Helen Rutter, Jane Alexander (Aqualinc Research Ltd) and Catherine Moore (GNS)
The first barrier for preventing waterborne illness is to protect the drinking water sources from contamination. To do so requires delineation of source water risk management areas (SWRMAs), within which activities are controlled. The SWRMAs reflect the risk of source water contamination based on the time for contaminants to travel to the abstraction point, and also the time needed for some contaminants (e.g., bacteria) to attenuate or become inactive. The National Environmental Standards for Drinking Water (2007) (NES-DW) were intended to support source water protection by providing national direction on how to manage activities that could impact the quality of treated drinking water. In the recent review of the NES-DW there was found to be significant variation in the methods used to define those zones.
A default methodology for delineating ‘source water risk management areas’ (SWRMAs) has been proposed to identify areas where activities have a higher likelihood of affecting source water (Lough et al., 2018). However, default zones may be too conservative in some cases, and not sufficiently conservative in others. There is a fine balance with conservatism. If the SWRMA is over-conservative, it could limit or restrict land use activities on highly productive land or lead to unnecessary barriers to the consenting and establishment of safe new community water sources. If not sufficiently conservative, then activities could be allowed within the SWRMA that could cause contamination of the source, or new sources could be allowed that are at risk. Simple approaches to defining SWRMAs must use a higher degree of conservatism than more robust methods. However, where risks are high, and/or there are large populations supplied by a well, then modelling-based methodologies have merit.
While SWRMA guidance (Moreau et al., 2014a) has referred to numerical modelling, there has been a lack of specific guidance on what makes a good model for SWRMA purposes. In particular, the prediction context is important: this relates to the level of risk being addressed (i.e., the risk that people could get sick; the loss of land use capability; etc.). Existing groundwater models may not be suitable for SWRMA delineation, and a poorly-constructed/constrained model may be worse than a simple/default method. Uncertainty quantification (UQ) and sensitivity analyses are a central part of any risk-based modelling, and predictions made by a model need to be accompanied by assessments of their uncertainties. Rutter and Moore (2021) developed guidelines for risk-based minimum model design and uncertainty quantification, to provide an indication of modelling and UQ approaches to be used.
This paper outlines different approaches to SWRMA delineation and covers some of the issues and pitfalls that can occur.