Region-specific statistics and national summary tables (Spreadsheets 1 and 2) referred to in this section are available as downloadable Excel files.
3.1 Site-specific assessments of state and trends
Site-specific median values and trend assessments are compiled in Spreadsheet 1. Site-specific medians and trends do not differ significantly for the time period 1995 to 2008 (this report) compared to the time period 1995 to 2006 (Daughney and Wall, 2007). In general, site-specific medians for the two time periods differ by less than ±10% except for a small proportion of sites where:
- Few samples have been collected, such that inclusion of data from additional samples collected between 2006 and 2008 leads to a significant change in the calculated median;
- The concentrations are near the analytical detection limit, such that a small change in absolute concentration between individual samples translates to a large relative (%) change; and/or
- There is a significant temporal trend at the site in question.
Similarly, site-specific median values do not differ substantially for the time period 1995 to 2008 compared to the 2008 year alone, except in the cases outlined above.
3.2 National overview of groundwater quality
National-level statistics related to state and trends in individual parameter categories for the period 1995 to 2008 are compiled in Spreadsheet 2 and summarised in Tables 3, 4, 5 and 6 (analogous to Tables 7, 8, 11 and 12, respectively, from Daughney and Wall, 2007). The national-level statistics show:
- National medians for most parameters (Table 3) are very similar to previously reported values (Daughney and Wall, 2007). National medians for major elements are intermediate between the global average for river water and the global average for groundwater, often being closer to the former.
- Only a small percentage of sites have median parameter values in excess of the relevant MAV, GV or TV (Table 4), except for a few previously recognised nationally significant groundwater quality issues, including NO3-N, Fe, Mn, and E. coli.
- A significant increasing or decreasing trend is detectable at about 25% of the monitoring sites for most of the parameter categories for which trends could be calculated, and the proportions of sites showing increasing and decreasing trends are the same to within about 10% (Table 5), in agreement with previous findings (Daughney and Wall, 2007).
- The national median values for absolute trend magnitude are less than 0.5 and 0.01 mg/L per year for most major and minor elements, respectively (Table 6). Compared to relevant median concentrations, this equates to national median relative trend magnitudes of less than ±2% and ±5% per year for most major and minor elements, respectively. Such rates of change are considered to be “slow”, as might occur naturally due to the process of water-rock interaction (Daughney and Reeves, 2006). There are no guidelines as to what rate of change in groundwater quality is acceptable from a resource management perspective, but for context it is relevant to note that a relative rate of change of greater than the arbitrary cut-off of ±1% is considered “ecologically meaningful” in rivers (Scarsbrook, 2006).
- Absolute rates of change are uncorrelated or only weakly correlated to median concentrations. For example, there is no relationship between the median concentration of NO3-N at a particular site and the rate at which NO3-N is changing over time.
Table 3. Calculated national percentiles and maximum values for groundwater quality parameters, based on site-specific median values determined for the period 1995 to 2008. Global average concentrations for river water and groundwater are given for comparison. All values in mg/L except E. coli (cfu/100 ml), Cond (μS/cm), pH (pH units) and Temp (°C).
|Parameter||New Zealand Groundwater (this report)||Global Averages1|
|n2||Percentiles||Max.||River Water||Ground Water|
|Minor or Biological||B||367||0.01||0.01||0.03||0.05||0.39||13.50||20-1000|
1 Global averages are provided for comparison to 50th percentile values in New Zealand groundwater. Global averages are taken from Turekian (1977), Hem (1985) and Langmuir (1997). Ranges of values are taken from Dragun (1998).
2 Total number of sites for which statistics could be calculated (i.e. parameter had been measured at least three times).
3 Similar percentiles are obtained for measured TDS, but statistics are based on fewer measurements and so are not tabulated here.
Table 4. Percentage of New Zealand monitoring sites at which median concentrations calculated for the period 1995 to 2008 are in excess of water quality standards or guidelines.
|Reason||MAV or GV||%Sites Exceeding1||Reason2||TV||%Sites Exceeding1|
|Minor or Biological||B||mg/L||Health||1.4||2.2||Toxicity||0.37||5.2|
|E. coli 4||cfu/100 ml||Health||1||23.1||Livestock||100||2.4|
1 Percentage of monitoring sites at which median exceeds the water quality standard or guideline, relative to the total number of sites for which a median could be calculated for the parameter in question.
2 The listed ANZECC TVs pertain either to direct toxicity to biota, or to non-toxicity related threat to aquatic ecosystems, or to the safe threshold for stock drinking water. Note that exceedence of an ANZECC TV in groundwater will not necessarily lead to adverse ecological consequences in adjacent surface waters on all occasions, because groundwater discharging to a surface water body may mix with the surface water, leading to dilution and reduction of the concentration of the parameter of concern.
3 Similar results are obtained for measured TDS, but statistics are based on fewer measurements and so are not tabulated here.
4E. coli is the only microbiological parameter that is considered in the drinking water standards (Ministry of Health, 2005).
Table 5. Number of monitoring sites (n) across New Zealand at which trend tests could be performed for the period 1995 to 2008, and percentages without significant trends (%N) or with significant increasing (%INCR) or significant decreasing (%DECR) trends (at 95% confidence level).
|Minor or Biological||B||276||8.3||13.4||78.3|
Table 6. National absolute and relative rates of change in groundwater quality parameters for sites with statistically significant trends. Relative median rates of change calculated by dividing the median absolute trend by the relevant median concentration from Table 3.
|Parameter||Units||Absolute Trend (units per year)||Relative Median Trend (% per year)|
|Minor or Biological||B||mg/L||-0.016||-0.001||0.264||-4.36|
|E. coli||cfu/100 ml||-37.03||-0.24||4.16||-|
3.3 Key indicators of groundwater quality
Figure 1 (a-l) displays the national and regional percentiles and exceedence and trend statistics for the six key groundwater quality indicators, based on all available data collected within the period 1995 to 2008. Figure 2 shows the year-by-year change in the percentage of sites exceeding relevant DWSNZ and ANZECC guidelines. Figures 3 (a-o) to 8 (a-o) show the changes in the percentiles of the key indicator parameters, by region and by calendar year. The information displayed in Figures 1 to 8 is also provided in table form in Appendices 1 to 15 (all appendices are downloadable from the Ministry for the Environment website). Table 7 presents the results of trend tests conducted to determine if there are year-by-year changes in the national and regional percentile values, or year-by-year increases or decreases in the percentage of sites exceeding DWSNZ or ANZECC guideline values.
This report has revealed a national median concentration of 1.7 mg/L for NO3-N based on all data collected in the period 1995 to 2008 (Figure 1a, Table 3), slightly higher than the national median of 1.3 mg/L NO3-N reported for the period 1995 to 2006 (Daughney and Wall, 2007). The slight increase in national median does not indicate an increase in NO3-N contamination of New Zealand’s aquifers in the period 2006 to 2008, but rather is caused by the unavailability of data from the Gisborne region for this investigation. Gisborne is dominated by oxygen-poor groundwater with low concentrations of NO3-N (Daughney and Wall, 2007), and the exclusion of the Gisborne SOE data yields a slightly higher national median concentration of NO3-N. Perspective on the calculated national median is provided by previous studies, which have estimated 0.3-1.0 mg/L for median NO3-N concentration in unimpacted groundwaters in New Zealand (Burden, 1982; Morgenstern et al., 2004; Daughney and Reeves, 2005). Daughney and Reeves (2005) also defined NO3-N thresholds of >1.6 mg/L and >3.5 mg/L as “probably” and “almost certainly” indicative of human influence, respectively.
The regions with the highest median NO3-N concentrations are Waikato (4.2 mg/L), Southland (3.4 mg/L) and Canterbury (3.4 mg/L) (Figures 1a and 3c, j and m). NO3-N in groundwater is a known concern in these regions, and hence more investigations are being undertaken in at-risk aquifers, and SOE sites with high NO3-N are being added to these regional monitoring programmes over time. This report also considered data from a small number of non-SOE sites in Waikato and Northland that are sampled specifically to assess NO3-N contamination in groundwater and/or saltwater intrusion, which may bias the statistical compilations for these regions. Note also that approximately 90% of the SOE monitoring sites in the Waikato, Southland and Canterbury regions are typified by oxygen-rich conditions (Daughney and Wall, 2007), which favour the persistence of any introduced NO3-N. The lower median concentrations of NO3-N observed for other regions might indicate a lower degree of human influence, and/or the predominance of oxygen-poor conditions that would cause introduced NO3-N to be converted to some other form of nitrogen (e.g. NH4-N, N2, N2O). For example, the majority of SOE sites in the West Coast are typified by oxygen-rich groundwater (Daughney and Wall, 2007), meaning that the relatively low regional median NO3-N concentration (1.1 mg/L) is indicative of a presently generally low level of human influence. In contrast, regions such as Auckland, Manawatu-Wanganui and Hawke’s Bay are dominated by oxygen-poor groundwater. The low median NO3-N concentrations for these
Table 7. Trends in annual percentiles (units per year) and exceedence levels (% sites above standards or guidelines) for selected water quality indicators. Positive and negative numbers indicate rates of change for significant increasing and decreasing trends, respectively (95% confidence level). Null entry (-) indicates that trend test was performed but no significant trend was detected. Blank entries indicate cases where the trend test could not be performed due to lack of data. The asterisks (* and **) indicate cases where rate of increase or decrease is greater than 0.1 and exceeds 10% of the corresponding median, respectively.
|Rates of change in annual percentile and % exceedence values (units per year)|
|% > MAV||-||-||0.3||-||-||-||-||-0.5||-||-||-1.0||-||-1.0||-|
|% > TV||0.4||0.7*||-||1.1||-||-||-||-||-0.6||-||-||-0.9||-0.8||-0.2||-|
|% > GV||-||-||-||-0.1||-||-||-||0.4*||0.3||-||-||-||-||-||-|
|% > TV||-||0.4||-||-||-||-||-||-||0.5*||-||-0.8||-||-||-||-|
|% > GV||-1.8||-||-||-||-||-||-||-||1.9*||-2.3||-||-0.6||-||-2.6||-|
|% > MAV||-2.2||-||-||-||-||3.3||-||-||-||-1.7||-||-||-||-1.4||-0.3|
|% > GV||-1.1**||-||-||-||-||-||-||-||0.6||-||-||-||0.5*||-||-|
|% > TV||-0.1||-||-||<0.1||-||-||-||0.7*||-||-||-||-0.4||-||-||-|
|% > MAV||-||-||-||-||-||-||-||-4.2**||-||-||-||-||-|
|% > TV||-||-||-||-||-||-||-3.5**||-1.1**||-||-||-||-||-|
* indicates cases where rate of increase or decrease is greater than 0.1 of the corresponding median
**indicates cases where rate of increase or decrease exceeds 10% of the corresponding median
regions (<0.1 mg/L – see Figure 3a, e and f) do not necessarily show that the groundwater isn’t polluted or never was polluted, but might instead indicate that the evidence of pollution has been “erased” by natural processes.
Nationally, 4.8% and 13.2% of monitoring sites have median NO3-N above the MAV defined in the DWSNZ (11.3 mg/L) and the toxicity-related TV specified in the ANZECC guidelines (7.2 mg/L), respectively (Figure 1a, Table 4). A much greater proportion of monitoring sites (73.2%) have median NO3-N above the ANZECC TV defined for ecosystem protection (in surface waters). While the use of ANZECC TV guidelines provides useful context here, it must be emphasised that the actual concentrations of NO3-N in any surface water body receiving NO3-N-rich groundwater will be entirely dependent on site specific factors such as dilution potential. The regions with the highest proportion of sites with median NO3-N concentration above the DWSNZ and/or ANZECC guidelines are Waikato, Taranaki, Southland, and Canterbury (Figure 1a).
Significant time trends in NO3-N concentration are detectable at roughly one third of the monitoring sites across New Zealand, and of these, roughly twice as many sites display increasing trends compared to decreasing trends (Figure 1b, Table 5). Previous studies in New Zealand and overseas have shown that for parameters such as NO3-N, the proportion of sites showing increasing trends exceeds the proportion sites showing decreasing trends, presumably due to intensification of agricultural activity (Frappaporti et al., 1994; Daughney and Reeves, 2006). The absolute rates of change are generally slow (cf. Daughney and Reeves, 2006), i.e. less than ±0.3 mg/L NO3-N per year, for the majority of sites. Nationally, there is no significant year-by-year change in the proportion of sites with median NO3-N exceeding the MAV, but the proportion of sites exceeding the toxicity-related TV (7.2 mg/L) is increasing slowly at 0.4% per year (Figure 2a, Table 7).
Regional trends are most pronounced in the West Coast, where half of the monitoring sites exhibit significant temporal trends, and all of these reveal increases in NO3-N concentration over time at rates that are significantly above the national average (Figure 1b). Identification of time trends in NO3-N concentration in the West Coast region is facilitated by the constancy of the structure of the regional monitoring programme, with few changes in sampling methods or sites within the period 1998 to present. The Canterbury, Waikato and Marlborough regions also display a greater proportion of sites with increasing compared to decreasing trends in NO3-N, and/or rates of change slightly above the national average. On a year-by-year basis, there are clear increases in the regional percentiles in NO3-N concentration in the West Coast (Figure 3o) and Canterbury (Figure 3c), and slower increases in the upper percentiles in Hawke’s Bay (Figure 3e) and Northland (Figure 3h) (Table 7). Canterbury is the only region with a significant year-by-year increase in the proportion of monitoring sites at which median NO3-N concentration exceeds the health standard (Table 7).
The national median concentration of NH4-N is 0.01 mg/L, based on all data collected in the period 1995 to 2008 (Figure 1c, Table 3). The regional assessments reveal the expected inverse correlation between NO3-N and NH4-N. The regions with the highest median concentrations of NH4-N are Manawatu-Wanganui (0.3 mg/L), Hawke’s Bay (0.1 mg/L) and Auckland (0.1 mg/L). As noted above, the SOE programmes of these regions are dominated by monitoring sites with oxygen-poor groundwater, in which nitrogen exists predominantly as NH4-N and NO3-N concentrations tend to be low. Regions previously noted to have the highest median concentrations of NO3-N have among the lowest median concentrations of NH4-N, e.g. Canterbury, Southland, and the West Coast.
Nationally, 3.8% of monitoring sites have median NH4-N above the GV defined in the DWSNZ (1.5 mg/L). Nationally, 5.3% and 37.7% of monitoring sites have median NH4-N above the toxicity-related and the ecosystem protection TVs specified in the ANZECC guidelines (0.9 and 0.01 mg/L, respectively; Figure 1c, Table 4). There is no nationwide year-by-year trend in the percentage of sites exceeding the DWSNZ or ANZECC guidelines (Figure 2b, Table 7). Manawatu-Wanganui, Hawke’s Bay and Taranaki are the regions with the highest proportions of sites having median NH4-N above the DWSNZ and/or ANZECC guidelines.
Time trends in NH4-N concentration are detectable at 21% of the monitoring sites considered in this report, of which about two thirds show significant increases in concentration over time (Figure 1d, Table 5). Absolute rates of change are generally less than ±0.05 mg/L NH4-N per year across all regions. The Manawatu-Wanganui region is the exception, where the few sites with increasing trends display a rate of increase in NH4-N concentration that is significantly faster than the national average. Other notable trend patterns are evident in Canterbury (Figure 4c), Northland (Figure 4h) and Otago (Figure 4i), where there are observable year-by-year increases in the percentile values and, for the latter two regions, also year-by-year increases in the percentage of sites exceeding DWSNZ and/or ANZECC guidelines (Table 7). For these latter three regions, the trend patterns appear to reflect the addition of monitoring sites with high NH4-N concentrations to the SOE programmes from about 2000 onwards.
3.3.3 E. coli
Calculated statistics for E. coli concentrations must be assessed with caution due to historical differences in sampling records and because proxy microbiological parameters such as total coliforms or faecal coliforms were employed by some regions for some time periods. For example, the Wellington region analysed total coliforms at 3 to 14 sites per year from 1995 to 2002, then switched to analysis of E. coli at 40+ sites per year from 2003 onwards (Figure 5). Similarly, the Bay of Plenty (Figure 5b) and Hawke’s Bay (Figure 5e) regions have historically measured faecal coliform counts, which may not be directly suitable as a proxy for E. coli concentrations. In this report, the main justification for combining E. coli counts with other proxy microbiological parameters is to ensure that the maximum amount of data can be used for median and trend calculations. There is no straight-forward reliable method for groundwater samples that can be used to convert total coliform counts into E. coli counts. The presence of E. coli always indicates faecal contamination by warm-blooded animals, whereas the total coliform group includes organisms of faecal and environmental origin, and hence total coliform counts are always greater than E. coli counts. This is clearly visible in the Wellington region (Figure 5n), and hence the grouping of these two variables leads to a calculated regional median that is relatively high compared to other regions (Figure 1e). Some regions have always monitored E. coli (as opposed to some other microbiological parameter), but only at a highly irregular interval. For example, E. coli data for the Marlborough (Figure 5g) and Taranaki (Figure 5k) regions were only available for the 2008 and 2006-2007 years, respectively and hence calculated regional medians may not be particularly robust. Canterbury, with a median E. coli concentration of 0.7 cfu/100 ml, is really the only case where a sufficient density of data was available over several calendar years.
The national median for E. coli is less than 1 cfu/100 ml, based on all data collected in the period 1995 to 2008 (Figure 1e, Table 3). The calculated percentiles of E. coli concentration indicate a heavily skewed distribution: the majority of results at most sites are near or below the detection limit (1 cfu/100 ml), but occasionally a much higher result is reported. The rare elevated microbiological counts might reflect a flooding event at a site, for example after heavy rain, or perhaps contamination that has occurred at a site with poor well-head protection or during sampling. The calculated national percentiles are also strongly biased by the irregularity of sampling record and analytical methodology for the various regions (see above).
Nationally, 23.1% and 2.4% of the monitoring sites across New Zealand have median E. coli concentrations above the MAV for human consumption (1 cfu/100 ml) and the TV for livestock consumption (100 cfu/100 ml), respectively (Figure 1e, Table 4). For comparison, Daughney and Wall (2007) reported a national exceedence level of 20% based on the MAV and data collected in the period 1995 to 2006, and Sinton (2001) reported exceedence levels of 9-60% for previous regional surveys. E. coli is somewhat unusual amongst the key indicators of groundwater quality in that its MAV is the same as its detection limit. Hence compared to the other indicators of groundwater quality, a measured E. coli concentration that exceeds the MAV has a greater chance of being caused by contamination during sampling or analysis. The greatest proportions of sites that exceed the MAV and/or TV are found in Taranaki (70%), Auckland (33%), Otago (30.6%), Waikato (25.0%) and Northland (25%) (Figure 1e). E. coli is the only microbial indicator parameter that is actually considered in the DWSNZ; all other microbial parameters are considered to be proxies for E. coli, and so exceedence data from the Wellington, Hawke’s Bay and Bay of Plenty regions must be considered with caution.
Significant time trends in E. coli concentration are detectable at only 2% of the monitoring sites considered in this report (Figure 1f, Table 5). Nationally, there is no significant year-by-year change in the proportion of sites with median E. coli concentrations exceeding the health standard (Figure 2c). These results likely reflect that trend tests have low power to detect changes in E. coli concentration over time because the historical monitoring record is sparse, non-continuous and irregular at most monitoring sites, and because elevated E. coli counts tend to occur sporadically, possibly due to flooding events at sites with poor well-head protection or to contamination during sampling or analysis.
Most regions do not have significant time trends in E. coli concentration (Figure 5, Table 7). The only exceptions occur for Wellington (Figure 5n), where the observed decrease over time is an artefact of using total coliform counts as a proxy for E. coli prior to 2003, and for Northland (Figure 5h) and Otago (Figure 5i), which show a decreasing percentage of sites exceeding the health standard over time, perhaps as a result of improved sampling methods in recent years that minimise the possibility of contamination.
3.3.4 Iron and manganese
Iron and manganese are considered together in this section because they tend to co-occur in groundwater (both are soluble only under oxygen-poor conditions) and hence, as indicators, they yield similar information.
The calculated national medians for Fe and Mn are 0.03 and 0.01 mg/L, respectively (Figure 1g and i, Table 3), in good agreement with previously reported values (Daughney, 2003; Daughney and Reeves, 2005; Daughney and Wall, 2007). The regions with the highest median concentrations are Manawatu-Wanganui (Fe 0.36 mg/L, Mn 0.27 mg/L), Hawke’s Bay (Fe 0.16 mg/L, Mn 0.05 mg/L) and Bay of Plenty (Fe 0.10 mg/L, Mn 0.03 mg/L). As noted previously, the SOE programmes of these regions include many sites with oxygen-poor groundwater in which Fe and Mn are soluble; the oxygen-poor conditions mean that these same regional SOE programmes also tend to have high NH4-N and low NO3-N. Conversely, regions with low median concentrations of Fe and Mn, such as Canterbury, Southland, Waikato and West Coast, have SOE programmes dominated by monitoring sites with oxygen-rich groundwater in which Fe and Mn are generally insoluble. Note that regional comparisons of Fe and Mn concentrations may be complicated by differences in sampling procedure, particularly in terms of protocols for field filtration. For example, the introduction of field filtration in the Wellington region in 2004 dramatically reduced the measured concentrations of both Fe and Mn (Figures 6n and 7n).
Nationally, 21.3% and 26.9% of the sites considered in this report have median concentrations of Fe and Mn above their respective aesthetic GVs defined in the DWSNZ (Figure 1g and i, Table 4). Overall, 9.9% of sites have median Mn concentration above the health-related MAV specified in the DWSNZ (0.4 mg/L), but only 1.4% of the sites had median Mn above the TV for ecosystem protection defined in the ANZECC guidelines (1.9 mg/L). The regions with the highest proportion of sites with median Fe and/or Mn concentration above the DWSNZ and/or ANZECC guidelines are Manawatu-Wanganui, Hawke’s Bay, Bay of Plenty, Auckland and Taranaki (Figure 1g and i).
Significant time trends in Fe and/or Mn are evident at about 15% of the sites considered in this report (Figure 1h and j, Table 5). For Mn, there are roughly equal proportions of sites showing increasing and decreasing trends, whereas for Fe, the proportion of sites with decreasing trends is roughly twice as high, which is unusual compared to other indicator parameters considered in this report and probably arises from the widespread adoption of field filtration of samples to be analysed for Fe and Mn from about 2004 onwards. Rates of change are more variable for Fe than Mn, but are generally less than 0.05 mg/L per year for both parameters at most sites (Figure 1h and j, Table 6). Nationally, there are significant year-by-year decreases in the percentage of sites with median Fe or Mn concentrations exceeding the relevant guidelines (Figure 2d and e). Certain regional trend patterns are evident (Table 7), including a year-by-year decrease in Fe and Mn percentiles in Wellington (Figure 6n and 7n), which is an artefact of the introduction of field filtration in 2004, and a year-by-year increases in Fe and Mn percentiles in Northland (Figure 6h and 7h) and Otago (Figure 6i and 7i), which may reflect modification of the regional SOE programmes from about 2000 onwards to include a greater proportion of monitoring sites with oxygen-poor groundwater.
Electrical conductivity provides a measure of the TDS concentration in a groundwater sample, and so it provides a useful indicator for spatial and/or temporal changes in abstraction, salt water intrusion, recharge mechanism, etc. The national median for electrical conductivity is 204 μS/cm, but there are substantial variations between regions (Figure 1k, Table 3). The regions with the highest median values for electrical conductivity are Northland (520 μS/cm), Manawatu-Wanganui (464 μS/cm) and Auckland (316 μS/cm). The SOE programmes of these regions include many monitoring sites that tap coastal aquifers with some level of salt water influence (Northland particularly), or deep and confined aquifers in which groundwater is known to be chemically evolved with relatively high TDS (Daughney and Wall, 2007). The lowest regional median values for electrical conductivity are for West Coast (85 μS/cm), Marlborough (151 μS/cm) and Canterbury (174 μS/cm). The low median electrical conductivity for West Coast groundwater probably reflects the diluting effect of the high regional rainfall. The low regional median electrical conductivity for Canterbury and Marlborough may reflect the importance of river seepage as a recharge mechanism, which tends to have lower TDS than recharge derived from rainfall that accumulates dissolved salts as it passes through the soil zone (Daughney and Reeves, 2005).
There are no DWSNZ or ANZECC guidelines for electrical conductivity (Table 4). However, electrical conductivity is correlated to TDS, and aesthetic guidelines exist for individual parameters such as Cl, Na and SO4. High TDS or high concentrations of the aforementioned ions are not pervasive issues in New Zealand aquifers—the relevant aesthetic GVs are exceeded at only 2-4% of all sites (Table 4).
Significant increasing and decreasing trends in electrical conductivity are detectable at 18.4% and 13.4% of monitoring sites considered in this report, respectively (Figure 1l, Table 5). Rates of change are slow at most sites, i.e. less than ±10 μS/cm per year. Not surprisingly, the median rate of change is positive for regions with a predominance of sites with increasing trends, such as Auckland, Canterbury, Marlborough, Southland and Tasman, and negative for regions with a predominance of sites with decreasing trends, such as Bay of Plenty and Otago (Figure 1l, Table 6). Trends in electrical conductivity in many regions are mirrored (in direction and relative magnitude) by trends in NO3-N. More research on these relationships may indicate that electrical conductivity and NO3-N can be used in combination as indicators of human influence.
Year-by-year changes in electrical conductivity are evident in several regions (Table 7). Annual increases in most percentile values occur in Canterbury (Figure 8c), Wellington (Figure 8n), Marlborough (Figure 8g) and Northland (Figure 8h), whereas year-by-year decreases in percentile values are apparent in Otago (Figure 8i) and Tasman (Figure 8l). In all cases, these observed year-by-year changes in the percentiles of electrical conductivity appear to be caused by changes in the set of sites comprising the SOE networks. To illustrate, note that the Tasman SOE network included ten sites from 1995 to 2001, with six additional sites added to the network from 2002 onwards. On average, the six additional sites had lower electrical conductivity than the sites in the original SOE network, and so the percentiles calculated for the set of 16 SOE sites are in general less than the percentiles determined for the original ten SOE sites.
3.4 Factors controlling groundwater quality
The aim of this section is to identify and explain statistically significant relationships between the key indicators of groundwater quality and factors of potential influence such as well depth, aquifer characteristics, or surrounding land use and land cover.
3.4.1 Well depth and aquifer confinement
Well depth and aquifer confinement must be assessed together because they are correlated: for the sites considered in this report, there is a statistically greater proportion of shallow wells (less than 10 m deep) in unconfined than confined aquifers (Daughney and Wall, 2007). There are no statistically significant relationships between rates of change in the key indicator parameters and well depth or aquifer confinement for any of the parameters considered in this report. However, the median values of key indicators are related to well depth and aquifer confinement, as reported previously (Daughney and Wall, 2007):
- NO3-N concentrations are higher for shallow wells (less than 10 m deep), especially in unconfined aquifers, but there are also many instances where high NO3-N concentrations are found in deep wells and vice versa (Figure 9);
- E. coli concentrations are most often above the detection limit (>1 cfu/100 ml) for wells less than 10 m deep, especially in unconfined aquifers, but there are also many instances where E. coli is detected in deep wells and confined aquifers (Figure 10); and
- Electrical conductivity and concentrations of NH4-N, Fe and Mn are higher for wells more than 50 m deep, especially in confined aquifers, but there are many instances where these parameters are elevated in shallow wells or unconfined aquifers (Figure 11).
The first two relationships almost certainly arise from human influence. On one hand, these relationships are to be expected, because unconfined aquifers are more susceptible to contamination, and proximity to the surface (the source of NO3-N and microbial pathogens) is logically an important influence. However, it is clear from Figures 9 and 10 that deep wells in confined and semi-confined aquifers can also be susceptible to contamination by NO3-N and/or microorganisms. Cases of microbial contamination in deep wells or confined aquifers probably reflect poor well-head protection more than the susceptibility of a particular type of aquifer (Sinton, 2001). Cases of NO3-N contamination in deep wells or confined aquifers may indicate poor well-head protection, but can also arise in certain New Zealand aquifer systems where the groundwater tends to remain oxidised for a long distance along the flow path (e.g. Canterbury), possibly due to a low concentrations of organic matter required as the substrate for microbial denitrification (Langmuir, 1997). In Canterbury, NO3-N contamination is also observed in deep wells in unconfined aquifers if the well screen is near the water table.
The third relationship probably reflects natural processes of water-rock interaction. Water-rock interaction tends to increase TDS, to which electrical conductivity is correlated (Langmuir, 1997). In addition to increases in TDS as groundwater moves along a flow path, natural processes also often (but not always) lead to the depletion of oxygen, which in turn favours the accumulation of dissolved Fe, Mn and NH4-N (Langmuir, 1997, Daughney and Reeves, 2005).
3.4.2 Aquifer lithology
There are no significant relationships between aquifer lithology and the rates of change of any the key indicator parameters considered in this report. E. coli concentrations are likewise not related to aquifer lithology. However, aquifer lithology controls the persistence of oxygen, which has an indirect relationship on the median concentrations of the indicators NO3-N, NH4-N, Fe and Mn, as observed previously (Daughney, 2003, Daughney and Wall, 2007). Specifically, the transition from oxygen-rich to oxygen-poor status in an aquifer is mediated by microbial respiration, which requires the presence of a reductant—organic carbon in most cases. In aquifer lithologies with low concentrations of organic carbon, such as ignimbrite, rhyolite, and some gravels (e.g. Canterbury), the groundwater remains oxygen-rich, and this favours the persistence of NO3-N and prevents the transformation to or accumulation of NH4-N and dissolved forms of Fe and Mn. The opposite case applies to lithologies that do contain abundant organic carbon, which is why concentrations of NH4-N, Fe and Mn are often observed to be high (and NO3-N concentrations low) in lignite, clay and some sand aquifers.
3.4.3 Surrounding land use and land cover
This report has not revealed any systematic significant relationships between land use or land cover and any of the key indicators of groundwater quality (state or trends). This result applies whether all monitoring sites are considered together, or if the data set is limited to the sites less than 10 m deep (at which the impact of land use would probably be most apparent). The lack of detectable relationship between land use and groundwater quality has been observed in several previous studies (Close et al., 1995; Reijnders et al., 1998; Broers and van der Grift, 2004; Daughney and Reeves, 2005, 2006; Daughney and Wall, 2007). Daughney and Wall (2007) stated that relationships between groundwater quality and land use are difficult to elucidate because:
- land use observations are usually made by eye and may not accurately describe land use or land use intensity;
- the groundwater at the monitoring site might not have entered the aquifer in the area where the land use observation was made;
- impacted groundwater might not have had time to travel all the way from its source area to the monitoring site; and/or
- substances indicative of land use impact (e.g. NO3-N) might have been transformed or degraded before reaching the monitoring site.