Groundwater HIA post edit - FreshwaterLife

abstraction, all assumptions must be recognised and taken into account. It is often
useful to undertake some form of sensitivity analysis in order to understand the effects
of ranges in parameter values on derived quantities (see Box 2.1).
Box 2.1: Sensitivity analysis
Example of simple sensitivity analysis to illustrate the effects of ranges in parameter values on
derived quantities, using the Theis equation for unsteady-state flow in confined aquifers
(Kruseman and de Ridder 1990):
s = (Q/4πT).W(u)
where u = r 2 S/4Tt, and W(u) is a function of u (commonly known as the well function), with s
being the drawdown at a radius r from the pumping well at time t, in an aquifer of
transmissivity T and storativity S, and abstraction taking place at a rate Q. Suppose that the
equation is being used to predict the drawdown at a sensitive wetland, using aquifer
parameters estimated from previous tests. The quantities Q, r and t are known with
reasonable accuracy, and we are using estimated values of T and S to predict s. Let’s say
Q = 1,000 m 3 /d, r = 500 m, t = 100 days, T = 400 m 2 /d and S = 1x10 -4 . This gives a prediction
for drawdown (s) of 1.63 m. The measured range for T might be 200 to 600 m 2 /d (even
ignoring the fact that the true range may be much greater), and let’s say the range for S is
from 5x10 -5 to 5x10 -4 . Keeping S at the original value, using the extremes for T gives a range
for s of 1.14 to 2.98 m. Keeping T at the original value, using the extremes for S gives a
range for s of 1.31 to 1.77 m. However, combining the uncertainties (varying both T and S in
the combinations that give the greatest extremes) results in a possible range for s of 0.93 to
3.26 m. Which drawdown turns out to be the ‘true’ value could have dramatic implications for
the wetland.
Some types of uncertainty are easier to reduce than others. For example, drilling more
observation boreholes for a pumping test, or conducting tests in several boreholes, will
help to reduce the data and sample uncertainty; using a radial flow model with layers
(as opposed to a simple analytical equation) to analyse the results will reduce the
model uncertainty. However, reducing knowledge uncertainty may require extended
scientific study; and environmental or natural uncertainty is impossible to reduce, and
must just be recognised.
2.3 Conceptual modelling
Conceptual modelling is at the heart of both CAMS and the Water Framework Directive
(see Appendix 1), and its importance to HIA cannot be overemphasised. In the water
resources context, a conceptual model can be defined as a synthesis of the current
understanding of how the real system behaves, based on both qualitative and
quantitative analysis of the field data. Some people take the view that conceptual
models are based upon a purely qualitative understanding, with quantitative
assessment only coming in during subsequent analytical or numerical modelling.
However, in this report, the term conceptual modelling definitely includes quantitative
analysis.
A real hydrogeological system is so complex that it will never be possible to study
everything in detail; a conceptual model is therefore bound to be a simplification of
reality. The important question is to determine what needs to be included in the study
and what can be safely ignored. In other words, what observed behaviour do we want
the conceptual model to get right, and what don’t we mind the model getting wrong?
For example, if we are investigating the mechanisms that operate during periods of low
flow in a Chalk stream, we may not mind being wrong about the mechanisms that
operate during groundwater flooding events (Environment Agency 2002a). Or, when
developing a regional groundwater resources model of a coastal aquifer we may
choose to ignore the difference in density between fresh and saline water in order to
6 Science Report – Hydrogeological impact appraisal for groundwater abstractions