Factsheet: Calculating water quality trends

LAWA displays trends for the last five and ten years for the river water quality data. Data are evaluated to determine whether water quality indicators are showing improving or degrading trends.  An indeterminate trend is given when there are insufficient data to determine the direction of the trend.

Why do we calculate trends?

LAWA calculates trends to show how the quality of water in rivers is changing at each site over time. Changes in water quality often take several years to be seen and longer trends are generally more reliable.  This is because some chemical measures of water quality can be influenced by flow, seasonal and climatic cycles (weather patterns). 


What river water quality parameters does LAWA show trends for?

LAWA displays river water quality trends for eight water quality indicators: water clarity, turbidity, E. coli, total nitrogen, total oxidised nitrogen, ammoniacal nitrogen, dissolved reactive phosphorus and total phosphorus.  


How do we calculate water quality trends? 

Trends are calculated for the last five and ten years. Data are evaluated to determine whether water quality is showing improving, degrading, or indeterminate trends.

 The data used to calculate water quality trends for rivers is collected monthly, quarterly or bi-monthly. Sites that are sampled monthly will have more data than sites that are sampled quarterly. Generally, the more data points we have available for a site, the more robust the trend analysis will be. LAWA displays trends for the last five and ten years for sites with monthly data, but only ten-year trends for sites with quarterly and bimonthly data. 

 Sites were excluded from the analysis if they had less than 90% of the data expected in any one year of the data record, or if more than 30% of the data was less than or greater than laboratory detection limits. 

To determine whether water quality at a river site is showing improving, degrading or indeterminate trends, LAWA uses the ‘Seasonal Kendall Trend Test’. This test compares the water quality data of each season separately (January with January, February with February, etc) which means if any changes are detected they are not due to seasonal patterns.  


A bit more detail on river trend analyses 

The river trend analysis looks at the confidence intervals on either side of the estimated slope for each site and parameter combination. If zero is outside the 95th confidence interval range then the trend direction is determined with confidence, which means that we can be certain that water quality is showing signs of degrading or improving trends at a certain site for a certain parameter. However, if zero is within the confidence interval range, it is concluded that there are insufficient data to determine the trend direction - this is when LAWA reports an indeterminate trend. This approach is statistically powerful and used in recent national trend analysis summaries (e.g. Larned et al. 2015).  

Ideally, data should be flow adjusted before trend analysis to remove any effects of variation in stream flow. However, many councils do not measure river flow at their water quality sampling sites which is why all data used for trend analysis on LAWA are not flow-adjusted. Larned et al. (2015) showed that flow adjustment makes relatively little difference to any trends identified for most sites when data is considered at a national level. 


LAWA shows three different trend categories for rivers: 


1 Improving Trend

 Improving trend_Apr2018


 This symbol is used for sites that show an improving trend in water quality. An improvement is generally a reduction in concentration of a water quality parameter, for example, phosphorus concentrations, except, for visual clarity where an improvement is an increase in clarity.  



2 Indeterminate Trend

 Indeterminate trend_Apr2018


 This classification is given to sites where there is insufficient evidence to clearly say if water quality trend is improving or degrading. An indeterminate trend does not mean that the water quality at a site has stayed the same, but rather that insufficient data are available to report a certain trend with confidence. 



3 Degrading Trend

Degrading trend_Apr2018


 This symbol is used for sites that show a degrading trend in water quality over time. A degrading trend is generally an increase in concentration of a water quality parameter, except, for visual clarity where a decline is a reduction in clarity. 




As part of the annual update of data on LAWA, the lake trend analysis methodology will be updated and matched with the current river water quality trend analysis in September 2018. This means that LAWA will be moving from the current p-value approach to the confidence interval approach. Watch this space to find out in detail how LAWA analyses lake trends. 



Where do I find more information? 

Larned S, Snelder T, Unwin M, McBride G, Verberg P, McMillan H 2015. Analysis of water quality in New Zealand lakes and rivers. Prepared for Ministry for the Environment. NIWA Client Report No. CHC2015-033. 

 Our fresh water 2017.  A joint report by the Ministry for the Environment and Stats New Zealand.