Statistical models for evaluating suspected artefacts in long-term environmental monitoring data

Long-term water quality monitoring is of high value for environmental management as well as for research. Artificial level shifts in time series due to method improvements, flaws in laboratory practices or changes in laboratory are a common limitation for analysis, which, however, are often ignored. Statistical estimation of such artefacts is complicated by the simultaneous existence of trends, seasonal variation and effects of other influencing factors, such as weather conditions. Here, we investigate the performance of generalised additive mixed models (GAMM) to simultaneously identify one or more artefacts associated with artificial level shifts, longitudinal effects related to temporal trends and seasonal variation, as well as to model the serial correlation structure of the data. In the same model, it is possible to estimate separate residual variances for different periods so as to identify if artefacts not only influence the mean level but also the dispersion of a series. Even with an appropriate statistical methodology, it is difficult to quantify artificial level shifts and make appropriate adjustments to the time series. The underlying temporal structure of the series is especially important. As long as there is no prominent underlying trend in the series, the shift estimates are rather stable and show less variation. If an artificial shift occurs during a slower downward or upward tendency, it is difficult to separate these two effects and shift estimates can be both biased and have large variation. In the case of a change in method or laboratory, we show that conducting the analyses with both methods in parallel strongly improves estimates of artefact effects on the time series, even if certain problems remain. Due to the difficulties of estimating artificial level shifts, posterior adjustment is problematic and can lead to time series that no longer can be used for trend analysis or other analysis based on the longitudinal structure of the series. Before carrying out a change in analytic method or laboratory, it should be considered if this is absolutely necessary. If changes cannot be avoided, the analysis of the two methods considered, or the two laboratories contracted, should be run in parallel for a considerable period of time so as to enable a good assessment of changes introduced to the data series.     

Beyond the Guidelines: Practical lessons for monitoring

A series of workshops have provided extensive feedback on a recently published manual, Monitoring Guidelines to Evaluate Effects of Forestry Activities on Streams in the Pacific Northwest and Alaska (Guidelines) (MacDonald et al., 1991). These workshops and other discussions have led to the identification of fourteen additional lessons for monitoring. These lessons are concepts which either were not incorporated into the Guidelines, were not sufficiently emphasized, or which are needed to put the Guidelines in context. The topics include: monitoring as a continuum; defining objectives and hypotheses; peer review; uncertainty and risk; upslope vs. instream monitoring; photo sequences; scale considerations; data storage, data interpretation, and data base management; activities monitoring; and personal commitment as a critical component in monitoring projects. Many of these lessons might appear self-evident, but our experience indicates that they are often ignored. Like the Guidelines, these lessons are widely applicable and should be explicitly recognized when formulating and conducting monitoring projects.     

Assessing the toxic threat of selenium to fish and aquatic birds

A procedure is given for evaluating the toxic threat of selenium to fish and wildlife. Toxic threat is expressed as hazard, and is based on the potential for food-chain bioaccumulation and reproductive impairment in fish and aquatic birds, which are the most sensitive biological responses for estimating ecosystem-level impacts of selenium contamination. Five degrees of hazard are possible depending on the expected environmental concentrations of selenium, exposure of fish and aquatic birds to toxic concentrations, and resultant potential for reproductive impairment. The degree of hazard is given a numerical score: 5 = high hazard, 4 = moderate hazard, 3 = low hazard, 2 = minimal hazard, and 1 = no identifiable hazard. A separate hazard score is given to each of five ecosystem components; water, sediments, benthic macroinvertebrates, fish eggs, and aquatic bird eggs. A final hazard characterization is determined by adding individual scores and comparing the total to the following evaluation criteria: 5 = no hazard, 6–8 = minimal hazard, 9–11 = low hazard, 12–15 = moderate hazard, 16–25 = high hazard. An example is given to illustrate how the procedure is applied to selenium data from a typical contaminant monitoring program.     

Application of a risk assessment based approach to designing ambient air quality monitoring networks for evaluating non-cancer health impacts

An ambient air quality monitoring network has been establishedusing risk assessment techniques to evaluate adverse health effects from exposures to airborne contaminants. The risk assessment method was compared to traditional methods of establishing air quality monitoring networks: identifying maximumconcentration impacts or maximum total population. Results suggest that the health risk method best predicted the locationof adverse, non-carcinogenic respiratory illnesses during the evaluation period. Spearman Rank Correlation Coefficient, r _s, values obtained using the risk assessment method werestatistically greater than the values obtained using theconcentration and population methods. The concentration methodwas the least accurate predictor of adverse effects.     

Risk assessment near uncontrolled hazardous waste sites: Role of monitoring data

The traditional approach of coupling estimates of human exposures to individual chemicals with laboratory studies of the toxicity of the chemicals as the basis for quantitative assessments of risk is not adequate when considering problems near hazardous waste sites. For example, there are (a) too many chemicals and mixtures involved, (b) containment uncertainties, and (c) future exposure problems associated with groundwater and soil contamination. Among the items to be considered in expanding the dimensions of risk assessment methodologies of the past are (a) identification of chemical groups or of dominant toxic chemicals of principal concern and the likely man-made as well as natural pathways for environmental migration, (b) the role of biological monitoring in addition to traditional monitoring approaches, and (c) the coupling of monitoring data with past, current, and future population activity patterns and with epidemiological and other health data.Monitoring data is a key in risk assessments since there is little likelihood that materials balances or modelling will provide authoritative information concerning exposure levels. The use of monitoring data from control areas and from nationwide baseline surveys in developing comparative risk assessments is particularly important. Finally, recent experience provides us with a number of practical guidelines for designing and carrying out monitoring programs that will provide authoritative and useful data. *** DIRECT SUPPORT *** AZ802019 00002     

An approach to the assessment of the quality of environmental monitoring data

This paper reports an approach to the assessment of the validity of environmental monitoring data--a 'data filter'. The strategy has been developed through the UK National Marine Analytical Quality Control (AQC) Scheme for application to data collected for the UK National Marine Monitoring Plan, although the principles described are applicable more widely. The proposed data filter is divided into three components: Part A, 'QA/QC'--an assessment of the laboratory's practices in Quality Assurance/Quality Control; Part B, 'fitness for purpose'--an evaluation of the standard of accuracy that can be demonstrated by activities in (A), in relation to the intended application of the data; and Part C, the overall assessment on which data will be accepted as usable or rejected as being of suspect quality. A pilot application of the proposed approach is reported. The approach described in this paper is intended to formalise the assessment of environmental monitoring data for fitness for a chosen purpose. The issues important to fitness for purpose are discussed and assigned a relative priority order on which to judge the reliability/usefulness of monitoring data.     

Integrated exposure monitoring for health risk assessment

Risk estimation and risk evaluation in society are often done separately for pollutants in ambient air, water, food and industrial air. The need for evaluating total exposure is pointed out. Ongoing international programs are referred to. Biological monitoring in combination with environmental monitoring constitutes a promising step towards an integrated exposure monitoring for health risk assessment. The rationale for such an approach is discussed. The need of monitoring not only single substances but a combination of substances is exemplified. A strengthening of the basic research in related fields will widely increase future possibilities for health-related surveillance.     

Retrospective monitoring of synthetic musk compounds in aquatic biota from German rivers and coastal areas

The polycyclic musk compounds HHCB (Galaxolide) and AHTN (Tonalide) are commonly used as synthetic fragrances in personal care products and household cleaners. These and other synthetic musk fragrances were quantified in different aquatic samples from the German Environmental Specimen Bank (ESB). While HHCB and AHTN were found in almost all samples, most of the other musk fragrances were detected only in a few samples and mostly at lower concentration levels. Blue mussels from the North Sea showed varying levels of 0.5-1.7 ng g(-1) ww for HHCB and 0.4-2.5 ng g(-1) ww for AHTN (ww, wet weight) in the period from 1986 to 2000, while blue mussels from the Baltic Sea were only slightly contaminated with synthetic musk fragrances. Lipid weight-related concentrations of synthetic musk compounds in blue mussels were higher than in eelpout muscles, bladder wrack and herring gull eggs. In comparison to the marine specimens, muscles of bream from German rivers had higher concentrations of HHCB and AHTN. The ranges of HHCB and AHTN concentrations in bream from the Elbe River were 545-6400 ng g(-1) lw and 48-2130 ng g(-1) lw, respectively (lw, lipid weight; five sampling sites, period 1993-2003). In the Rhine River, HHCB and AHTN levels of bream muscles were highest at the Iffezheim site (up to 9750 ng g(-1) lw HHCB, 1998). Even higher synthetic musk levels were detected in bream from the rivers Saale and Saar. In recent years, levels of both compounds determined in bream from most sampling sites have decreased from maximum values in the 1990s. As the concentrations of AHTN have decreased faster, the ratio of HHCB to AHTN increased from 2-4 in the 1990s to 10-20 in recent years.     

Statistical and nonstatistical considerations for environmental monitoring studies

In environmental studies statistics is too often used as a salvage operation, or as an attempt to show significance in the absence of any clear hypothesis. Good design is needed, not fancier statistics. Too often we pursue short-term problems that are in fashion rather than study long-term environmental deterioration that really matters. Since change-often unpredictable change-is an intrinsic part of nature, it is pointless to fight all environmental change. We must choose our level of concern and then influence environmental change where we can. The judgement on whether a given change is bad cannot be left to the statistician or to statistical tests; the politician in consultation with the ecologist are responsible for it. The statistical significance of a hypothesized impact-related change should be tested against year-to-year variation in the unimpacted situation rather than against replicate sampling error. This is another argument for long-term studies. Attributes of good design and appropriate criterion and predictor variables are discussed.Paper presented at a Symposium held on 20–21 April 1982, in Edmonton, Alberta, Canada.     

Framework for enhancing the statistical design of aquatic environmental studies

Aquatic environmental studies can be categorized by the breadth of their scope and the types of desired results. The use of this categorization coupled with a clear specification of objectives and a judicious knowledge of the environmental variability should lead to more statistically efficient studies. This paper discusses the types of lacustrine studies commonly encountered in terms of their categorization. It provides examples of how the intrinsic environmental variability can influence their design and stresses the importance of properly stated objectives, the developing of testable hypotheses, the design of robust and powerful studies, and the importance of evaluating the implication of changes as critical factors for conducting effective and efficient environmental studies.An early version of this paper was presented at the First International Conference on Environmentrics, Cairo, Egypt, April 4–7, 1989, under the title Framework for the Design of Aquatic Environmental Studies.     

Integrating probabalistic and targeted compliance monitoring for comprehensive watershed assessment

Environmental monitoring typically falls into one of two broad categories. Targeted designs, utilizing fixed stations, focus on describing and quantifying impacts, tracking trends, and assessing compliance with regulatory guidelines or limits. Probabilistic designs, in contrast, draw sampling stations at random from an area or region, and the stations are used to describe conditions in the region of interest based on a subpopulation of sites. These two design approaches are usually viewed as mutually exclusive, with randomized designs used for broader regional assessments of overall ambient condition and targeted designs for demonstrating regulatory compliance and/or characterizing specific, localized impacts. Combining elements of both approaches into a single design provides benefits not available from either design alone. Embedding targeted monitoring within the framework of a probabilistic design enables data from targeted stations to be viewed in a more accurate regional context and provides a consistent background against which to identify characteristic regional patterns of contamination and impact. We use the San Gabriel River Regional Monitoring Program, recently implemented in southern California, to illustrate the structure of a hybrid design and how it enables data analyses and assessments that provide a more complete picture of conditions in the watershed. For example, the hybrid design showed that approximately 80% of the metals levels at compliance sites were below the 25th percentile of the overall watershed condition as indicated by the probabilistic sampling.     

Cumulative effect assessment in Canada: a regional framework for aquatic ecosystems

Sustainable development of the aquatic environment depends upon routine and defensible cumulative effects assessment (CEA). CEA is the process of predicting the consequences of development relative to an assessment of existing environmental quality. Theoretically, it provides an on-going mechanism to evaluate if levels of development exceed the environment's assimilative capacity; i.e., its ability to sustain itself. In practice, the link between CEA and sustainable development has not been realized because CEA concepts and methods have developed along two dichotomous tracks. One track views CEA as an extension of the environmental assessment (EA) process for project developments. Under this track, stressor-based (S-B) methods have been developed where the emphasis is on local, project-related stressors, their link with aquatic indicators, and the potential for environmental effects through stressor-indicator interactions. S-B methods focus on the proposed development and prediction of project-related effects. They lack a mechanism to quantify existing aquatic quality especially at scales broader than an isolated development. This limitation results in the prediction of potential effects relative to a poorly defined baseline state. The other track views CEA as a broader, regional assessment tool where effects-based (E-B) methods specialize in quantification of existing aquatic effects over broad spatial scales. However, the predictive capabilities of E-B methods are limited because they are retrospective, i.e., the stressor causing the effect is identified after the effect has been measured. When used in isolation, S-B and E-B methods do not address CEA in the context necessary for sustainable development. However, if the strengths of these approaches were integrated into a holistic framework for CEA, an operational mechanism would exist to better monitor and assess sustainable development of our aquatic resources. This paper reviews the existing conceptual basis of CEA in Canada including existing methodologies, limitations and strengths. A conceptual framework for integrating project-based and regional-based CEA is presented.     

A comparative assessment of molecular biological and direct microscopic techniques for assessing aquatic systems

Molecular techniques have clear value for community characterization; however, almost all previous datasets are based upon non-molecular measurements and it is hard to compare "old" data with "new" data because few correlations have been made. Therefore, the purpose of this evaluation was to simultaneously use molecular and non-molecular methods within the same sampling program to determine how data compare. Three methods were used for characterizing microbial populations in Lake 260 (L260) at Experimental Lakes Area (Ontario, Canada) during a whole-lake exposure study. Methods included whole-cell microscopic counts (for bacteria, cyanobacteria, algae and zooplankton), chlorophyll a, and small sub-unit (ssu)-rRNA hybridization using EUB-338, EUB-785, CYAN-785-a/b, EUCA-1379 and UNIV-1390 gene probes. Strong correlations were found between the EUB-785 probe signal and "bacteria minus cyanobacteria" direct counts, and the EUB-338 probe signal and "bacteria plus cyanobacteria" counts. Furthermore, the difference in probe signal between EUB-338 and EUB-785 (a presumptive signature for cyanobacteria and plastids) correlated with cyanobacterial direct counts and also with chlorophyll a. However, EUCA-1379 probe signal did not correlate with algae counts, and UNIV-1390 probe signal only correlated with total bacteria counts. Results suggest that, although ssu-rRNA methods are fast, reproducible, and specifically detect "viable" organisms, their use may be limited to non-eukaryotic populations unless new probes are developed that are more specific.     

River habitat monitoring and assessment in Germany

Whereas assessing the biological and chemical quality of water is a standard environmental procedure in many countries, the use of habitat survey methods that assess the ecomorphological quality of rivers is relatively new. In Europe, the EC Water Framework Directive requires such assessment from all EU Member States. In Germany, the first river habitat assessments were introduced in the late 1990 s. Each federal state develops its own river habitat map using the 'On-site Survey' and/or the 'Overview Survey'. The assessment describes the difference of the actual condition from a previously defined reference condition. In practice, a defined 'potential for restoration', a more realistic condition, makes restoration activities much easier and more successful. In Germany, the first River Habitat Map 2001 was published in 2002. The survey covered 33,000 km of river length, which equates to 10% of all rivers. A wide range from 'Undisturbed' (class 1) to 'Totally Disturbed' (class 7) river units exists; 77% of them are 'Clearly Disturbed' (class 4) or in worse condition. These result reflects extensive anthropogenic impact on the environment in general, but also past intense technical river 'improvements' that focused on the protection of settlements and traffic routes from flooding, better shipping conditions, the use of water power, and drainage of floodplains for agriculture and urban development. For comparability of survey results between EU Member States, a harmonization of national survey methods is in progress. A crucial point here is the definition of the reference condition for each river (near-natural conditions), since it influences the survey results.     

A proposed coast-wide reference monitoring system for evaluating wetland restoration trajectories in Louisiana

Wetland restoration efforts conducted in Louisiana under the Coastal Wetlands Planning, Protection and Restoration Act require monitoring the effectiveness of individual projects as well as monitoring the cumulative effects of all projects in restoring, creating, enhancing, and protecting the coastal landscape. The effectiveness of the traditional paired-reference monitoring approach in Louisiana has been limited because of difficulty in finding comparable reference sites. A multiple reference approach is proposed that uses aspects of hydrogeomorphic functional assessments and probabilistic sampling. This approach includes a suite of sites that encompass the range of ecological condition for each stratum, with projects placed on a continuum of conditions found for that stratum. Trajectories in reference sites through time are then compared with project trajectories through time. Plant community zonation complicated selection of indicators, strata, and sample size. The approach proposed could serve as a model for evaluating wetland ecosystems.     

Comparison of data from two vegetation monitoring methods in semi-natural grasslands

Two vegetation-monitoring methods were compared: subplot frequency analysis (SF) and visual estimation of percentage cover (VE). Two independent observers collected data from two semi-natural, species-rich grasslands on three different occasions during the growth-season. During the last data collection period, survey times were also recorded. The two different data sets from the two methods were compared using partial Redundancy Analyses. The purpose of the comparison was to identify the method that explains most of the relevant variation in biodiversity-monitoring (inter-and intra-site variation), and the variation irrelevant when evaluating data (systematic inter-observer variation and variation due to phenological changes). Compared with VE data, more variation in SF data could be explained by spatial variables, while less variation depended on the observer and time of year surveyed. SF also found more species per plot but took on average five times longer to complete than VE. In conclusion, the different methods are suitable for different purposes: SF is more suitable for purposes demanding high accuracy and high precision, such as long-term biodiversity-monitoring when the identification of small changes has high priority, while VE might be more suitable for a one-time mapping of a large area.     

Early-season agricultural drought: detection, assessment and monitoring using Shortwave Angle and Slope Index (SASI) data

Early season or crop-planting-period (ES/CPP) drought conditions have become a recurrent phenomenon in tropical countries like India, due to fluctuations in the time of onset and progression of monsoon rains. ES/CPP agricultural drought assessment is a major challenge because of the difficulties in the generation of operational products on soil moisture at larger scales. The present study analyzed the Shortwave Angle Slope Index (SASI) derived from Near Infrared and Shortwave Infrared data of Moderate Resolution Imaging Spectroradiometer, for tracking surface moisture changes and assessing the agricultural drought conditions during ES/CPP, over Andhra Pradesh state, India. It was found that in-season progression of SASI was well correlated with rainfall and crop planting patterns in different districts of the study area state in both drought and normal years. Rainfall occurrence, increase in crop planted area, and decrease in SASI were in chronological synchronization in the season. Change in SASI from positive to negative values is a unique indication of dryness to wetness shift in the season. Duration of positive SASI values indicated the persistence of agricultural drought in the crop planting period. Mean SASI values were able to discriminate an area which was planted in normal year and unplanted in drought year. SASI thresholds provide an approximate and rapid estimate of the crop planting favorable area in a region which is useful to assess the impact of drought. Thus, SASI is a potential index to strengthen the existing operational drought monitoring systems. Further work needs to be on the integration of multiple parameters—SASI, soil texture, soil depth, rainfall and cropping pattern, to evolve a geospatial product on crop planting favorable areas. Such products pave the way for quantification of drought impact on agriculture in the early part of the season, which is a major inadequacy in the current drought monitoring system.     

Re-construction of the shut-down PM10 monitoring stations for the reliable assessment of PM10 in Berlin using fuzzy modelling and data transformation

A dense monitoring network is vital for the reliable assessment of PM10 in different parts of an urban area. In this study, a new idea is employed for the re-construction of the 20 shut-down PM10 monitoring stations of Berlin. It endeavours to find the non-linear relationship between the hourly PM10 concentration of both the still operating and the shut-down PM10 monitoring stations by using a fuzzy modelling technique, called modified active learning method (MALM). In addition, the simulations were performed by using not only raw PM10 databases but also log-transformed PM10 databases for skewness reduction. According to the results of hourly PM10 simulation (root mean square error about 13.0 μg/m³, correlation coefficient 0.88), the shut-down stations have been appropriately simulated and the idea of dense monitoring network development by the re-construction of the shut-down stations was realised. The results of simulations using raw and log-transformed databases showed that data transformation has no significant effect on the performance of MALM in the simulation of shut-down PM10 stations. By the combination of the 11 still operating stations and the 20 re-constructed stations, a dense monitoring network was generated for Berlin and was utilised for the calculation of the reliable monthly and mean annual PM10 concentration for five different PM10 zones in Berlin (the suburban-background, urban-background, urban-traffic, rural-background and suburban-traffic areas). The results showed that the mean annual concentration of PM10 at the five zones increased by about 13.0% in 2014 (26.3 μg/m³) in comparison with 2013 (23.3 μg/m³). Furthermore, the mean annual concentration of PM10 in the traffic lanes of the suburban (2013 25.0 μg/m³, 2014 26.9 μg/m³) and urban (2013 27.7 μg/m³, 2014 31.3 μg/m³) areas is about 14 and 20% higher than the PM10 concentration of suburban-background (2013 21.3 μg/m³, 2014 24.5 μg/m³) and urban-background (2013 23.0 μg/m³, 2014 26.1 μg/m³) areas, respectively.     

Interpretation of roadside PM10 monitoring data from Sunderland,United Kingdom

Roadside PM₁₀ has been monitored by Partisol® at three sitesin Sunderland between August 1997 and February 1998. The sites chosen were an inner city kerbside site; a roadside site adjacentto a dual carriageway on the outskirts of Sunderland with an openaspect; and a rural site.The results indicate that there is a seasonal variation in the relationship between the sites in terms of monitored PM₁₀.In the winter there is a poor correlation between the sites whereas in the summer significant correlations are obtained. Of the sites monitored PM₁₀ is consistently highest at the inner city roadside site. During the summer, exceedances of theU.K. 50 g m^-3 standard (DETR, 2000) are associated with conditions suitable for the build-up of photochemical pollutionhowever during the winter period exceedances are recorded duringa variety of weather conditions.At the dual carriageway site PM_2.5 has also been recorded and contributions to measured PM₁₀ are 77% in summer and68% in winter. The results illustrate a number of inconsistencies between this study utilising the Partisol® andothers reporting results where PM₁₀ has been monitored by TEOM®.