• About this site|
• Site index|
• Contact us

Publications

• About the Ministry
• Air quality
• Biodiversity
• Climate change
• Energy
• General
• Hazardous substances
• Land
• New and genetically modified organisms
• Oceans
• Resource management
• State of the environment
• Sustainable development
• Treaty and iwi
• Urban
• Waste
• Water

“Trigger” values for New Zealand rivers

NIWA Client Report, MfE002/22 May 2000

Executive Summary

Trigger values for ecosystem protection are needed in the new ANZECC guidelines that are currently being drafted. If median water quality at monitoring sites exceeds these trigger values, then it is intended to “trigger” a management response. Ministry for Environment has asked NIWA to estimate such trigger values for New Zealand rivers using percentiles taken from the National Rivers Water Quality Network (NRWQN) datasets. Of the 77 sites in the NRWQN, 32 are considered baseline or pseudo-baseline (minimally or lightly impacted) and 11 of these sites are lake-fed or affected by alpine headwaters. The remaining 21 sites were used for estimation of trigger values (as 80 or 20 percentiles) after being further categorised into 3 lowland sites (elevation at monitoring site < 150 m) and 18 upland sites. Variables for which trigger values were estimated were: total phosphorus, dissolved reactive phosphorus, nitrogen, oxidised nitrogen (nitrate + nitrite-N), ammoniacal nitrogen (ammonia + ammonium-N), dissolved oxygen as % saturation, pH, black disc visibility, turbidity, and temperature. In order to obtain a meaningful trigger value for temperature, data for February alone (high insolation, low flow) were used. The percentiles obtained for pH and especially for %DO, may require professional judgement in order to derive trigger values because these variables are subject to diurnal fluctuation. The percentiles are presented and discussed with reference to trigger values for SE Australian rivers.

Introduction

Trigger values for ecosystem protection are needed for the new ANZECC guidelines that are currently being drafted. These trigger values indicate marginal water quality for supporting ecosystem health (Hart et al. 1999). Running medians of water quality data measured in monitoring programmes may be compared with these trigger values. If the median value of a water quality variable for a particular site exceeds the trigger value, then it is intended to “trigger” a response on the part of water managers, which might be to initiate special sampling or carry out an investigation of reasons for the degraded water quality.

Several ways of specifying trigger values are envisaged by Hart et al. (1999), but in most cases, and for most water quality “issues” (and therefore variables), the use of statistics from datasets for “reference” sites with minor or minimal impact is recommended. The 80 percentile has been (arbitrarily) chosen where high concentrations or values indicate degraded water quality, and the 20 percentile where a low benchmark is appropriate.

The Ministry for Environment (MfE, Mr Bob Zuur) has asked NIWA to estimate trigger values for New Zealand rivers using percentiles calculated for the National Rivers Water Quality Network (NRWQN) dataset. The NRWQN involves sampling of 77 river sites on 49 different rivers at monthly intervals for a variety of water quality variables (Smith & McBride 1990). This report describes the approach taken for site selection and categorisation, presents the relevant percentiles for selected sites in the NRWQN, and gives a brief commentary on the percentiles, including a comparison with SE Australian trigger values.

Description of the Approach

Ten years of data from the NRWQN have been used to calculate percentiles in order to estimate “trigger” values for the new edition of the ANZECC guidelines. Both 80 and 20 percentiles were calculated: the upper percentiles provide an estimation of the trigger value for most variables, but lower percentiles are needed for some variables (e.g. visual clarity) and both for others (dissolved oxygen, pH).

Of the 77 river sites in the NRWQN, 32 sites are regarded as “baseline” (essentially un-impacted) or “pseudo-baseline” (lightly impacted) (Smith and McBride 1990). Data from these sites were used for the calculation of percentiles.

Temperature is one variable that is strongly seasonal in its distribution, so data for February visits (late summer) at which time a combination of comparatively low flows and still high insolation gives seasonal temperature maxima, were used to derive percentiles for this variable. Note that February temperatures are typically higher than January temperatures in the NRWQN dataset.

The minimally or slightly impacted sites were categorised as “upland” (site elevation > 150 m) and “lowland” (< 150 m) following the approach taken in Australia. Unfortunately, most un-impacted to slightly impacted sites are at relatively high elevation, and only four suitable lowland sites are available in the NRWQN: The Waipapa River (Northland), the Upper Waipa and Ohinemuri Rivers (Waikato Region) and the Haast River (Westland). The last river is alpine in its headwaters, and so the percentiles were re-calculated with Haast data removed.

There are fully 28 upland un-impacted to slightly impacted sites in the NRWQN. However, 10 of these are either lake fed (e.g., Tarawera @ lake outlet, Waikato @ Reids farm) or have alpine headwaters with glaciation (e.g., Waimakariri @ the Gorge, Shotover @ Bowens) or both (Waitaki @ Kurow, Clutha @ Luggate), and so the analysis was repeated with these sites removed, leaving only 18 sites.

Appendices 1 and 2 list the baseline and pseudo-baseline sites, give their elevations and other data, and record notes about their catchment characteristics.

Site categorisation and the estimation of percentiles was done in DataDesk which apparently uses a slightly different algorithm for percentile interpolation from MS EXCEL. Appendix 3 summarises the DataDesk output.

Comments on the New Zealand data

Table 1 summarises the percentiles calculated for a range of variables for the different categories of baseline or pseudo-baseline sites. Percentiles for visual clarity (black disc range, m), turbidity, and temperature are included, as well as for the variables (five nutrient measures, dissolved oxygen, pH) covered for various regions of Australia (trigger values for SE Australia, taken from the draft ANZECC guidelines, are given for comparison in Table 1).

The dissolved oxygen (DO) percentiles seem unlikely to be very useful for estimating trigger values, since they are for daytime sampling whereas diurnal minima usually occur near dawn. For a similar reason pH minima may not be very useful. The temperature percentiles for February alone (e.g., 80 percentile temperature for upland sites = 18°C, and 21.5°C for lowland rivers) are probably more useful as trigger values than year-round values. For all three diurnally (and seasonally) varying variables (temperature, pH and DO), professional judgement might need to be used for deriving trigger values.

Unfortunately, the NRWQN does not include data for chlorophyll a in periphyton, although observations of periphyton as % cover are routinely made. Nor are water column chlorophyll a analyses made, so no trigger values can be given for Chla in the water or on the bed of New Zealand rivers.

Note that turbidity and visual water clarity are closely, inversely, related, and the 80 percentile for turbidity is broadly consistent with the 20 percentile for visibility and vice versa.

Comparison with SE Australian Values

A comparison was made with trigger values for SE Australia (including Tasmania) a region that may be more comparable climatically with New Zealand than the other (tropical and or arid) regions of Australia. This comparison can only be made on “face value”, because analytical methods may differ.

Trigger values for SE Australia, for both total phosphorus (TP) and total nitrogen (TN), are similar to the 80 percentiles from the NRWQN, which is “comforting”. Percentiles for dissolved reactive phosphorus (“DRP” equivalent to FRP) and ammoniacal-N (“NH4”) are also similar to, but somewhat lower than, for SE Australia.

Of the nutrients, the greatest difference between NZ and SE Australia is for oxidised-N (“NO3”). The 80 percentiles for oxidised-N in Table 1 are appreciably higher than for SE Australia, particularly for lowland sites. This raises questions about the designation of these three lowland sites as “baseline” or “pseudo-baseline”. Nitrate-N is a good indicator of land use and it seems likely that the Ohinemuri River in particular (designated as “pseudo-baseline”) is actually appreciably impacted by agricultural runoff. I notice that the trigger value (190 ppb) for upland rivers in Tasmania, which may be climatically more comparable with NZ, is much higher than for upland SE Australia generally (15 ppb).

The 20 and 80 percentile values for % saturation DO are very close, suggesting a very "tight" distribution of the data around 100% saturation. The 80 percentile pH for upland NZ rivers is 8.00 pH units, appreciably above that (7.5) for SE Australia. Other trigger values for pH and DO are well within SE Australian values. As noted above, the DO and pH percentiles may not be very useful as trigger values because of diurnal and seasonal variation.

The 80 percentiles for turbidity in NZ are low compared to the wide range given for SE Australia. No comparative visibility data are available for SE Australia.

Acknowledgements

Thanks to Mark O’Donohue, Monash University, Victoria, for explaining the concepts behind trigger values, and to Bob Zuur, Ministry for Environment, for providing liason. Graham Bryers provided data from the NRWQN and carried out preliminary site sorting and data processing. Graham and Bryce Cooper reviewed the report.

References

Hart, B.T.; Maher, B.; Lawrence, I. (1999). New generation water quality guidelines for ecosystem protection. Freshwater biology 41: 347-359.

Smith, D.G.; McBride, G.B. (1990). New Zealand’s National Water Quality Monitoring Network- design and first year’s operation. Water resources bulletin 26: 767-775.

Table 1: Trigger Values for New Zealand rivers

View trigger values for New Zealand rivers (large table)

Appendix. 1.  NRWQN baseline or pseudobaseline site data

View NRWQN baseline or pseudobaseline site data (large table)

Appendix 2. NRWQN site description

View NRWQN site description (large table)

Appendix 3: DataDesk summaries of percentiles for baseline or pseudo-baseline NRWQN sites

1. Lowland sites (< 150 m elevation) a. Four baseline lowland sites: WH1 (Waipapa), HM1 (Waipa @ Otewa), HM6 (Ohinemuri), GY4 (Haast).

Summaries No
Selector
Percentile 20

b. Three baseline lowland sites (Haast removed being alpine affected)

Summaries cases selected according to

Select NOT Haast

Percentile 20

c. Temp summary is not very useful because of seasonality, so we select just February data

Summary of Temp cases selected according to Select Feb 474 total cases of which 436 are missing Percentile 20

Count 38

Mean 17.8842

Median 18.75

Lower ith %tile 14.67

Upper ith %tile 20.28

Summary of Temp cases selected according to Feb, NOT Haast 474 total cases of which 444 are missing Percentile 20

Count 30

Mean 19.4633

Median 19.25

Lower ith %tile 17.75

Upper ith %tile 21.45

2. Upland sites (> 150 m elevation)

d. 28 Upland baseline sites

Summaries

No Selector

Percentile 20

e. 18 Upland baseline sites (alpine and/or lake-fed sites removed)

Summaries cases selected according to Select NOT glacial, lake Percentile 20

f. Temp summary is not very useful because of seasonality, so we select just February data

Summary of Temp cases selected according to Select Feb 3404 total cases of which 3126 are missing Percentile 20

Count 278

Mean 15.9004

Median 16

Lower ith %tile 13.6

Upper ith %tile 18.09

Summary of Temp cases selected according to Feb, NOT glacial, lake 3404 total cases of which 3226 are missing Percentile 20

Count 178

Mean 15.4309

Median 15.5

Lower ith %tile 13.2

Upper ith %tile 17.9