is contained. It acts destructively upon those patterns, most likely, because it is formed in ignorance or disregard of them. On the other hand, a solution is good if it is in harmony with those larger patterns. As I understand it, and further paraphrasing Berry, this means that a good design will have to meet a number of requirements more appropriate to a post-modern society. Among these a number may be mentioned, even though few of them, if activated at all, could be attributed directly to continuous modeling philosophy. Among other requirements, a good design will:
(Not all points have been listed; only the first five would probably be implicated by continuous modeling, assuming that the continuous output would be further processed by aquatic ecologists; the last two distinctly relate to post-modern society, and are included from the larger list out of interest; the emphases are those of the present writer.) The point is that continuous modeling makes it difficult to avoid ecosystem concerns, while the use of event hydrology makes it difficult to consider them. This point has also been developed by Abbott (1993). Opting for event hydrology, then, given the computational environment of late 1994, is akin to a deliberate decision to choose ignorance all-the-way-down-the-road, to invite eco-disaster. In this sense, perhaps wrecklessly, we may say that it has become simple-minded, in the words of the title of this paper. PROCESSES RELEVANT TO ECOSYSTEMS In this section, we mention by way of example, four sets of processes or procedures that are not commonly found in popular event models used in design applications for managing the impacts of urban stormwater. All four sets rank at the top priority for future enhancements, in the writer's judgment, and the discussion helps develop an argument for new code that may be added to existing codes, perhaps through suitable shells. Ecosystem restoration In my judgment, the most important BMPs (how many, one wonders, could simultaneously be best when so many are demonstrably detrimental?) have not yet been coded into the widely-distributed and supported surface water quality models, and hence these BMPs are not appearing in alternative water pollution control strategies. The most important BMPs are those which redirect us, back to the ecosystems which existed before so-called "white-man's" agriculture or urbanization, BMPs which by their design objectives seek to direct runoff back into the ground, to remove pollutants at the source, and add feed-stock for aquatic ecosystems. Deciduous canopy is an important BMP in urban areas, to cool overheated cities. Similarly, infiltration BMPs, particularly those with sand filters, would help reduce the loss and destruction of cold water fish habitats. We have not yet developed design methods for BMPs that add the "contaminants" that are essential for aquatic life, such as flies that have aquatic larval stages. Thermal enrichment Temperature modeling is very inadequate in our existing models. Not even HSPF can model heat accumulation in blacktop paving and roof tiles, nor do any urban runoff models cover the wash-off of thermal energy from paving and roofs. Temperature is the number one determinant of ecosystem types downstream of urbanization. Available models do not deal with solar thermal enrichment, yet all aquatic chemical and biochemical processes are temperature-dependent. Areas such as Detroit are effectively enormous, exposed, black-body solar receptors, and rainfall/runoff is an efficient transporter of heat from hot pavement and roofs. In this discussion, thermal enrichment is taken to mean the increase in thermal energy, carried from the existing or proposed development to receivers, over what it would have been had there been no significant anthropogenic activity, including agriculture and clearance of deciduous canopy. Enrichment should not merely denote potency in terms of existing conditions - sediment constituents, for example, because the sediments have been seriously degraded since the time of the original mixed forest canopy. My group has done some studies on paving, parking lots and on the city of Guelph as a whole, modeling the thermal enrichment due to storm water. Rain cell dynamics In the Great Lakes region we experience summer convective rains approximately once every four days. We do not make enough use of readily available data that deals with the inherent spatial and transient variability of thunderstorms. Yet it is the hot weather and the associated short-duration, high-intensity rain that causes the flooding, thermal enrichment and erosion downstream. Convective summer thunderstorms are dynamic: they age, have preferred trajectories and preferred directions, they are not stationary on the average, and they are multicellular. In fact there is still almost no modeling of thunderstorm dynamics, even though all surface water quality modeling is wrong if the rain is wrong. All of it. Modelers end up calibrating for bad rain data by fiddling with infiltration rates, which is really funny, when you think of it. In North America, weather radar data is readily available. Weather radar units are becoming economical, compared with networks of recording rain gages. Urban runoff in sewers is now being controlled in Japan by small-scale radar. Software exists for converting radar reflectivities to rainfall intensities at a scale of 50 meters and in time steps of 1 to 6 minutes, approaching the resolution and averaging that we need for urban hydrology. My group is using ArcInfo to convert radar reflectivities to rainfall intensities directly, and to model storm cell dynamics, as a part of the spatial data management for the SWMM program. Inherent fuzziness Fuzzy logic is being quite widely applied in hydrology these days. We certainly need to present computed urban pollutants as fuzzy output. Since the fundamental equations in surface water quality models are not analytical, we should not strive as we do, to present computed results as a smooth, single-valued relationship, especially for pollutant build-up and wash off. Modellers are not relaying the inherent fuzziness of the processes in their results. Fuzzy logic will require a re-examination of fundamental principles. Simple elementary measures, like thermal enrichment, will no doubt benefit from the approach. DECISION SUPPORT SYSTEMS Most, if not all, of the above processes are relevant to ecosystem concerns. As argued before, they also seem to imply a need for continuous modeling, which in turn requires better shells and decision support code. What are the requirements of decision support systems (DSS) that favor such continuous modeling? For a start, such DSS would need, at a minimum, to provide:
Such DSSs must become a normal part of the modeling procedures, so that uncertainty due to poorly defined processes and information are duly and responsibly revealed to the end-users. Beyond the topic of this paper, calibration and error analysis are important attributes of such a shell. In passing, it should be remembered that calibration requires only short term records of input and objective functions, and thus a short-term monitoring effort; design however requires long-term input functions that may in fact be reasonably transposed from elsewhere. There seems to be no doubt then that our deterministic surface water quality models could benefit from more statistical tools. For design purposes, very-long-term input functions may be synthesized from shorter records using stochastic rainfall generators, for example. Considering the inherent variability of rainfall, more statistical manipulations are necessary to build 75 years of data, and to present the results of 75 years of flow and many pollutants at many points. End-users of 3G modeling could not comprehend such results without a fair amount of statistical manipulation. The melding of statistics with deterministic models will have another advantage: it will facilitate the removal of the artificial distinction between data gatherers (field personnel) on the one hand and data consumers (modellers) on the other (or between monitoring people and analytical laboratory people on the one hand, and the modellers on the other). Provided such models incorporate suitable sensitivity analyses, very long term complex deterministic models with comprehensive statistical tools provide useful management tools for data collection programs. They are the best means for filling in missing data. Ranking the sensitive parameters helps rank priority for selecting chemicals, sampling frequencies and locations, and accuracies of determination. Widely distributed, integrated databases are about to become useful. Our surface water quality models should be tied in with suitable data and user networks, such as the Great Lakes Information Network (GLIN) and the Consortium for International Earth Science Information Network (CIESIN). We also need to encourage widespread use of our models, in education, decision making, as well as engineering and research. For instance, we should write code that helps make our models available in different languages. Not all North American modellers speak English. Apart from minority groups, there is a great need to translate HSPF and SWMM into Spanish and French, because about 60 million people who drink the water that about 300 million of us are all polluting today, speak those two languages, rather than English, and could benefit from using good SWMM programs just as much as we do. PCSWMM PCSWMM for Windows is a shell - code that is written around existing code, usually to provide better human interaction. In this case the existing code is SWMM4.3 - a collection of programs comprising hundreds of files, hundreds of routines, and scores of thousands of lines of FORTRAN source, object and executable code; more specifically SWMM4.3 refers to the U.S.EPA official issue, compiled and linked using Lahey or Ryan-McFarland products, and downloaded during or after June, 1994. It makes little sense to test a program for SCEA. Written for a drag-and-drop Windows environment, the shell PCSWMM for Windows (James, 1994) aims especially at sensitivity, calibration and error-analysis for design applications using large-scale, long-term, continuous modeling at high spatial and temporal resolution. SCEA is rendered semi-automatic for version 4.3 of the U.S.EPA's Storm Water Management Model. Fuzzy logic is used to manage the complexity, and to interpret the ranked parameter sensitivity in various RUNOFF state-variable spaces. So far as is known, this is the only shell to focus especially on SCEA for design applications using large-scale (say 1000 and more elements) and long-term (say 75 years) continuous water quality modeling at high resolutions (elements of say 100 ft length, 1 acre extent and integration intervals of 60-seconds). Contrariwise, several papers in the peer-reviewed literature, claim that these specifications are impossible to achieve. Of course the shell also includes the usual graphics user interfaces that are windows, interactive, menu-driven and pointing-device oriented (GUI-WIMP; note that these gooey wimps are not what they sound like - they were created to be robust). Pre- and post-processors for data input and output interpretation are included. But most significantly, SCEA is rendered semi-automatic for approximately 102 different hydrologic parameters and a total number of parameters in the order of 104, depending on the processes active, and the spatial resolution. James and James (1993) describe the genesis of PCSWMM. Various objective functions, performance evaluation functions and error functions are used in PCSWMM for Windows. Since the error analysis is linear, with first and higher order sensitivity gradients, the procedures depend on user-intervention at various levels. Intensive, interactive user-dialog demands that the terms used are carefully defined, especially if they differ in important ways from much of the literature, which has heretofore been understandably loose. That is why, for clarity, the terms used have once again been defined throughout this paper. CONCLUDING DISCUSSION Principal among the purposes of the modeling effort, is to design an optimum array of BMPs - landscape interventions that temporally divert, store or treat urban stormwater runoff to remove pollutants, reduce flooding or provide other amenities. The recommended way to design them in the future is by using continuous 3GM. But....despite almost 30 years' of arguments in favour of continuous modeling, the method has not yet become routine in design offices. Amazingly, most stormwater and flood design manuals that have been published over the past year or two, still do not recommend continuous over event modelling. Thus recourse to existing engineering design manuals does not help. In any case, adoption of a long-term time-series should not be the insurmountable challenge that it seems to have been: because of:
- there can no longer be any case for event-hydrology methods. The old argument that continuous modeling is not computable is no longer true: 75 years at one-hour time-steps for 40 subcatchments, 40 conveyances, and 15 pollutants can be handled by a 486DX2 easily within an 8-hour day. Thus, instead of using the one-in-fifty-year-storm, simply use one fifty-year-storm. For continuous, fuzzy modeling, appropriate decision support systems are recommended: PCSWMM for Windows is such a decision support system - code that manages the simulation system, as opposed to the internal process codes (collectively called the engine). This methodology also involves:
In the final analysis, the method must be computable - a simulation that can be performed in a working day (eight hours) using, and providing computed output that can be contained on, typical engineering office workstations, comprising (say) 500 megabyte (Mb) hard disk and a 486-66 motherboard with 16 Mb dynamic memory (DRAM). Finally, in keeping with our philosophy of continuous modeling for ecosystem concerns, we have to redefine potency factors and enrichment, so that we can relate our results back to pre-forest-clearance conditions, since impacts of the change from indigenous forest to agriculture were often as bad as the change from agriculture to urbanization. In closing, we need code for cost-effective, comparative evaluation of eco-restorative BMPs, and the decision support systems that encourage their evaluation, preferably their adoption. And remember: never use event hydrology for design. ACKNOWLEDGMENTS The shell PCSWMM for Windows described in this paper was written by Rob James. Former graduate students of W. James who worked on similar codes are: Karen Dennison, and Al Dunn (a sensitivity analysis framework for CREAMS and SWMM respectively); Mark Robinson (PCSWMM3 - a user interface); Taymour El-Hossieny (who worked on intelligent database interfaces); and Tony Kuch (PCTOOLS - a sensitivity analysis shell). Small parts of this paper have been taken and modified (here and there) from a paper given this summer at the ASCE Hydraulics 94 conference in Buffalo, NY. Also much of this presentation was given orally (but more graphically) at a workshop sponsored by the US EPA at Heidelberg College, in Tiffin, Ohio, in September 1993. REFERENCES Abbot, M.B. 1991. Hydroinformatics: information technology and the aquatic environment. Avebury Technical. ISBN 1-85628-832-3. 157pp. Field, R. and Lager, J.A. 1975. Urban runoff pollution control - state of the art. J.Environmental Eng'rg, ASCE EE1 pp.107-125. Huber, W.C. and Dickinson, R.E. 1988. Storm water management model user's manual, ver 4. EPA/600/3-88/001a. (NTIS PB88-236641/AS) U.S.EPA, Athens, Ga. James, W., in press. Channel and Habitat Change Downstream of Urbanisation. Eng'rg Fndn Conf. on Urban Runoff and Receiving Streams. Crested Butte, Colorado. Aug. 4-9, 1991. -----, 1994. Rules for responsible modeling. Draft monograph and documentation for PCSWMM for Windows. Published by CHI Guelph, ON Canada. ca. 150 pp. -----, 1994. A semi-heuristic, fuzzy, sensitivity, parameter-optimization and error-analysis shell for very long-term, large-scale water quality models. NABEC-94 paper 94113. U of Guelph. July 25-28, 1994. -----. 1993. Need for 3-generation stormwater quality impacts models. USEPA Invited Conference on Model-building for Great Lakes Water Quality problems of the 2000's. Heidelberg College, Tiffin, OH. Aug.4-6. copy available from author. -----, and James, W. Robert C. In press. PCSWMM4 - a public domain, graphics-oriented shell for easing the design-use of SWMM4. Runoff Quantity and Quality Model Conference. ASCE. Reno NV. Nov. 8-9 1993. 6 pp. -----, and Robinson, M.A. 1986a. Continuous Deterministic Urban Runoff Modelling. Int. Symp. in Urban Drainage Modelling, Dubrovnik, Yugoslavia, published by Pergamon Press, London, pp.347-378, Apr. 1986. -----, and -------. 1986b. Continuous Modelling of Downstream River Erosion caused by 40 Years of Urbanization. Water Forum '86, ASCE, Long Beach, pp.1710-1717, Aug.4-6, 1986. -----, and Shivalingaiah, Boregowda. 1986. Continuous Mass Balance of Pollutant Build-up Processes. Proc. NATO Advanced Research Workshop on Urban Runoff Pollution, Montpellier, France. Aug.25-30, 1985, published by Springer-Verlag in Urban Runoff Pollution (Series G. Vol.10), pp.243-271,1986. Kennedy, P. 1993 Preparing for the 21st century. Random Hse, ISBN 0-394-58443-0, 445pp. Klemes, V. 1987. Empirical and causal models in hydrologic reliability analysis. Procs NATO Institute on Engineering reliability and risk in water resources. Tucson, Ariz. May 19-Jun 2, 1985. ISBN 90-247-3492-4. pp.391-403. Kuch, A.J. and James, W. 1993. A sensitivity analysis tool for three generation continuous urban storm water quality management modelling. Runoff Quantity and Quality Model Group Conf, Sponsored by ASCE. Reno NV, Nov.8-9, 1993, in press. Lazaro, T.R. 1990. Urban hydrology - rev ed. Technomic. ISBN 87762-547-6. 250pp. Leopold, L.B. 1968. Hydrology for urban land planning - a guidebook on the hydrologic effects of land-use. USGS. Circ. No. 554. 18pp. Navarro, J.E., McCauley, D.J. and Blystra, A.R. 1994. Instream flow incremental methodology (IFIM) for modelling fish habitat. In: Current practices in modelling the management of stormwater impacts. Lewis Pub. ISBN 1-56670-052-3 pp.1-22. Orr, D.W. 1992. Ecological literacy. State U. of NY. ISBN 0-7914-0874-4 224pp. Footnotes: 1 Professor of Environmental and Water Resources Engineering, University of Guelph, Guelph, ON, Canada N1G 2W1, fax 519 767 2770, email: wjames@uoguelph.ca 2 This part of the paper has been abstracted from the user documentation for PCSWMM for Windows - see James, 1994. |