Cover Image
close this bookConducting Environmental Impact Assessment in Developing Countries (UNU, 1999, 375 pages)
close this folder5. EIA tools
close this folder5.1 Impact prediction
View the document(introduction...)
View the document5.1.1 Application of methods to different levels of prediction
Open this folder and view contents5.1.2 Informal modelling
View the document5.1.3 Physical models
View the document5.1.4 Mathematical models
View the document5.1.5 Modelling procedure
View the document5.1.6 Sensitivity analysis
View the document5.1.7 Probabilistic modelling
View the document5.1.8 Points to be considered when selecting a prediction model
View the document5.1.9 Difficulties in prediction
Open this folder and view contents5.1.10 Auditing of EIAs
View the document5.1.11 Precision in prediction and decision resolution

5.1.11 Precision in prediction and decision resolution

Much recent research has focused on auditing the accuracy of predictions, although the term "predictive techniques audit'' is not generally used. These studies tend to focus on the scientific and technical aspects. The main issue identified by most research workers is the degree of precision possible in assessing and predicting environmental parameters, most researchers concluding that improving prediction accuracy is at the heart of successful EIA.

Several research workers have concluded that predictions need to be written in a manner that facilitates auditing, and should therefore be presented as falsifiable hypotheses. In addition, they should present information on: (i) the variable subject to an impact; (ii) the magnitude, extent, and time scale of the impact; (iii) the probability of its occurring and its significance; and (iv) the confidence to be placed in the prediction.

There is also a general belief (and India is no exception) that prediction and assessment in EIA is acceptable or improved only when it is presented in a quantitative manner. This has increased the application of mathematical models (notably air quality or dispersion models, followed by water quality models with an emerging interest in consequence modelling for risk assessment) as well as increasing the utilization of assessment schemes on the basis of scoring (or rating) and ranking.

Application of mathematical models sometimes requires non-routine data. If it is not generally available, then assumptions have to be made. Typical assumptions in the case of air dispersion models refer to atmospheric stabilities for which data is not available on a routine basis. Emission data supplied by the project proponent is treated as accurate.

The models themselves have problems of incompleteness and it is not correct to attach a significant amount of certainty to the model's estimate. For example, well-constructed dispersion models with good a database are noted to have a variance of the order of 60%. Most of the water quality models applied consider the presence of complete mixing of pollutants near the discharge point in the x, y, and z directions. The consequence models generally used treat a chemical plume as having a density comparable to air with no reactions and heat exchange, which is not the case in reality.

The common mathematical models used in EIA enable calculations up to ambient concentrations and not directly to the estimation of impacts, for which one needs to take the opinion of experts, use guidelines (standards), or apply valuation models. Further, models are not available for a variety of other issues like socio-economic and health impacts. Despite these limitations, it is noticed that the EIA documents sometimes include a dominating section on the use of a mathematical model and the review members spend more time in checking the equations, coefficients, etc., of the model applied. Intensive debates are not uncommon, for example whether the 8 hourly sulphur dioxide concentration at 3 km downwind (where there is a sanctuary) is 50 mg/m3 or 70 mg/m3 (60 being the standard value) when it is possible that the model estimate can be anywhere between 5 and 100 mg/m3! Mature application of mathematical models is rarely seen.

It should be remembered, however, that precision is preferable, but striving for precision could result in a prediction defined too narrowly to be of practical use. Also, it is often unreasonable to expect a high degree of accuracy from predictions. Added to this doubt is the question of how accurate the prediction needs to be on an operational level: if underlying processes are sufficiently understood, then the appropriate management response may still be evoked by an impact even if it is not precisely predicted.

It is suggested that trying to make predictions rigorously testable in a context where decision-making is done at a practical level is, in most cases, not appropriate. Obviously, precision is preferable when it is possible and appropriate. However, the philosophy of "better qualitative and useful than quantitative and wrong'' may be more appropriate in most cases in developing contries.

What this statement means in terms of auditing is that the totally pragmatic evaluation of a prediction would become "Did it result in appropriate management action being taken?'' compared with the totally scientific evaluation of "Did it correctly describe the type, magnitude, extent, and duration of the impact?'' The focus should be to try and incorporate aspects of both these approaches in an auditing programme. In other words, we should be concerned with whether the EIA procedural framework results in good management, as well as showing how well the individual steps in the process perform. At a practical level, individual steps can be no more effective than the whole of the process. What point is there in correctly predicting 99% of impacts if none of them gets managed properly?