Factsheet: Calculating water quality state for rivers

Why are we recording the current state of our freshwaters?

We measure the current state of our rivers, streams and lakes to get an idea about whether our water is suitable for use for various purposes, and about the effect of different pressures (e.g. urbanisation, farming, etc.) on the condition of our freshwater resources.

Some influence from people on our waterways is necessary to support our communities and agriculture, and it is not possible to return most water bodies in New Zealand to their pre-human state. Even where this is almost achievable, such as within national parks, introduced aquatic species such as trout, as well as native and exotic birds, insects and pest mammals, and natural processes such as erosion and atmospheric pollution, impact our waterways to some extent.

How is State calculated for our rivers and streams?

State in the LAWA Rivers Water Quality topic is described using two independent methods: state quartiles and NOF band scores.  The state quartiles are a relative measure, comparing sites against each other, while the NOF band scores are an absolute measure, evaluating each site against expectations of water quality values in the National Policy Statement for Freshwater Management 2020 (Freshwater NPS 2020).

State Quartiles

The state quartile value presented for the river water quality sites on LAWA is based on the median of monitoring results from the last five years (2015-2019).   A state median is calculated when there is at least 50% of the data available over this time-period, (i.e. at least 30 monthly samples over a five-year period), and the quartile ranking is derived by comparison of each site’s median with others. The quartile ranking describes how each site ranks relative to the others – the 1st quartile being the best quarter, the 4th quartile being the worst quarter.  

The results are then presented in four groups from those that are within the best 25% (Quartile 1), to those in the worst 25% (Quartile 4):


  Q1: Best 0-25% of sites  


  Q2: Best 25-50% of sites   


 Q3: Worst 25-50% of sites  


  Q4: Worst 0-25% of sites 


For river water quality indicators, quartiles can be derived for each site by comparing its median values with all other sites around the country, or compared to sites with similar land use and/or altitude. Thus a single site may have various quartile scores, depending on the group it’s being compared against.

The land cover classification for a reach reflects the dominant upstream land cover, grouped as forest (indigenous and exotic), urban or rural. For example, if the catchment of a reach is 50% or more covered in ‘forest’, the reach’s land cover is classified as forest, however, if the catchment of a reach is 15% or more covered by urban land cover, the reach’s land cover is classified as urban. These land cover definitions are based on the New Zealand River Environment Classification (REC).  At some sites, the REC classification does not reflect current land cover, and in these cases councils have provided LAWA with an updated land cover classification.


NOF Band Scores

The Freshwater NPS 2020 requires the management of freshwater in a way that ‘gives effect’ to Te Mana o te Wai.  The National Objectives Framework (NOF) in the Freshwater NPS 2020 sets up requirements for regional councils and unitary authoritaries in setting objectives, policies and rules to manage freshwater in their regions.  One component in the NOF is defining of 'attribute states' or 'bands', which range from A to D (or E).  ‘Attributes’ in the Freshwater NPS 2020 are characteristics of the water that need to be managed by regional councils, the equivalent of LAWA's 'indicators'.

Monitoring data from river water quality indicators are evaluated against NOF attribute states, and LAWA displays this as the A, B, C D or E NOF bands (along with descriptions for context) so you can see which attribute state the water is currently in.


For ammoniacal nitrogen, dissolved reactive phosphorus and E. coli, LAWA requires:

  • a minimum of five years of data
  • data represented from all quarters, and
  • at least 54 of 60 samples available over the last five-year period.


Attribute states for indicators are defined by the lowest band from the numeric attribute state metric (e.g. the annual median, the 95th percentile or the annual maximum result).  For example if the ammoniacal nitrogen annual median result was an A NOF band and the annual maximum result was a B NOF band, then LAWA shows this as a B NOF band.

An example of the descriptions for the attribute bands (e.g. ammoniacal nitrogen) is given below:  

Councils, along with their communities, need to set ‘target attribute states’ (NOF bands) for sites in their region.  For many attributes there is a national bottom line.  That means councils need to aim for the bottom line or better, unless it is considered appropriate to set the NOF band below the national bottom line (e.g. the existing freshwater quality is caused by naturally occurring processes).  As part of setting target attribute states, councils will also set timeframes as to when these will be achieved. 


The difference between State Quartiles and NOF Band Scores

State quartiles are a useful comparative ranking.  They show us how a site compares against other monitored sites. The NOF band provides an absolute ‘score’ for some water quality indicators. 

For some sites the water quality may be good (and score “A” under NOF), yet still be in the 4th quartile. This happens when all the sites are good (and meet the NOF “A” definition); some will still be better, leaving others in the 4th quartile. On LAWA that looks like this:


 State quartile and NOF view


River ecological health monitoring sites

LAWA shows three indicators (Macroinvertebrate Community Index (MCI), taxonomic richness and %EPT) which are used to measure the ecological health of river sites.

The state presented for these sites uses the median calculated from data over the last five years (2015-2019).  A minimum of three data points over the last five years is required for a median value to be calculated for sites that are sampled once a year.  Sites that are sampled twice per year need a minimum of six samples over the last five years.


National Picture Summary - Additional Notes

The Land Cover classes (Native vegetation, Exotic forest, Pasture, and Urban) used in the national picture summary were grouped using the following approach:

LAWA land cover data was acquired on a per-site basis, using the NZREACH that each site fell on, to extract land cover from the REC.  For the native land cover category, we grouped wetland, tussock, scrub and bare (alpine) ground along with indigenous forest, and kept REC’s other categories (exotic forest, pastoral and urban).  The REC determined the land cover category for each reach by calculating the most influential land cover in the catchment upstream, using data from the New Zealand Land Cover Database (LCDB), and acknowledging that less urban cover is required (by area) for it to be most influential.

The REC assigned a category to each reach based on dominance rules for the catchment land cover upstream. “The Land-Cover type that was present in the greatest percentage was deemed to be the dominant land cover unless Pastoral exceeded 25% in which case the class was set to P or if Urban exceeded 15% in which case the class was set to U. If both Pastoral and Urban exceeded their respective thresholds then Urban was assumed to be the most dominant.[1]


[1] New Zealand River Environment Classification User Guide, Snelder T., Biggs, B., Weatherhead M. (2004, updated 2010) MfE, NIWA. https://www.mfe.govt.nz/sites/default/files/environmental-reporting/about-environmental-reporting/classification-systems/rec-user-guide-2010.pdf