USE AND MISUSE OF COMPLEX MODELS: EXAMPLES FROM WATER DEMAND MANAGEMENT1

ABSTRACT: The power of computers has increased in recent decades, and one might expect improved management to result because decisions can be made with understanding available only via models. However, there is potential for quite the opposite: poor decisions due to unrealistic model output generated by users without access to appropriate training in the use of models. We discuss and, by reference to water demand models (IWR-MAIN, MWD-MAIN), illustrate three areas in which unintended errors of judgment by untrained personnel may cause difficulty:

• * Attributes of management models; if output from any type of model has no measure of confidence, then results may be over- or undervalued
• * Input data; with complex models, problems here typically will be difficult to detect.
• * Calibration and history-matching (verification); if these steps or data are combined, then users should be less trustful of model output than otherwise.

Because all models have weaknesses and because there always is uncertainty about output from any model, we end with suggestions for coping with complex models. Monitoring programs play a central role in such efforts because they can identify discrepancies between model predictions and actual events and because they can ensure time is available to develop solutions for problems unanticipated in the modeling effort.     

Membrane-based wet electrostatic precipitation

Emissions of fine particulate matter, PM2.5, in both primary and secondary form, are difficult to capture in typical dry electrostatic precipitators (ESPs). Wet (or water-based) ESPs are well suited for collection of acid aerosols and fine particulates because of greater corona power and virtually no re-entrainment. However, field disruptions because of spraying (misting) of water, formation of dry spots (channeling), and collector surface corrosion limit the applicability of current wet ESPs in the control of secondary PM2.5. Researchers at Ohio University have patented novel membrane collection surfaces to address these problems. Water-based cleaning in membrane collectors made of corrosion-resistant fibers is facilitated by capillary action between the fibers, maintaining an even distribution of water. This paper presents collection efficiency results of lab-scale and pilot-scale testing at FirstEnergy's Bruce Mansfield Plant for the membrane-based wet ESP. The data indicate that a membrane wet ESP was more effective at collecting fine particulates, acid aerosols, and oxidized mercury than the metal-plate wet ESP, even with approximately 15% less collecting area.     

The Applicability of National Critical Loads Data in Assessing Designated Sites

Critical loads have been successfully used within Europe in the development of effects-based policies for pollution abatement, including the Second Sulphur Protocol and the Protocol to abate acidification, eutrophication and ground-level ozone (CLRTAP, 1979). This success has encouraged the UK Environment Agency and Conservation Agencies to use the national critical load maps as a screening tool in assessing the threats from acidification and eutrophication to designated (Natura 2000) sites. The UK maps of critical loads are based on national-scale data sets appropriate for national-scale assessments, and were never intended for use at the site-specific level. Site-based assessments are often targeted at Special Areas of Conservation, a sub-set of the UK Natura 2000 sites. The spatial data available includes the boundaries of the sites but not the location of the designated features. Ancillary data is variable from one site to another; habitat types may be described in detail with cross-reference to classes of the National Vegetation Classification (NVC: Rodwell, 1991 et seq), but information available on soils and geology is generalised and has not been related to the habitats or species being protected. Hence it can be difficult to relate the individual sites to the national maps, even where appropriate to do so. This paper examines the underlying uncertainties in the national critical load maps showing how the maps could give misleading results if used for site-specific assessments. It also includes advice on how to determine when the national data may be appropriate as a policy-tool at the site-level.     

Using Multiple Chemical Criteria for Critical Loads of Acidity in Maritime Regions

Critical loads have become a well-establishedpart of the work programme of the UnitedNations Economic Commission for Europe (UNECE) Conventionon Long-Range Transboundary Air Pollution (CLRTAP). Thelinking of ecosystem response to deposition level is thecentral principal of the critical loads approach. Foreach ecosystem, a biological indicator is chosen, asuitable chemical criterion selected and a criticalchemical limit assigned. The Bc:Al ratio is the mostwidely used chemical criterion for setting criticalloads. However, critical loads based on this criterionare very sensitive to marine deposition. In regions whichreceive high depositions of marine-derived base cations,such as the Republic of Ireland (ROI) and the UnitedKingdom (UK), critical loads based on the Bc:Al criterionwill inevitably be high. Therefore, it is proposed thatcritical loads are estimated using multiple chemicalcriteria with appropriate critical limits to protect thechosen biological indicators. The range of publishedchemical criteria have been applied to the ROI and theUK. The chemical criterion corresponding to the mostsensitive critical load have been mapped and thecontribution of each to the final maps investigated. Thesimulations indicate that the most sensitive criteria forsetting critical loads are based on specifying critical Hor Al concentrations. However, the choice of critical limits andmodel parameters will ultimately effect the criticalloads. Therefore, it is important that appropriate criticallimits are chosen to protect the biological indicator andreceptor ecosystem from long-term damage.     

The Role of Mineral Weathering Rate Determinations in Generating Uncertainties in the Calculation of Critical Loads of Acidity and their Exceedance

Current legislation within Europe aimed at limitingecosystem damage resulting from inputs of atmosphericpollution is based on the critical load concept. Mineralweathering rates are central to the calculation ofcritical loads (acceptable levels) of acid deposition.The authors have undertaken a number of studies whichillustrate the complications and limitations inherent inpredicting mineral weathering rates and the implicationswhich these have for critical loads calculations andmapping. Calculated weathering rates and critical loadsfor two acid-sensitive parent materials (greywackes andgranites) are presented and are used to illustrate theimpact that uncertainty can have on critical loadexceedances. The results have obvious implications forportraying the uncertainties of critical loads to policy makers.     

Influence of soil hydrological pathways on stream aluminium chemistry at Llyn Brianne, mid-wales

Storm runoff in afforested catchments at Llyn Brianne is acidic and Al-bearing. At baseflows, stream water is well-buffered with low Al levels. This paper presents the results of a study into how hydrological pathways account for these variations in stream-water chemistry. The investigation was carried out in the LI1 catchment; a 0.4-ha subcatchment covered by stagnohumic gley soils was monitored between October 1988 and September 1989. An instrumented hill-slope was established to identify the hydrological pathways that control the hydrochemistry of storm runoff draining from the subcatchment. Perched watertables developed in the surface horizons of the soil during storm episodes and produced lateral flow above the impeding subsoil. This near-surface flow path was responsible for generating acid, Al-rich storm runoff. Some water drained vertically through the soil profile into the underlying slope drift; seepage from groundwater in the drift sustained baseflows. Buffering reactions in the groundwater zone reduced the acidity and Al levels of baseflows. These hydrochemical characteristics are likely to be representative of other areas of stagnohumic gley soils, which cover 19% of the LI1 catchment: these soils may therefore provide a substantial source of acid, Al-bearing storm runoff in LI1 and similar afforested catchments.     

Rapid Screening and Identification of Airborne Carcinogens of Greatest Concern in California

This paper describes a method for establishing a priority list of airborne carcinogens within a state Jurisdiction. In this case it was necessary to identify, from among hundreds of potential candidates, the five to ten materials of greatest potential concern in California as airborne carcinogens.

Because no previous inventory of carcinogens in California existed, published lists, rankings, and assessments of national scope were used to identify candidates. By systematic manipulation and comparison of these data sources, 47 materials of some notoriety were chosen for closer scrutiny. This selection was pared to 22 candidates largely by eliminating those which had very little production and use in California. (Substances primarily used as pesticides were excluded from the scope of this study.) The remaining candidates were then ranked by additive . and multiplicative algorithms and by a panel of experts. The results of these rankings were combined to produce a single selection of 11 priority candidates. In alphabetical order, they are arsenic, asbestos, benzene, cadmium, carbon tetrachloride, chloroform, ethylene dibromide, ethylene dichloride, N-nitrosoamines, perchloroethylene, and polycyclic aromatic hydrocarbons. In continuing studies, a baseline emissions inventory is being prepared, and a source testing program is being designed.     

Distinguishing between Lévy walks and strong alternative models

Lévy walks are a widely used but contentious model of animal movement patterns. They are contentious because they have been wrongly ascribed to some animal species through use of incorrect statistical methods and because they have not been adequately compared against strong alternative models, such as composite correlated random walks. This lack of comparison has been partly because the strong alternative models do not have simple likelihood functions. Here I show that power-spectra and the distribution of the first significant digits (the leading non-zero digits) of the step lengths can distinguish between Lévy walks and composite correlated random walks. Using these diagnostic tools, I bolster previous claims that honey bees use a movement strategy that can be approximated by Lévy walks when searching for their hive or for a food source.     

The Impact of Conifer Harvesting on Stream Water Quality: A Case Study in Mid-Wales

A 12-year record of water quality data for runoff from a spruce forested hillslope with podzolic soils shows the impacts of conifer harvesting and replanting in relation to nitrate generation and its influence on surface water acidification. With felling, nitrate increases from a background of 18 Eq/l to about 50 Eq/l after 1 to 2 years and then declines to background levels over the next 1 to 2 years and to lower concentrations thereafter. This change is mirrored by an acidification process as manifest by a change in Gran alkalinity, acid neutralization capacity (ANC) and aluminium concentrations as well as pH. For example, Gran alkalinity and ANC, which start at negative concentrations prior to felling (about –20 and –50 Eq/l, respectively), become more negative (–30 and –100 Eq/l, respectively) at high nitrate concentrations. Correspondingly, pH decreases from about 4.7 to 4.5 and aluminium concentrations increase from about 14 to 16 M. Subsequently, the acidification is reversed as nitrate concentrations decline and after five years post-felling the system has higher pH, Gran alkalinity and ANC together with lower aluminium concentrations than even before the felling took place (the post-felling values are about 4.9, -15 Eq/l, –20 Eq/l and 7 M/l, respectively).Other determinands show clear changes over time. For example, there is a marked increase in sodium and chloride prior to and around the time of felling (200 to 300 and 230 to 400 Eq/l, respectively), with a subsequent decline in concentration to pre-felling and to lower values of around 160 and 170 Eq/l, respectively, thereafter. This change is probably associated with abnormally high inputs of sea-salts from the atmosphere during the first quarter of the year of felling, and dilution thereafter, rather than a direct consequence of the felling activity itself: this change in sea salt loading has had an impact on stream acidity. Dissolved organic carbon and iron also change with concentrations increasing over time (60 to 200 and 1.0 to 1.5 M/l, respectively) and this mirrors a general pattern observed across the Plynlimon catchments irrespective of whether or not there has been felling activity.The implications of the findings are discussed in relations to environmental management and hydrochemical processes.