Presentation is loading. Please wait.

Presentation is loading. Please wait.

Regional assessment of water quality trends in the Wellington region

Similar presentations


Presentation on theme: "Regional assessment of water quality trends in the Wellington region"— Presentation transcript:

1 Regional assessment of water quality trends in the Wellington region
RIGHT OF REPLY OF ANTONIUS HUGH SNELDER ON BEHALF OF WELLINGTON REGIONAL COUNCIL May 2018

2 Contents Clarification of changes to statistical analyses presented the joint witness statement. Analysis of the representativeness of GWRC’s river water quality monitoring network Quantification of statistical uncertainty associated with regional water quality improvement Point 1 clarifies issues that were addressed by my primary evidence and were discussed in the joint expert statement. Points 2 and 3 were raised by Dr Canning in his primary evidence.

3 Basis for the evidence Analysis of water quality trends for rivers and lakes of the Wellington Region (Snelder, 2017b) Assessment of trends in 20 water quality measures at 62 sites in the region Analysis of regional-scale river water quality trends in the Wellington Region; period 2007 to (Snelder, 2017a) Helicopter view of regional water quality trends

4 Clarification – binomial tests
Binomial tests to establish significant improving regional trends in 6 variables for 5 and 10 years periods and no significant degrading trends In his evidence, Dr Canning raised issue of “false discovery” in association with my analysis The joint witness statement contained results that were adjusted for false discovery After adjusting for false discovery, the majority of original regional improving trends remained significant and there were no regional degrading trends Ten-year period. Before adjustment significant improving regional-trends in Clar, TP, NO3-N, NNN, and TN and Chla. There were no significant degrading regional-trends. After adjustment improving trend in Chla did not reach significance at 0.05 level but was significant when significance was relaxed to 0.1. Five year period Before adjustment significant improving regional-trends in Clar, Turbidity, NNN, and TOC and E. coli. There were no significant degrading regional-trends. After adjustment, improving regional trend in Clar, Turbidity, and E.coli when significance was relaxed to 0.1.

5 Clarification - pseudo replication
In his evidence, Dr Canning raised issue of pseudo- replication in association with my analyses I conducted sensitivity analyses to assess whether my results were affected by pseudo replication The sensitivity analysis indicated my original conclusions are not affected by pseudo replication Joint witness statement concluded: analyses are robust there is no evidence of region-wide degradation over the ten-year or five-year time periods Dr Canning suggested “pseudo replication” of monitoring sites because the monitoring network has catchments with multiple sites. Dr Canning suggested that sites located in the same catchment are influenced by the same conditions and a component of the water at downstream sites is measured at the upstream sites and that this raises the possibility that my assessment was unduly influenced by pseudo replication.

6 Additional analysis - representativeness
In the joint witness statement (paragraph 12), we agreed that GWRC’s monitoring network may not be adequately representative of the Wellington Region Here, I present results of analysis of the representativeness of GWRC’s river water quality monitoring network

7 Variable distributions
Red bars = segments of the river network Blue bars = monitoring site network The character of rivers in the Wellington region was described using six environmental variables that are strongly associated with variation in water quality (Table 1 - RoR). These environmental variables are available for each segment of a digital river network that represents all of New Zealand. The rivers of the Wellington region are represented by 18,000 segments (Snelder 2017b). Each site in the monitoring network was associated with the same six environmental variables corresponding to the segment on which the site was located. Most paired bars (red and blue) shown in the Figure (Figure 1 RoR) have reasonably closely matched values on the y-axes. This indicates a reasonable distribution of monitoring sites over the range of each of the environmental variables represented by the entire river network. There are some exceptions to this, which indicate a degree of bias in the monitoring sites’ representation of regional variation for some environmental variables. For example, Figure 1 indicates that there is a lower proportion of monitoring sites that are associated with very high values of usPastoral (blue bars) than network segments (red bars). Conversely, there are a lower proportion of monitoring sites that are associated with very low values of usIFS (blue bars) than network segments (red bars).

8 Representativeness - conclusion
There is no perfect monitoring network The above analysis shows the existing network is reasonably representative of the range of river environments in the Wellington region I conclude that my analyses are reasonably representative of river water quality in the Wellington region over the past decade

9 Additional analysis – uncertainty of ‘overall water quality trends’
I evaluated ‘overall water quality trends’ based on the proportion of sites with improving trends This indicated a dominance of improving trends (i.e., >50% of sites) for most variables I concluded this is strong evidence of overall water quality improvement at the regional level over the past decade. Statistics were derived using all trends, irrespective of the confidence in the trend direction In evidence, Dr Canning had low confidence in my conclusions because of the inclusion of ‘uncertain’ site trends This opinion was not based on an analysis of the uncertainty

10 Uncertainty of ‘overall water quality trends’
I conducted a Monte Carlo analysis to determine the uncertainty of the ‘overall water quality trends’ Analysis for 10-year period indicates 62% of all site + variable trends are improving (compared to my original 63%) 95% confidence interval is 52% - 71% Only 3 (of 18) variables have a dominance of degrading trends Therefore, high confidence that the majority of site and variable trends are improving

11 Overall conclusion In our expert statement, we agreed that there is no evidence of region-wide degradation over the ten-year or five-year time periods False discovery and pseudo replication were addressed as part of expert witnessing To address comments by Dr Canning I conducted further analyses of representativeness and uncertainty These indicate that my original conclusions are robust I do not resile from my original conclusions. There is strong evidence of water quality improvement across the region over the past decade Water quality has degraded at some sites and for some variables. However, degradation is isolated rather than occurring in a consistent and regional scale manner


Download ppt "Regional assessment of water quality trends in the Wellington region"

Similar presentations


Ads by Google