View all publications

Modelled river water quality

River water quality guidelines

Having discussed the pressures on freshwater quality in New Zealand, we now examine the current state of water quality.

New Zealand has compulsory values that regional councils must set objectives and limits for – these values are for ‘ecosystem health’ and ‘human health for recreation’. The compulsory values, and their associated water quality attributes, are in the National Objectives Framework (NOF) that is part of the National Policy Statement for Freshwater Management (New Zealand Government, 2014). The framework consists of ‘attribute states’ that describe different effects pertaining to an attribute and a ‘bottom line’ for the minimum acceptable state, which should not be exceeded. These are regulations set by the Government.

We also look at the macroinvertebrate community index, a water quality indicator that scores a water sample in four groups from ‘poor’ to ‘excellent’. Macroinvertebrate taxa tolerate contaminants in different ways – the composition of invertebrates at a site provides information about the site’s water quality. This information is the basis for the macroinvertebrate community index.

We compare river water quality results against the attribute states for the compulsory value in the National Objectives Framework and the macroinvertebrate community index guidelines (see table 2; New Zealand Government, 2014; Stark & Maxted, 2007).

Table 2: River water quality guidelines and attribute states

River water quality compulsory value or guideline

Status

Water quality attribute or variable

What it means

Ecosystem health (trophic state) –National Objectives Framework national bottom lines in the National Policy Statement for Freshwater Management(New Zealand Government, 2014)

Regulatory

Algae (periphyton)

The attribute states consist of four bands, A–D, with A being the best state and D the worst. The national bottom line (the boundary between bands C and D) for periphyton is the level where regular and/or extended-duration nuisance blooms occur, reflecting high nutrient enrichment and/or significant alteration of the natural flow regime or habitat.

Ecosystem health (toxicity) –National Objectives Framework attribute states in the National Policy Statement for Freshwater Management (New Zealand Government, 2014)

Regulatory

Nitrate-nitrogen, ammoniacal nitrogen

The attribute states consist of four bands, A–D, with A being the best state and D the worst. The national bottom line (the boundary between bands C and D) for median nitrate toxicity and pH adjusted ammonia toxicity is the level that would have some growth effects on up to
20 percent of sensitive fish species.

Macroinvertebrate community index –Stark & Maxted, 2007

Non-regulatory

Macroinvertebrate community index

The macroinvertebrate community index is a water quality indicator that groups sites into four categories: excellent, good, fair, and poor. The macroinvertebrate community of a stream lives with the stresses and changes that occur in the freshwater environment, whatever their cause, human-induced or natural.

This report does not include the Australian and New Zealand Guidelines for Fresh and Marine Water Quality (ANZECC) because they are currently under review and the numbers are expected to change. We do report on ANZECC guidelines in our webpages, which can more readily be updated than this report (see Environmental indicators Te taiao Aotearoa – Fresh water [Stats NZ]).

Modelled data estimates water quality for all river segments in the digital river network

We used the River Environment Classification (Snelder & Biggs, 2002) to group river monitoring sites by land-cover class: pastoral, urban, exotic forest, and native. A large proportion of site monitoring is in the pastoral class. To reduce the impact of sites influenced by pastoral land cover being over-represented in our analysis, we used statistical models to estimate median water quality in each of about 560,000 unique river segments in New Zealand’s river network (Snelder & Biggs, 2002) for the period 2009–13 (Larned et al, 2017). The water quality medians were estimated using Random Forest models based on results from river water quality monitoring and a set of explanatory variables that represent catchment characteristics.

Modelled patterns of water quality in New Zealand for four water quality variables are shown in ; figure 7figure 8, figure 9, and figure 10 for the period 2009–13 (see Environmental indicators Te taiao Aotearoa – Fresh water [Stats NZ]​ for additional maps). Overall, the modelling showed that medians for all water quality variables tended to be worse in low-elevation areas on the east coasts of the North and South islands, inland Waikato, Wairarapa Valley, Rangitikei-Manawatu coastal plain, Taranaki ring plain, and the Auckland region (Larned et al, 2017). These areas coincide with land used for intensive agriculture and most of New Zealand’s urban centres. Medians, of all the water quality variables, were estimated to be better in the major mountain ranges, large areas of the Department of Conservation estate, and in smaller areas dominated by native forests in Northland and the Coromandel Peninsula.

For each modelled water quality variable, we compared estimated median concentrations or scores with threshold values (macroinvertebrate community index) and attribute states (nitrate toxicity and ammonia toxicity in the National Objectives Framework) for New Zealand river segments over the five-year period 2009–13, where available.

For a small proportion of total river length, estimated median nitrate-nitrogen and ammoniacal nitrogen concentrations did not meet the national bottom lines for toxicity, or had macroinvertebrate community index scores in the ‘poor’ category (1 percent or less of 560,000 river segments by length, see figure 5 and figure 6). These results suggest that with some localised exceptions, nitrate and ammonia levels in our rivers are not toxic to freshwater species (0 kilometres of river length for ammoniacal nitrogen, 2.3 kilometres for nitrate-nitrogen). Our rivers also appear to generally have good macroinvertebrate community index scores, with most river segments in the ‘excellent’, ‘good’, or ‘fair’ categories (4,606 kilometres of river length have a ‘poor’ score; this index does not tell us about the diversity or abundance of macroinvertebrate taxa present).

For more detail see Environmental indicators Te taiao AotearoaRiver water quality: nitrogen [Stats NZ], River water quality: phosphorus [Stats NZ], River water quality: clarity [Stats NZ], River water quality: Escherichia coli [Stats NZ], and River water quality: macroinvertebrate community index [Stats NZ].

Figure 5

river length grouped by National Objectives Framework bands for modelled ammonical nitrogen and nitrate-nitrogen for the period 2009–13
Click to enlarge view
 

This graph shows river length grouped by National Objectives Framework bands for modelled ammonical nitrogen and nitrate-nitrogen for the period 2009–13. Visit the MfE data service for the full breakdown of the data.

Note: Bands A through D represent different states, with A being the best state and D being the worst. The national bottom line is the boundary between bands C and D. At this level, negative impacts on growth and mortality of multiple sensitive species are expected.

Figure 6

river length grouped by categories for modelled macroinvertebrate community index for the period 2009–13
Click to enlarge view

This graph shows river length grouped by categories for modelled macroinvertebrate community index for the period 2009–13. Visit the MfE data service for the full breakdown of the data.

Figure 7

modelled concentrations of nitrate-nitrogen grouped by percentiles for the period 2009–13
Click to enlarge view

This map shows modelled concentrations of nitrate-nitrogen grouped by percentiles for the period 2009–13. Visit the MfE data service for the full breakdown of the data.

Figure 8

modelled concentrations of dissolved reactive phosphorus grouped by percentiles for the period 2009–13
Click to enlarge view

This map shows modelled concentrations of dissolved reactive phosphorus grouped by percentiles for the period 2009–13. Visit the MfE data service for the full breakdown of the data.

Figure 9

modelled macroinvertebrate community index scores grouped by categories for the period 2009–13
Click to enlarge view

This map shows modelled macroinvertebrate community index scores grouped by categories for the period 2009–13. Visit the MfE data service for the full breakdown of the data.

Figure 10

shows modelled concentrations of E.coli grouped by percentiles for the period 2009–13
Click to enlarge view

This map shows modelled concentrations of E.coli grouped by percentiles for the period 2009–13. Visit the MfE data service for the full breakdown of the data.

Models suggest 83 percent of total river length for large rivers was not expected to have regular or extended algal blooms

Many studies both nationally and internationally indicate that the responses of stream periphyton to nutrients are complex and influenced by many factors such as light, flow history, and species composition (Larned, 2010). Nutrient concentration thresholds to achieve periphyton objectives are therefore always uncertain and must be considered along other factors such as river flow regime and shading. In New Zealand, several guidelines and decision-support systems help define nutrient thresholds to manage periphyton growth. The most robust thresholds are specific to different stream types because of the strong influence that variation in flow regime, substrate, and other factors have on periphyton abundance.

To quantify the geographic extent of potential nutrient-periphyton issues, this report combined national nutrient thresholds with modelled nitrogen and phosphorus concentrations (Larned et al, 2015). The resulting analysis estimates the extent of river segments by length that do not achieve the national bottom line for periphyton in the National Objectives Framework (New Zealand Government, 2014).

National scale models of periphyton abundance were derived using the method of Snelder et al (2014) for the 77 National River Water Quality Network (NRWQN) sites. The models predict periphyton abundance as a function of total nitrogen, dissolved reactive phosphorus, hydrological indexes, measures of sunlight, and temperature. The model response variable (periphyton cover) was converted to chlorophyll-a concentration using a correlation between the two abundance measures described by Larned et al, (2015). The models were then used to derive total nitrogen and dissolved reactive phosphorus concentration thresholds for each of 23 River Environment Classification (REC) ‘source of flow’ classes (source of flow distinguishes different river types based on the catchment climate and topography) (Larned et al, 2015).

An assessment of every segment of the REC digital river network, stream order 3 and greater, was made by comparing estimated total nitrogen and dissolved reactive phosphorus concentrations (Larned et al, 2017) with the relevant total nitrogen and dissolved reactive phosphorus criteria. Only segments of order 3 and greater were included, because the input periphyton data to the model was from large river sites (NRWQN sites), so we do not have confidence in the model for lower stream orders.

The assessment first considered whether the segment was able to support conspicuous periphyton. This assessment classified bed substrate into fine or coarse using predicted bed material grain size as described by Snelder et al (2013). Segments classified as having fine bed substrate were assumed to not support conspicuous periphyton, so applying nutrient concentration criteria to manage periphyton biomass is not warranted for these segments. These segments occur in low gradient areas where fine sediments accumulate and stream power is insufficient to move larger material (eg gravel and cobbles).

The models underlying this analysis are uncertain and the thresholds were derived from periphyton and environmental data collected from a limited number of sites that are representative of large New Zealand rivers. The resulting analysis is therefore uncertain and should be considered as only indicative of periphyton issues nationally.

Of the total river length in New Zealand’s digital river network of stream order 3 and greater, estimated periphyton abundance in 17 percent of total river length (approximately 1,708 kilometres) did not meet the national bottom line for periphyton. In 60 percent of the modelled river length (5,988 kilometres) estimated periphyton abundance was estimated to be less than the national bottom line. The remaining 23 percent of river length (2,355 kilometres) has fine bed substrate, and are presumed to mean that periphyton growth is generally not supported.

Regional council periphyton monitoring has expanded since the National Objectives Framework in the National Policy Statement for Freshwater Management was released in 2014. In future reports, we intend to report on the results of those periphyton monitoring programmes.

Public health considerations for swimming in rivers

Animal and human faeces in fresh water can increase the risk of infection from faecal pathogens for those who come in contact with the water.

E.coli is used as an indicator of faecal matter present in fresh water. High concentrations of E.coli are associated with the risk of infection byCampylobacter(a gastrointestinal illness).

This section focuses on the public health risks for swimming in rivers , and not on other factors that may affect the ability to swim in a water body. Although we have yet to develop an indicator for tracking the change in risk from faecal contamination over time, we present the results of our initial work on rivers. In future we will expand this work to include lakes.

See Technical note on the initial assessment of modelled E.coli data for details on the work completed so far. This is an area where data collection and modelling practices are evolving, so it will take time to develop an indicator.

Many factors determine the health risks from recreational activities in rivers. E.coli is one of these. The concentration of E.coli varies over time in any given river. The exposure to risk is a complex interaction between a specific activity and its duration, and the E.coli concentration of the specific water body at the time of recreation. For example, the risk from E.coli would be different if a person was immersed in the water (such as swimming, where water may be swallowed) as opposed to being on the water (such as fishing), and whether it was a quick dip or a longer session.

To provide information on the risk of exposure to E.coli during recreational activities in rivers, it is necessary to define what we mean by ‘swimmable’ and how it relates to the risk of infection from pathogens. Our analysis used the grade and category system outlined in the Clean Water Package released in February 2017 (Ministry for the Environment, 2017), as opposed to current attribute states in the National Objectives Framework 2014. The Clean Water package assigns rivers to one of five grades (excellent, good, fair, poor, and very poor). The top three grades combine to a swimmable category; the last two grades would not be considered swimmable.

Snelder et al (2016) used statistical models to estimate E.coli concentrations in each of approximately 560,000 unique river segments in New Zealand’s river network. We only looked at river segments of stream order 4 and higher, as these are rivers considered large enough to swim in. The dataset used to inform the model is based on E.coli data collected between 1990 and 2013 (monitoring longevity varies across sites).

There are a number of statistics which describe and quantify E.coli values. Snelder et al, (2016) used four approaches for quantifying E.coli values:

  • median E.coli concentration (half the values are above/below this value)
  • E.coli concentration as a 95th percentile (95 percent of the numbers are below this value – this is the statistic used in the National Objectives Framework)
  • percent exceedances of the current upper threshold of 540 E.coli
  • percent exceedances of the lower threshold of 260 E.coli.

The 95th percentile statistic was the most imprecisely modelled of all four statistics, so we have excluded it from our results (see Technical note on the initial assessment of modelled E.coli data for commentary on its statistical performance) (analysed by Stats NZ). The remaining three statistics have such a high level of agreement that any combination will show an estimated 65–70 percent of segments fall under the ‘swimmable’ category.

This differs from the 72 percent figure stated in the Clean Water Package which includes both rivers and lakes, and takes into account adjustments made to reflect knowledge from regional councils and other sources.

The overall health risk from the three statistics in the swimmable category averaged over time is between 1 and 3.5 percent (Prime Minister’s Chief Science Advisor, 2017). The actual risk depends on the E.coli concentration at the time of activity, which is why councils undertake monitoring at popular swimming spots during summer. For the most up-to-date information on your local swimming spot, see Land Air Water Aotearoa [LAWA].

Along with pathogens from faecal contamination, many other factors affect the ability to swim in a particular water body. These include low water clarity, fast currents, rocks, dense weed beds, and toxic cyanobacteria (see Toxic cyanobacteria in Greater Wellington rivers).

Toxic cyanobacteria in Greater Wellington rivers

Toxic cyanobacteria in a water body is another public health concern when swimming. Toxic cyanobacteria form distinctive black or brown to dark green mats, and have an earthy, musty odour, especially when they break free from the riverbed and collect along the shoreline. People should avoid contact with cyanobacteria mats, and dogs and livestock should be kept away from them.

Many regional councils monitor cyanobacteria. Here we present a case study looking at the Greater Wellington region.

When five dogs died in the summer of 2005/06 after consuming toxic cyanobacteria, the Greater Wellington Regional Council began monitoring the abundance of cyanobacteria in its rivers and streams. Cyanobacteria, sometimes called blue-green algae, are an important and naturally occurring component of freshwater ecosystems. Some species produce toxins that can be harmful if present in drinking water or if humans come into contact with them through activities such as swimming. Dogs are particularly susceptible because they are drawn to the odour of some river cyanobacteria, and more than 70 dog deaths have been reported across New Zealand since 2006.

Monitoring and research by the Greater Wellington Regional Council showed large gravel-bed rivers, such as the Hutt and Waipoua rivers, have the largest and most frequent toxic cyanobacteria blooms. Combined low levels of dissolved reactive phosphorus and moderately elevated nitrate-nitrogen concentrations appear to foster the growth of cyanobacteria blooms in some Wellington rivers.

Source: Greater Wellington Regional Council, 2016