Determining the likelihood and cost of detecting reductions of nitrate-nitrogen concentrations in groundwater across New Zealand

Nitrate-nitrogen (NO₃-N) is a contaminant of concern in groundwater worldwide. Stakeholders need information on the ability to detect changes in NO₃-N concentrations to prove that land management practices are meeting water quality aims. We created a database of quarterly to monthly NO₃-N measurements in 948 sites across New Zealand; 186 of those sites had mean residence time (MRT) data. New Zealand has set a target of sufficient land use mitigations in the next 30 years to ensure steady state surface water concentrations do not exceed 2.4 mg L^-1. Here we assess whether the current monitoring network could identify the impacts of these mitigations, assuming that the mitigations are successfully implemented at the source. Only 41 % of the network could detect statistically significant reductions with the current standard quarterly sampling after 30 years of monitoring. The percentage of sites increased to 60 % with increased monitoring frequency (often weekly) but this required a 100-300 % increase in monitoring costs. However, policy makers and stakeholders typically require information on policy and mitigation effectiveness within 5-10 years. Detection within 5-10 years was very unlikely (0-20 % of sites) regardless of the sampling frequency. Importantly, these analyses include the impacts of groundwater lag and temporal dispersion on the likelihood of detecting change, ignoring these impacts, incorrectly, yields a much higher likelihood of detecting reductions. We conclude that the current monitoring network is unlikely to be fit for the purpose of detecting NO₃-N reductions within practical timeframes or budgets. Furthermore, we conclude that lag and temporal dispersion effects must be included in detection power calculations; we therefore recommend that MRT data is regularly collected. We also provide a python package to enable easy detection power calculations with lag and temporal dispersion impacts, thereby supporting the development of robust change-detection monitoring networks.