A cost-benefit approach to tracer testing Categorisation of Systems, Purposes, Processes & Tracers Examples of specific questions Critical protocols, QA & QC Cost / Benefit

Systems Categorisation Waste water treatment / distribution systems Drinking water treatment / distribution systems Gas / Steam distribution systems Non-environmental or industrial systems: Surface water Groundwater Hydrologic Systems: Associated Environmental & Geologic Systems: Contaminated land Oil field applications Gas phase tracing (atmospheric / soil)

Categorisation of Purpose Typical Tracer Test Objectives Groundwater connections Seepage velocities Hydraulic conductivity Kinematic porosity Groundwater flow mechanism Contaminant behaviour Establish flow horizons Characterise recharge Hierarchy of Tracer Test analysis Connectivity Travel time Statistical (mean, variance, moments) Transfer function (coefficients of transfer model) Analytical solutions (few parameters) Numerical solutions (few to many parameters) Constrain predictions Typical groundwater tracer test objectives are listed on the left. On the right hand side is reproduced a table from the BGS/EA collaborative tracer test manual (Ward et al. 1998), which is the standard UK reference work. The table lists seven levels of hierarchy of analysis, culminating in 'constraining predictions'. I argue that as one requires some form of conceptualisation throughout, that constraint prediction (hypothesis testing) should form a part of analysis at every stage. This point is also discussed further below. Ward et al. 1998.

Process Categorisation Mixing Dilution Advection / Dispersion Diffusion Chemical reaction Ion exclusion Adsorption Absorption Leakage Surface water / groundwater interaction Degradation Decay Note scale dependency of some processes eg: Dispersion (spatial scale) Diffusion (temporal scale) Processes are presented ahead of particular types of tracer as tracers of various forms are typically used to evaluate the relative extent and importance of these processes. WRT scale – it is important to be aware that the same tracer may respond differently at different scales, and thereby elucidate different processes. EG diffusion is a time-dependent phenomenon, and depending on the temporal scale of transport (the experiment) may provide information about fracture flow only (short times), fracture and matrix porosity (medium times) or bulk porosity (long times).

Tracer Categorisation Ambient tracers Geochemistry: Physico – Chemical parameters Major ion chemistry Trace element chemistry Anthropochemistry: Contaminant spills / leaks Atmospheric gases Artificial tracers Temperature Micro-organisms Particles Dyes Ions Gases Fluorocarbons Stable isotopes Radioisotopes Ambient tracers – those that are present in the environment already. Artificial tracers – those we might introduce to stress a system or produce a specific effect. Bottom graphic – increase in atmospheric concentration of CFCs and SF6 throughout 20th Century.

Tracer Categorisation Ambient tracers Geochemistry: Physico – Chemical parameters Major ion chemistry Trace element chemistry Anthropochemistry: Contaminant spills / leaks Atmospheric gases Artificial tracers Temperature Micro-organisms Particles Dyes Ions Gases Fluorocarbons Stable isotopes Radioisotopes LOTS OF CROSSOVER! Note that many of the tracers will elucidate common controlling processes. From a cost-benefit point of view, as many groundwater investigations will as a matter of course acquire, for example, major ion or trace element chemistry, information regarding processes of interest may already be held by the investigative team.

Examples of specific questions How do I discriminate between processes responsible for residual contaminant release? How long will contamination persist above a given concentration? What is the proportion of fracture / matrix flow? Is this SPZ realistic? How does the permeability vary spatially? Are these two points connected? What is the groundwater velocity? What should be the immediate response to a groundwater contamination incident ?

Know your question How do I discriminate between processes responsible for residual contaminant release? How long will contamination persist above a given concentration? What is the proportion of fracture / matrix flow? Is this SPZ realistic? How does the permeability vary spatially? Are these two points connected? What is the groundwater velocity? What should my immediate response to a groundwater contamination incident be? Varying levels of conceptualisation Questions should be appropriate for one's existing level of conceptual understanding Tracer TEST – test hypotheses (one is usually required to estimate parameters in order to conduct a test successfully – iterative processes) From the list of specific questions it is obvious that some are considerably more complex than others. Thus, there is a level of system conceptualisation required prior to asking any particular question. From an artificial tracer test point of view, even the most basic 'point – to – point' experiment to demonstrate connection requires one to consider how much dilution might be expected (in order to quantify amount of tracer to inject) and to estimate travel times – in order to be sampling at the right time, as well as assuming that a connection exists – ie flowpath assumptions. Thus, 'prediction constraint' or hypothesis testing should be present at whatever level of tracer test one employs. The graphs at bottom show changes in groundwater chemistry with distance (0 – 10,000m) from a sining river. These illustrate particular processes of dissolution (increases in HCO3 & EC) and ion exchange (increase in the Mg/Ca ratio). Simple chemistry.

Boundary conditions An 'artificial' tracer test should be very well defined Well-constrained input functions Can be conducted in conjunction with pumping tests or other controls on the flow field Generally well-understood properties of tracers (e.g. diffusion co-efficients) Preparatory column experiments can be used to establish aquifer-specific sorption isotherms if necessary Wide variety of contaminant analogues – tailored to your requirements All facilitated by adequate preparation WRT Artificial tracer tests, one of the principal advantages is the tight control available over the boundary conditions of the experiment, particularly the input function – in terms of concentrations/amounts, shape of input pulse, length of input time. On occasion the flow field can also be controlled to a greater or lesser degree (eg from laboratory column experiments to pumping wells) and this enables more precise analysis as well as potentially examining different responses under changing flow conditions. The diagram at bottom is a simple pipe network model of a karst aquifer showing principal flows in a particular system. To be able to constrain one's boundary conditions, one needs to have excellent protocols for QA/QC – see next slide Basic karst / fracture network model where: Qout = Q1 + Q2 + Qaq1 + Qaq2 Qlost = 0.5 Qin

Protocols, Quality Assurance & Quality Control Strict observance of protocols is required for successful tracer testing Quality Assurance: Sampling and analytical protocols Quality Control: Methods for indication / quantification of error (duplicates, trip blanks, field blanks, decontamination test blanks) Quantification of error (QC) necessitates & informs protocols (QA) (iterative process) Sources of error may be several and non-obvious Familiarity with analytical methods and laboratory processes is important. Good QA/QC provides the necessary assurance to both regulators and clients, and practitioners with robust results Protocols are critical for successful tracer tests, not only in order to provide the analyst/modeller with closely constrained boundaries and thus a better handle on model uncertainty, but critically to provide the regulator and/or client with assurance that the experiment will be conducted with all due regard for the safety of the hydrologic system in question.

Summary Define your objective within the context of your conceptual understanding Be aware of the processes likely to contribute to system behaviour Recognise the possibility of already having / intending to collect information that reflects the processes you are interested in Artificial tracer test: tightly constrained boundary conditions Be aware of critical protocols and QA / QC requirements More on Cost / Benefit soon... Speak with Water Tracing Services UK! Thanks for listening, please have a look at our website for more detail: WATERTRACING.COM

Breakthrough curves testing a hypothesised dipole flow field between a sinking stream and a pumping well. The linear part of the breakthrough curve 'tail' reflects the dominant process controlling contaminant release from the aquifer. Although the dipole model fitted the tailing precisely at the pumping well, it also fitted precisely the curve from a well elsewhere within the flow field, thereby indicating that the dipole odel (which should only have fitted at the pumping well) was not a good model of the system. Dual porosity behaviour and regional groundwater gradients were considered to have greater influence on the ongoing release of proxy contaminant (ie tracer).