Four sessions on “Monitoring Strategies: temporal trends in groundwater and
surface water quality and quantity” at the EGU conferences in 2012, 2013,
2014, and 2015 and a special issue of
The presence and dynamic behaviour of nutrients in groundwater and surface water is an important issue in water management, in particular in areas with intensive agriculture. This is, for example, reflected in EU directives such as the Nitrates Directive (EU, 1991), the Water Framework Directive (WFD; EU, 2000), the Groundwater Directive (GWD; EU, 2006), and the Monitoring Directive (EU, 2009). Member states are obliged to monitor and report on the environmental status of the water bodies and, if necessary, take measures to establish adverse trend reversal. As far as nutrients are concerned, the European directives focus on aquatic ecosystems and groundwater-dependent ecosystems. In order to meet the obligations, monitoring programmes have to cover a range of water quantity, water quality, and ecological parameters, and an understanding of dynamic nutrient processes is required for these programmes to be efficient and cost-effective. However, the design of monitoring strategies is often hampered by limited knowledge of, for instance, nutrient responses to weather conditions, land use, and agricultural practices. Moreover, the behaviour of nutrients shows large variability in both space and time (see, e.g., Campbell et al., 2015; Goyenola et al., 2015).
To satisfy the increasing demand for knowledge and information on the
dynamic behaviour of nutrients, the past 10–15 years have seen a rapid
development of observation devices and technologies for high-resolution
monitoring of nutrients and other solutes and isotopes at affordable cost,
encouraging researchers, and other stakeholders to perform studies in
experimental as well as operational settings. Thus, vast amounts of research
data have been collected on various water quality variables, allowing the
study of relevant biogeochemical processes and enabling comparisons between
the results obtained by the use of different monitoring devices. Thus,
awareness has increased about the advantage of using high-resolution
nutrient monitoring as a complementary tool next to traditional low-frequency monitoring. The sessions on “Monitoring Strategies: temporal
trends in groundwater and surface water quality and quantity” at the
EGU conferences in 2012, 2013, 2014, and 2015 clearly showed that high-frequency monitoring and strategies for nutrient monitoring are subjects
that attract great interest. Part of the work presented at these sessions is
now gathered in the 10 papers included in this special issue of What does the new monitoring technology have to offer and how can we develop
an optimal monitoring strategy? Can we assess and quantify the transport processes of nutrients, in
particular at a short timescale? How can we use high-frequency nutrient monitoring to achieve our management goals?
An overview of monitoring objectives and timescales for high-frequency
nutrient monitoring is given in Table 1. We distinguished between three main
groups of monitoring objectives:
to improve our understanding of the underlying hydrological, chemical, and
biological processes determining temporal and spatial patterns in nutrients
(process understanding: PU); to quantify nutrient loads and concentrations (quantification: Q); to support operational water and environmental management,
including evaluation of the effects of mitigation measures and predictions
(management: M).
It should be noted that some papers address more than one of these overall objectives.
Kirchner et al. (2004) addressed the new opportunities of high-resolution
monitoring for understanding the functioning of catchments, and they foresaw
a new era of technical progress and study of actual data, making full profit
of the newly acquired spectrum of signals from very short to longer timescales. A decade later, a large number of papers and presentations,
including those at the EGU sessions, have demonstrated that process
understanding has indeed improved significantly. We have made a subdivision
of the monitoring objectives focusing on process understanding (PU):
Overview of monitoring objectives and timescales for high-frequency nutrient monitoring.
Continued.
Quantification (Q) type monitoring objectives focus not on identifying and
understanding the processes but on the quantification of specified
quantities, such as averages, probabilities, and proportions of exceedance of
water quality standards. Typically, such objectives relate to policy
development and operational management, in particular relative to
EU directives such as the EU Nitrates Directive (EU, 1991) and the Water
Framework Directive (EU, 2000). Q type objectives are divided into five categories:
The central aim of the management (M) type monitoring objectives is an
evaluation of the impact of water and environmental management measures as
well as climate change on nutrient transport. M type objectives typically
involve the reaction of the catchment to man-made or natural changes of
nutrient sources, and the hydrological functioning or the biogeochemistry of
the system. We have defined three subgroups:
The scale at which information is required is termed “information scale”.
Information scale is important when designing monitoring systems and
choosing the methods and goals for data processing (Broers, 2002; Van Geer et
al., 2006). For instance, selection of monitoring equipment and choice of
methods for data smoothing require a properly defined information scale, and
the papers and abstracts are therefore grouped according to this (Table 1).
For each monitoring objective, the required information depends on the scale
at which the information is needed. The following three temporal scales are considered:
short-scale dynamics and extreme events (minutes to weeks); seasonal and annual patterns (months to several years); longer-term behaviour and trends (years to decades).
Specific monitoring objectives may require a specific information scale.
This we illustrate for the monitoring objective “characterizing groundwater
surface water interaction”. Typically, analysis of the response of nitrate
concentrations in surface water to rainfall events is of short temporal
scale (minutes or hours). To estimate average loads from shallow groundwater
towards surface water during the growing season, the information scale
required will involve one or several seasons. To evaluate the long-term
sustainability of groundwater-dependent aquatic ecosystems in a WFD
assessment, the information scale may cover several years or decades.
Irrespective of the timescale of the monitoring objective, observations contain variations at all timescales and the gathered data have to be processed and statistically filtered in order to obtain the correct trend information or system characteristics at the desired timescale (e.g. Lloyd et al., 2014).
Obviously, to obtain information at short timescales, high-frequent monitoring is required and data processing will include high-pass filters. Concentrations and loads of nutrients frequently show rapid changes over time as a result of rainfall events, emissions of effluents from point sources and unintended losses of manure or pesticides during application. Often, these rapid changes occur at timescales of less than 1 h and high-frequency monitoring is required in order to capture peaks and extreme values that would go undetected if applying only low-frequency monitoring (cf. Campbell et al., 2015; Skeffington et al., 2015; Van der Grift et al., 2016).
Also, if assessing the statistical characteristics of the concentration or the load of a solute (e.g. average and percentile values or the frequency of exceedance of a threshold), high-frequency monitoring is a valuable tool. In principle, statistical characteristics can be determined from low-frequency observations provided that the monitoring period is sufficiently long. However, in many cases the system shows statistically non-stationary behaviour over longer periods of time due to, for example, changes in land use management. High-frequency monitoring enables the estimation of trend characteristics in shorter periods, being less sensible for longer-term trends (e.g. Lloyd et al., 2014). Many studies focus on the interactions between groundwater and surface water, in particular the different flow paths of nutrients towards the surface water (cf. Poulsen et al., 2015b; Rozemeijer et al., 2010b). The weather conditions appear to be the major driving force for the temporal distribution of fluxes along the different flow paths, including quick components like discharges from point sources, tile drain water, and overland flow and slow components such as discharges from deeper groundwater. The quick components have response times of the order of magnitude of hours, days, or weeks. Therefore, the response of nutrient fluxes and loads to precipitation is a complex function (e.g. Van der Velde et al., 2010). To estimate this complex response function and to unravel the contributions of the different flow paths, high-frequency monitoring is a prerequisite (cf. Campbell et al., 2015).
An example of an objective with a seasonal information scale is the estimation of average or typical nutrient concentrations during the growing season. An example of a long-term monitoring objective is found in the WFD, which includes elucidating the trends in water quality status towards the 2027 compliance with good chemical status and meeting the environmental objectives for aquatic and terrestrial ecosystems (cf. Rozemeijer et al., 2014; Erntsen et al., 2015; Skeffington et al., 2015). As with groundwater, an equivalent timescale is required for demonstrating the trend reversal in concentrations of nitrate (Visser et al., 2007). Although high-frequency information (days to weeks) is not required for the analysis of seasonal and annual patterns and long-term behaviour, high-frequency monitoring can be beneficial, because often statistical characteristics and input-response relations can be inferred reliable from a shorter monitoring period. Individual observations of water quality are the result of variation at a wide range of frequencies. High-frequency variations (noise) tend to obscure the low-frequency signal. High-frequency monitoring enables filtering out the noise (low-pass filter) during relatively short monitoring periods in order to elucidate the long-term trend (Bierkens et al., 1999; Halliday et al., 2012; Aubert et al., 2013; Lloyd et al., 2014; Van der Grift et al., 2016).
Overview of monitoring methods and instruments applied in the session abstracts and special issue papers.
Several types of sensors have been developed in recent years. Some are based on in situ laboratory (mobile or stationary) analysis of water samples, while others utilize, for instance, light or infrared (UV) spectra to measure chemical parameters (e.g. turbidity, nitrate, DOM) or materials capable of passive adsorption of chemicals (e.g. SorbiCells). Some sampling methods produce point observations in time, whereas others derive flow- or time-weighted concentrations over a time period. A number of studies (e.g. Rozemeijer et al., 2010c; Cassidy and Jordan, 2011; Jordan et al., 2013; Huebsch et al., 2015) compare several sampling instruments and monitoring strategies (Table 2). Various continuous monitoring methods, in particular those described in the papers presented in this special issue, are listed in Table 2.
Based on the observations and findings described at the five EGU sessions together with the 10 papers included in the present special issue, some general conclusions can be drawn.
Several research groups in Europe and beyond are undertaking pilot studies on the use of high-frequency monitoring of nutrients. During the past decades, there has been growing awareness of the fact that the quality of the aquatic environment is threatened by high concentrations and loads of nutrients in groundwater and surface water. At the same time, development of observation equipment enabling high-frequency monitoring at affordable cost has been extensive and, accordingly, assessment and quantification of the dynamic behaviour of nutrients at very small timescales (minutes to hours) are now feasible. Most testing has been devoted to process understanding (PU) and quantification of concentrations and loads (Q) (Table 1). Quantification of concentrations and loads to be used in the status assessments required by the EU Water Framework Directive has received much attention by several European research groups during the last five years. However, only few papers and contributions cover aspects of the monitoring effects of river basin management plans that have been implemented to reduce pollution by nutrients or climate change impacts. Although full-scale application of high-frequency monitoring at national or regional scale may not always be reported in scientific papers, we believe that its use in operational water management is still limited. The papers listed in Table 1 show that different monitoring methods have been successfully implemented and tested, and it is a step forward towards implementation of these kinds of applications in national or regional monitoring programmes in the coming years.
Some papers present comparisons between different observation methods and equipment, and others discuss the technical issues related to the observation devices, and it appears that sensors and other equipment have measurement errors differing from those of traditional laboratory analyses. This may, for example, be due to the required regular calibration and the often high maintenance effort of equipment.
High-frequency monitoring produces time series that enable us to unravel the transport processes of nutrients, for example the contribution of different flow routes or the ratio between statistically stationary fluctuations and structural trends. The fast-growing amount of data requires development of new analysis techniques to handle the large data sets. The error statistics of the new equipment in combination with the large amount of data require also new techniques for QA/QC.
Research into high-frequency nutrient monitoring will continue. Here, we focus on the development expected for the near future.
Today, high-frequency monitoring of nutrients is subject to research and pilot studies, but we expect a transition from research to implementation in operational practice. This transition requires the design of efficient and cost-effective monitoring programmes, for which research is needed to identify the best combination of observation devices and how to best integrate the data from these devices with dynamic models describing the evolution of nutrients in time and space. Well-defined monitoring objectives are a prerequisite for optimum monitoring strategies (observation devices, spatial and temporal distribution).
High-frequency monitoring will become part of the routine workflow of agencies within groundwater and surface water quality management, and vast amounts of data will be generated. Often long time series are necessary, for example to assess trends over longer periods of time. Therefore, a robust system for data storage, QA/QC and easy access data availability is of great importance (e.g. Neal et al., 2011). Today, data processing (e.g. to assess trends) is hampered by the short duration of the time series. However, with increasing availability of long time series, application of advanced statistical time-series analysis methods becomes feasible (Lloyd et al., 2014). We expect that more research will be conducted into the application of statistically based techniques, such as transfer function–noise models, to deduce the characteristics of the series and to quantify the relationship with other hydrological variables (e.g. Van der Grift et al., 2016). Examples of characteristics may be typical seasonal behaviour, the memory of the system, and the trend. Examples of relationships are the response of nutrients to meteorological variables or to water management. Such time-series analysis techniques will have applications in studying the effects of climate change on the functioning of catchments, e.g. by elucidating the changing response times of water and solutes towards precipitation and drought events.
High-frequency data will in the future assist in achieving a better understanding about in-stream processes such as nitrogen and phosphorus assimilation, sedimentation, and resuspension processes. Moreover, water quality models will be challenged when calibrated against high-frequency data, which in turn will force models to be more dynamic (run at lower time steps) and improve their internal process descriptions.
High-frequency monitoring data will also be able to assist water managers in getting a true picture of nutrient loadings and sources that will enable river basin managers to implement more targeted and thereby cost-effective decisions when fulfilling the requirement under the EU directives directed at water management such as the Water Framework Directive, the Nitrates Directive, and the Groundwater Directive.
The future will likely see more emphasis on multi-variable analysis, in
which monitoring set-up, data collection, and data processing are not made
for one variable at a time but within a multi-variate framework. Such a
framework can include the dynamic modelling of travel times, the age dating
of contributing flow routes (e.g. Gilmore et al., 2016), and the inclusion of
other tracers of flow processes that can be monitored at high resolution,
including isotopes of water (
Future research into observation devices will probably concentrate on the combination of different types of high-frequency sensors to improve our knowledge of biogeochemical processes, such as nitrate attenuation processes and phosphorus retention, in groundwater and surface waters. Development of equipment (sensors) will likely continue in the coming years, in particular to create cost-effective, more precise, more robust, and more low-maintenance monitoring devices.
The work is a contribution to the BufferTech project under the Innovation Foundation in Denmark (grant no. 1305-00017B) and supported by the Strategic Research Funding of TNO. Edited by: M. Giudici Reviewed by: three anonymous referees