Road runoff may adversely affect aquatic receiving environments. Contaminants in road stormwater discharges are complex. They include fuels, additives, oil and brake and tyre residues containing a variety of toxic and ecotoxic components, such as heavy metals and organic compounds. Receiving environments from road runoff include streams, rivers, lakes, wetlands, estuaries, harbours and the open coastline. The characteristics of these different types of water body influence the fate of contaminant inputs, how they are assimilated and therefore their sensitivity. Reflecting the need for cost-effective ways of prioritising the management of road runoff in relation to the risk of adverse effects, the NZ Transport Agency commissioned MWH in association with NIWA to develop a screening model that addressed the following research question: Under what conditions is stormwater run-off likely to cause adverse environmental effects? Building on earlier research, the aim of this one-year study was to revise and enhance the Transport Agency’s vehicle kilometres travelled (VKT) screening tool for road runoff to allow its wider application to rivers/streams and coasts/estuaries, and to be able to factor in the effects of pathway attenuation, traffic congestion and non-road contaminant sources. While the principal intention was to develop an improved screening method for road networks, the study also considered how this could be extended to provide for an absolute risk assessment in relation to established effects thresholds. The road stormwater screening (RSS) model developed in this study provides a robust, consistent method for establishing the relative risk of adverse effects from road runoff that can be applied anywhere in New Zealand using existing datasets. Other applications include i) a ‘drill down’ facility to screen the road network for contaminant load ‘hot spots’, ii) apportioning contaminant loads between local roads and state highways and iii) whole of catchment analysis for road contributions. The RSS model is to be used on a comparative rather than absolute basis for screening road networks and their likely risks to receiving environments. It is intended to provide guidance to network operators and road controlling authorities on the management of road runoff and the development of catchment management plans. Reflecting uncertainty in the estimation of contaminant loads, the relative contribution of road and non-road sources of copper and zinc in any given sub-catchment estimated by the RSS model should be considered indicative. While these load estimates provide a fit-for-purpose basis for supporting the risk assessment provided by the screening model, any management response is likely to require a targeted investigation of the relative importance of different contaminant sources in a given stormwater catchment. The US Federal Highways Administration and UK Highways Agency have developed relatively sophisticated models which combine the prediction of contaminant loads and assessment of risk based on the sensitivity of receiving waterbodies. In both cases these authorities have had access to large datasets of road runoff quality and have invested considerable resources in the analysis required to support the development of tools which deliver assessments of absolute risk. The literature review confirmed the suitability of a relative assessment method for the study due to lack of equivalent datasets and resourcing in New Zealand. It also supported adoption of a ‘source-pathway-receptor’ approach where the source is the road section or network that generates a contaminant load in runoff and the pathway is the route from the road to where the runoff enters the receptor or receiving waterbody (e.g. river, lake or estuary). The novel inclusion of factors to account for road traffic congestion, road drainage characteristics and urban land use as moderators of contaminant load and hence risk to the waterbody were considered to be feasible. The review favoured risk being assessed using scores based on source strength and receiving environment sensitivity. These aspects have been incorporated into the RSS model. Commonalities in approaches by US and UK agencies provide a template for a New Zealand-based absolute risk assessment method for road runoff as a potential future stage of research. The review focused on the future approach that regional councils might take for consenting stormwater discharges for road runoff and the potential value of a risk-based process to support consenting approvals/assessments. Of particular importance is the National Policy Statement for Freshwater Management 2014 (NPS-FM), given that it will provide regional councils and unitary authorities with the regulatory framework to set objectives, attributes and limits for water quality. The RSS model is closely aligned with the requirements of the NPS-FM, although it would benefit from inclusion of attribute tables for zinc and copper that may shortly be under consideration by the National Objectives Framework Reference Group. The RSS model has been developed using a combination of ESRI ArcGIS and MS Excel. ArcGIS is used to determine data inputs and to map results, adopting the River Environments Classification as the basis for analysis. Excel provides the platform for estimation of contaminant loads and risk. Key concepts underlying the RSS model are: • the adoption of a source-pathway-receptor conceptual model • a focus on potential effects of copper and zinc, using New Zealand road runoff sampling data collected under previous Transport Agency research and other complementary studies, recognising that these metals also function as proxies for a wider range of stormwater contaminants • the ability to assess risk associated with discharges to a range of receiving waterbodies including rivers and streams, and coasts and estuaries. The VKT screening tool, developed under a 2007 Transport Agency study, provided the starting point for the RSS model. The VKT tool also adopted a source-pathway-receptor approach but was limited to the assessment of depositional receiving environments and assessed risk using VKT by sub-catchment as a proxy for contaminant load. The RSS model has added a range of significant enhancements and new features to the VKT method. These include: 1 A road contaminant load module for estimation of zinc and copper loads at the sub-catchment level from road traffic, vehicle emission factors that vary in relation to traffic congestion, pathway attenuation and conversion of VKT to contaminant load 2 A non-road (urban) contaminant load module for estimating zinc and copper loads at the subcatchment level from the extent of urban non-road impervious surfaces and contaminant yields for residential and industrial/commercial areas, respectively 3 A method for assessing risk to streams and rivers, based on estimates of in-stream copper and zinc concentrations relative to guideline concentrations combined with a receiving environment sensitivity score indicated by modelled values of the macroinvertebrate community index 4 A method for assessing risk to coasts and estuaries, based on estimates of copper and zinc concentrations in sediments delivered to coastal discharge points relative to guideline concentrations combined with a receiving environment sensitivity score determined from the physical depositional characteristics of the water body. The RSS model was evaluated in a case study applied to Te Awarua-o-Porirua Harbour and its catchment. The area was chosen because it comprises a mix of local roads and state highways that discharge to streams of varying dilution potential as well as coastal waterbodies (Pauatahanui Inlet and Onepoto Arm). The sub-catchments of these two waterbodies contain markedly different urban/rural land use (with Onepoto more urbanised), allowing for model evaluation under widely different catchment conditions. Streams and rivers risk assessment: For the assessment of road-traffic risk to rivers and streams, the majority of reaches are classified as ‘lowest risk’. In most of these sub-catchments, the streams are able to adequately dilute the relatively small copper and zinc loads discharged in road runoff from the limited extent of roads present. Exceptions are the Onepoto Arm sub-catchments and those south of Pauatahanui Inlet containing SH1 and/or relatively dense local road networks, and which are drained by streams with limited dilution potential. Road traffic risk in these sub-catchments is assessed as ‘highest risk’. In contrast, most sub-catchments containing any urban land use are classified as ‘highest risk’ according to the urban risk assessment. Thus, even in sub-catchments containing relatively limited areas of urban impervious surfaces, loads of copper and zinc are high in relation to stream dilution potential. Only a small proportion of sub-catchments containing urban land are assessed as being in lower risk categories. These results show the importance of including contaminants from other urban sources as well as from road traffic when assessing stormwater risk. Considering road traffic derived contaminants in isolation has the potential to lead to a marked under-assessment of risks to aquatic environments. Coasts and estuaries risk assessment: The distribution of metal loads at coastal outlets around the Harbour reflects the mix of urban and rural land use and size of the catchments that discharge to these locations. The stormwater risk profile is determined by the combination of contaminant concentration and receiving environment sensitivity at each discharge point. For Pauatahanui Inlet, the highest risk occurs for discharges to Browns Bay, reflecting the combination of high copper and zinc loads and the Bay being partly enclosed and highly depositional. While the urban contribution drives the overall ‘highest’ risk score, road traffic makes a significant contribution at this location, with the traffic-related risk classified as ‘lower’ to ‘medium’. For Onepoto Arm, where land use is predominantly urban, no stormwater outlets were found with a ‘highest’ risk level. The load profiles for zinc and copper are both dominated by the relatively high (80% plus) contribution from Porirua Stream and its catchment that discharges to a single outlet (which includes the bulk of Semple Street stormwater catchment draining Porirua’s CBD). Despite having the highest metal load, the high sediment load discharged from the Porirua Stream catchment results in this outlet being ranked only 10th (for zinc) and 11th (for copper) in terms of sediment metal concentrations compared with all 13 outlets, with a resultant risk level of ‘medium’. Sensitivity analysis and validation: A sensitivity analysis assessed the extent to which uncertainty in model inputs influences the assessment of risk. Varying the road-derived loads of copper and zinc by ±15% was found to change the road-traffic risk classification of around 3% of stream reaches in the case study catchment. The adoption of higher residential roof zinc yields in the estimation of urban zinc loads resulted in an approximate 2% increase in the proportion of stream reaches assessed as ‘highest’ risk. The minor change in risk level indicates that assumptions made in estimating contaminant loads are unlikely to have a major bearing on the outcome of the risk assessment, thus providing assurance in the reliability of the RSS model output. Model validation compared modelled concentrations of copper and zinc with observations from water and sediment quality monitoring programmes in the case study area. While based on very limited samples, mean in-stream concentrations of modelled copper and zinc provided a reasonable reflection of observed stream water quality. Estimates of zinc concentrations were approximately an order of magnitude higher than those for copper. The relativity between sites for modelled and observed concentrations was found to be consistent. For the coastal validation, there are no data collected near outfalls with which a direct comparison could be made. Observed sub-tidal concentrations for zinc and copper in sediment were typically up to an order of magnitude lower than those modelled at coastal outlets. This is attributed to significant reworking and dilution of sediment from the point of discharge near the intertidal zone. Limited data on observed metal concentrations in sediment taken from nearby catchpits that discharge to one outlet suggest reasonable agreement with the ‘delivered’ sediment metal concentrations modelled at this discharge point. This study also looked at further development of the RSS model in providing for assessments of absolute risk. Key aspects of the such work would include: collection of additional data on road runoff quality to allow development of a more comprehensive set of vehicle emission factors related to traffic congestion and road characteristics; incorporation of methods for estimating the frequency with which acute toxicity guideline concentrations of copper and zinc in rivers and stream are exceeded; incorporation of methods for estimating metal accumulation in estuary bed sediments; and undertaking field surveys of stream and estuarine ecological condition. (Author/publisher)
Samenvatting