Florentina Moatar

Particulate organic carbon in the estuarine turbidity maxima of the Gironde, Loire and Seine estuaries: origin and lability.

A study of the particulate organic carbon (POC) in the estuarine turbidity maxima (ETMs) of the three major French macrotidal estuaries shows that the average contents are 1.5, 3.3 and 3.1% (expressed in % of dry suspended sediment) in the Gironde, Loire and Seine Estuaries, respectively. There is no seasonal variation of POC contents in the Gironde Estuary, whereas, they often increase in the Loire and the Seine Estuaries in spring and summer. The lability of the estuarine particulate organic matter was estimated by two analyses: 1/labile organic matter was measured as the organic carbon loss during incubation tests over one month; 2/ the hydrolysable organic fraction was determined after 6N HCl digestion. The organic fractions of the ETMs are mainly refractory. Any increase in the amount of POC as compared to the background levels (cited above) is always correlated to an increase of organic matter lability. The yearly average fluvial contributions by various particulate organic pools (soil and litter organic matter; organic matter of phytoplanktonic and human origin) that enter the three estuaries were quantified. In the Garonne River, soil and litter are the major (90%) POC sources. In the Loire system, due to the eutrophication of the river water, phytoplankton contributes up to 50% of the total POC load. In the Seine river, soil and litter contribute 70% of the total POC input; POC of human origin is also significant (10%), due to the impact of the City of Paris (10 million inhabitants). The lability of the different types of organic matter ranks as follows: phytoplankton ∼litter > human-origin organic matter >> soil. By combining the POC budgets and the lability of each type of organic fraction, it was possible to explain why the POC of the three ETMs is different and characterizes its refractory vs. labile nature.

Uncertainties in assessing annual nitrate loads and concentration indicators: Part 1. Impact of sampling frequency and load estimation algorithms.

The objectives of this study are to evaluate the uncertainty in annual nitrate loads and concentrations (such as annual average and median concentrations) as induced by infrequent sampling and by the algorithms used to compute fluxes. A total of 50 watershed-years of hourly to daily flow and concentration data gathered from nine watersheds (5 to 252 km²) in Brittany, France, were analyzed. Original (high frequency) nitrate concentration and flow data were numerically sampled to simulate common sampling frequencies. Annual fluxes and concentration indicators calculated from the simulated samples were compared to the reference values calculated from the high-frequency data. The uncertainties contributed by several algorithms used to calculate annual fluxes were also quantified. In all cases, uncertainty increased as sampling intervals increased. Results showed that all the tested algorithms that do not use continuous flow data to compute nitrate fluxes introduced considerable uncertainty. The flow-weighted average concentration ratio method was found to perform best across the 50 annual datasets. Analysis of the bias values suggests that the 90th and 95th percentiles and the maximum concentration values tend to be systematically underestimated in the long term, but the load estimates (using the chosen algorithm) and the average and median concentrations were relatively unbiased. Great variability in the precision of the load estimation algorithms was observed, both between watersheds of different sizes and between years for a particular watershed. This has prevented definitive uncertainty predictions for nitrate loads and concentrations in this preliminary work, but suggests that hydrologic factors, such as the watershed hydrological reactivity, could be a key factor in predicting uncertainty levels.

pH modelling by neural networks. Application of control and validation data series in the Middle Loire river

Artificial neural networks (ANNs) are applied as a new type of model to estimate the daily pH of the Middle Loire river. The model is used for pH measurement screening, error detection (abnormal values, discontinuities and recording drifts) and validating the collected data. The measured values of pH are compared with the values estimated by the ANN model using statistical tests to verify homogeneity and stationarity. River water pH is affected by numerous processes: biological, physical and geochemical. Examples are: CO2 pressure equilibrium with the atmosphere, photosynthesis, respiration of plants, organic matter degradation, geological and mineral background, pollution etc. Inter-relationships between these processes and pH values are complex, non-linear and not well understood. As a neural network provides a non-linear function mapping of a set of input variables into the corresponding network output, without the requirement of having to specify the actual mathematical form of the relation between the input and output variables, it has the versatility for modelling a wide range of complex non-linear phenomena. For this reason the neural network approach has been selected and tested for pH modelling. We used the classical multilayer perceptron model (MLP). River discharge and solar radiation variables are used as inputs to the MLP model. The choice of these variables is dictated by what are perceived to be the predominant processes that control pH in the Middle Loire river, which is typically eutrophic during the low-flow summer period. The influence of the previous day’s flows and radiation has been evaluated in the calibration and verification test. The best network found to simulate pH was one with two input nodes and three hidden nodes. The inputs are: daily discharge and a variable called ‘Index of anterior radiation’, i.e. calculated as an exponential smoothing of the daily radiation variable. When calibrated over 4 years of data and tested (i.e. verified) for a one-year independent set of data, the model proved satisfactory on pH simulations, with accuracies in the order of 86%. After elaborating the pH model, the Student test and the cumulative Page‐Hinkley test were applied for automatic detection of changes in the mean of the residuals from the ANN pH model. This analysis has shown that such tests are capable of detecting a measurement error occurring over a short period of time (1‐4 days).

Severe and contrasted polymetallic contamination patterns (1900–2009) in the Loire River sediments (France)

The Loire River basin (117,800 km(2), France) has been exposed to multiple sources of metals during the last 150 years, originating from major mining districts (coal and non-ferrous metals) and their associated industrial activities. Geochemical archives are established here from the analysis of a 4m sediment core in the downstream floodplain and then compared to stream bed sediments from pristine monolithological sub-basins and from bed and bank sediments in impacted tributaries. The contamination is assessed for 55 major and trace elements through their enrichment factors to Al (EF), normalized to the pre-anthropogenic background. Archives from 1900 to 2009 show enrichment (EF<1.3) not only for Ba, Be, Cs, Ga, Rb, REE, Sr, V, and Zr but also for U and Th, despite U mining activities until the 1990s. From 1900 to 1950, the level of contamination is severe for Hg, Au, Ag (10<EF<30), important for Sb and Sn (3<EF<7) and moderate for Cu, Pb and Zn (1.5<EF<3). This state was mostly attributed to coal uses and metal mining. During the period 1950-1980, severe polymetallic contamination is noted for Hg (EF up to 53), Cd (23), Ag (18), Zn (6.2), Cu (6.0), Sn (5.6), Pb(4.8), Sb(4.4) and for new impacted elements as Bi (23.8), As (3.7), Cr (3.4), W (3.1), Mo (2.6), Ni (2.8), Co (1.65) due to mines, smelters, industries and from urban sewers, collected mostly after 1950 (total population of 8.4 million people). The limited dilution by detrital material (Loire sediment load about 1.5 Mt/year) is an additional cause of such severe contamination. After 1950, river eutrophication is well marked by the general increase of endogenic calcite (EF (Ca)=4), diluting all other elements by 20%. From 1980 to 2009, all contaminants, except Au (EF=100), decrease steadily.

Elemental properties, hydrology, and biology interact to shape concentration‐discharge curves for carbon, nutrients, sediment, and major ions

To investigate the prevalence and cause of concentration-discharge (C-Q) relationships for carbon, nutrients, major ions, and particulates, we analyzed 40 years of water quality data from 293 monitoring stations in France. Catchments drained diverse landscapes and ranged from 50 to 110,000 km2, together covering nearly half of France. To test for differences during low and high flows, we calculated independent C-Q slopes above and below the median discharge. We found that 84% of all catchment-element combinations were chemodynamic for at least half of the hydrograph and 60% of combinations showed non-linear C-Q curves. Only two or three of the nine possible C-Q modalities were manifest for each parameter, and these modalities were stable through time, suggesting that intrinsic and extrinsic elemental properties (e.g. solubility, reactivity, and source dynamics) set basic C-Q templates for each parameter, which are secondarily influenced by biological activity during low flows, and the interaction between hydrology and catchment characteristics at high flows. Several patterns challenged current C-Q views, including low-flow chemostasis for TSS in 66% of catchments, low-flow biological mediation of NO3- in 71% of catchments, and positive C-Q for dissolved organic carbon independent of catchment size in 80% of catchments. Efforts to reduce nutrient loading decreased phosphorus concentration and altered C-Q curves, but NO3- continued to increase. While C-Q segmentation requires more data than a single analysis, the prevalence of non-linear C-Q slopes demonstrates the potential information loss associated with linear or monotonic analysis of C-Q relationships, and conversely, the value of long-term monitoring. This article is protected by copyright. All rights reserved.

A quality-control method for physical and chemical monitoring data. Application to dissolved oxygen levels in the river Loire (France)

Abstract A quality-control method is proposed for examining continuous physical and chemical measurements, including temperature, dissolved oxygen, pH and electrical conductivity. Firstly, measurement consistency is evaluated by various modelling approaches: internal series structure, inter-variable relations or relations with external variables, spatial coherence and deterministic models. Secondly, outliers or systematic errors are detected using classical statistical tests. The method was evaluated for dissolved oxygen concentrations (DO) in the river Loire at Dampierre over a 5-year period (1990–1994), using data records containing fictitious errors, and raw data for the year 1995. The results demonstrate the effectiveness and advantages of a multi-model approach. In the case of dissolved oxygen for example, slow continuous drifts are always detected in under 4 days.

Assessing N emissions in surface water at the national level: comparison of country-wide vs. regionalized models.

Many countries are developing models to estimate N emissions in rivers as part of national-scale water quality assessments. Generally, models are applied with national databases, while at the regional scale, more detailed databases are sometimes available. This paper discusses pros and cons of developing regionalized models versus applying countrywide models. A case study is used to support the discussion. The model used, called Nutting-N (NUTrient Transfer modelING-Nitrogen), relies on a statistical approach linking nitrogen sources and watershed land and river characteristics and aims to evaluate the risk of water bodies failing to reach quality objectives defined by national and federal policies. After calibration and evaluation at the national scale (France), the predictive quality of the model was compared with two regionalized models in a crystalline massif (Brittany, western France, 27,000 km(2)) and in a sedimentary basin (Seine, Paris basin, 78,000 km(2)), where detailed regional databases are available. The national-scale model provided robust predictions in most conditions encountered in France (efficiency=0.69). Terrestrial retention was related mainly to specific runoff, and its median value was estimated at 49% of the N surplus, whereas median river retention represented 18% of incoming N discharge. Regionalizing the model generally improved goodness-of-fit, as the root mean squared error was reduced by 6-24%. However, precision of parameter estimates degraded when too few monitoring basins were available or when variability in land and river characteristics was too low in the calibration dataset. Hence, regional-scale models should be advocated only after the trade-off between improvement of fit and degradation of parameter estimates is examined.

Estimation of local extreme suspended sediment concentrations in California Rivers.

The total amount of suspended sediment load carried by a stream during a year is usually transported during one or several extreme events related to high river flow and intense rainfall, leading to very high suspended sediment concentrations (SSCs). In this study quantiles of SSC derived from annual maximums and the 99th percentile of SSC series are considered to be estimated locally in a site-specific approach using regional information. Analyses of relationships between physiographic characteristics and the selected indicators were undertaken using the localities of 5-km radius draining of each sampling site. Multiple regression models were built to test the regional estimation for these indicators of suspended sediment transport. To assess the accuracy of the estimates, a Jack-Knife re-sampling procedure was used to compute the relative bias and root mean square error of the models. Results show that for the 19 stations considered in California, the extreme SSCs can be estimated with 40-60% uncertainty, depending on the presence of flow regulation in the basin. This modelling approach is likely to prove functional in other Mediterranean climate watersheds since they appear useful in California, where geologic, climatic, physiographic, and land-use conditions are highly variable.

Trends and seasonality of river nutrients in agricultural catchments: 18 years of weekly citizen science in France

Agriculture and urbanization have disturbed three-quarters of global ice-free land surface, delivering huge amounts of nitrogen and phosphorus to freshwater ecosystems. These excess nutrients degrade habitat and threaten human food and water security at a global scale. Because most catchments are either currently subjected to, or recovering from anthropogenic nutrient loading, understanding the short- and long-term responses of river nutrients to changes in land use is essential for effective management. We analyzed a never-published, 18-year time series of anthropogenic (NO3- and PO43-) and naturally derived (dissolved silica) riverine nutrients in 13 catchments recovering from agricultural pollution in western France. In a citizen science initiative, high-school students sampled catchments weekly, which ranged from 26 to 1489km2. Nutrient concentrations decreased substantially over the period of record (19 to 50% for NO3- and 14 to 80% for PO43-), attributable to regional, national, and international investment and regulation, which started immediately prior to monitoring. For the majority of catchments, water quality during the summer low-flow period improved faster than during winter high-flow conditions, and annual minimum concentrations improved relatively faster than annual maximum concentrations. These patterns suggest that water-quality improvements were primarily due to elimination of discrete nutrient sources with seasonally-constant discharge (e.g. human and livestock wastewater), agreeing with available land-use and municipal records. Surprisingly, long-term nutrient decreases were not accompanied by changes in nutrient seasonality in most catchments, attributable to persistent, diffuse nutrient stocks. Despite decreases, nutrient concentrations in almost all catchments remained well above eutrophication thresholds, and because additional improvements will depend on decreasing diffuse nutrient sources, future gains may be much slower than initial rate of recovery. These findings demonstrate the value of citizen science initiatives in quantifying long-term and seasonal consequences of changes in land management, which are necessary to identify sustainable limits and predict recovery timeframes.

Uncertainties in Assessing Annual Nitrate Loads and Concentration Indicators: Part 2. Deriving Sampling Frequency Charts in Brittany, France

In water quality monitoring programs, standard sampling frequency schemes tend to be applied throughout entire regions or states. Ideally, the common standard among monitoring stations ought not to be the sampling frequency but instead the level of uncertainty of the estimated water quality indicators. Until now, there was no obvious way of doing this. This article proposes, for the first time, guidelines to select appropriate sampling frequencies to harmonize the level of uncertainty in the case of yearly nitrate indicators for the regional river water quality monitoring network in Brittany, France. A database of 50 watershed-year datasets (nine watersheds of 4 to 252 km2 in size) was used for which high temporal resolution data (hourly and daily) were available for flow and nitrate concentrations. For each dataset, the uncertainty levels were calculated by numerically simulating sampling intervals varying from 2 to 60 days. The precision limits of the uncertainties were successfully correlated to a hydrological reactivity index. The correlations were used to derive sampling frequency charts. These charts can be used by watershed managers to optimize the sampling frequency scheme for any watershed for a desired uncertainty level, provided that the dimensionless local hydrological reactivity can be calculated from previous records of continuous flow rates. The sampling frequency charts also suggest that, depending on the hydrological reactivity, expected uncertainties generated by monthly sampling range between ±6% and ±14% for the annual load and between -5% and +2.5% to +7.2% for the annual concentration average.