Long-term behaviour and cross-correlation water quality analysis of the River Elbe, Germany

>Water quality prediction of artificial intelligence model: a case of Huaihe River Basin, China

Accurate prediction of water quality contributes to the intelligent management of water resources. Water quality indices have time series characteristics and nonlinearity, but the existing models only focus on the forward time series when long short-term memory (LSTM) is introduced and do not consider the parallel computation on the model. Owing to this, a new neural network called LSTM-multihead attention (LMA) was constructed to predict water quality, using long short-term memory to process time series data and multihead attention for parallel computing and extracting feature information. Additionally, water quality indices have the issues of multiple data types and complex data correlations, as well as missing data and abnormal data problems in water quality data. In order to solve these problems, this study proposes a water quality prediction model called GRA-LMA-based linear interpolation, gray relational analysis and LMA. Two experiments are carried out to verify the predictive performance of the GRA-LMA with the water quality data of the Huaihe River Basin as a case study sample. The first experiment focuses on data processing, including the processing of missing data and abnormal data of water quality data, and the correlation analysis of water quality indices. Linear interpolation is adapted to process the missing data, while a combination of boxplot and histogram is adopted to analyze and eliminate the abnormal data, which is then repaired the abnormal data with linear interpolation. The gray relational analysis is adopted to calculate the correlation between different water quality indices, and water quality indices with high correlation are retained to determine the input variables of the water quality prediction model. The data processing results demonstrate that repairs can be made using linear interpolation without altering the pattern of data change and the model by using the gray relational analysis to reduce the quantity of data it needs as input. In the second experiment, the predictive capacity of GRA-LMA and existing models such as backpropagation neural network (BP), recurrent neural network (RNN), long short-term memory (LSTM), and gate recurrent unit (GRU) was evaluated and compared using different numerical and graphical performance evaluation metrics. Comparative experimental results show that the mean square error of pH, dissolved oxygen, chemical oxygen demand, ammonia nitrogen, electrical conductivity, turbidity, total phosphorus, and total nitrogen of GRA-LMA is reduced to 0.05890, 0.40196, 0.32454, 0.04368, 14.71003, 8.13252, 0.01558, and 0.14345. The results indicate that GRA-LMA has superior adaptability for predicting various water quality indices and can significantly lower the induced prediction error.

Twenty-eight years of ecosystem recovery and destabilisation: Impacts of biological invasions and climate change on a temperate river

Most river ecosystems are exposed to multiple anthropogenic stressors affecting the composition and functionality of benthic communities. Identifying main causes and detecting potentially alarming trends in time depends on the availability of long-term monitoring data sets. Our study aimed to improve the knowledge about community effects of multiple stressors that is needed for effective, sustainable management and conservation. We conducted a causal analysis to detect the dominant stressors and hypothesised that multiple stressors, such as climate change and multiple biological invasions, reduce biodiversity and thus endanger ecosystem stability. Using a data set from 1992 to 2019 for the benthic macroinvertebrate community of a 65-km stretch of the upper Elbe river in Germany, we evaluated the effects of alien species, temperature, discharge, phosphorus, pH and abiotic conditional variables on the taxonomic and functional composition of the benthic community and analysed the temporal behaviour of biodiversity metrics. We observed fundamental taxonomic and functional changes in the community, with a shift from collectors/gatherers to filter feeders and feeding opportunists preferring warm temperatures. A partial dbRDA revealed significant effects of temperature and alien species abundance and richness. The occurrence of distinct phases in the development of community metrics suggests a temporally varying impact of different stressors. Taxonomic and functional richness responded more sensitively than the diversity metrics whereas the functional redundancy metric remained unchanged. Especially the last 10-year phase, however, showed a decline in richness metrics and an unsaturated, linear relationship between taxonomic and functional richness, which rather indicates reduced functional redundancy. We conclude that the varying anthropogenic stressors over three decades, mainly biological invasions and climate change, affected the community severely enough to increase its vulnerability to future stressors. Our study highlights the importance of long-term monitoring data and emphasises a careful use of biodiversity metrics, preferably considering also community composition.

Lagrangian profiles of riverine autotrophy, organic matter transformation, and micropollutants at extreme drought

On their way from inland to the ocean, flowing water bodies, their constituents and their biotic communities are exposed to complex transport and transformation processes. However, detailed process knowledge as revealed by Lagrangian measurements adjusted to travel time is rare in large rivers, in particular at hydrological extremes. To fill this gap, we investigated autotrophic processes, heterotrophic carbon utilization, and micropollutant concentrations applying a Lagrangian sampling design in a 600 km section of the River Elbe (Germany) at historically low discharge. Under base flow conditions, we expect the maximum intensity of instream processes and of point source impacts. Phytoplankton biomass and photosynthesis increased from upstream to downstream sites but maximum chlorophyll concentration was lower than at mean discharge. Concentrations of dissolved macronutrients decreased to almost complete phosphate depletion and low nitrate values. The longitudinal increase of bacterial abundance and production was less pronounced than in wetter years and bacterial community composition changed downstream. Molecular analyses revealed a longitudinal increase of many DOM components due to microbial production, whereas saturated lipid-like DOM, unsaturated aromatics and polyphenols, and some CHOS surfactants declined. In decomposition experiments, DOM components with high O/C ratios and high masses decreased whereas those with low O/C ratios, low masses, and high nitrogen content increased at all sites. Radiocarbon age analyses showed that DOC was relatively old (890-1870 years B.P.), whereas the mineralized fraction was much younger suggesting predominant oxidation of algal lysis products and exudates particularly at downstream sites. Micropollutants determining toxicity for algae (terbuthylazine, terbutryn, isoproturon and lenacil), hexachlorocyclohexanes and DDTs showed higher concentrations from the middle towards the downstream part but calculated toxicity was not negatively correlated to phytoplankton. Overall, autotrophic and heterotrophic process rates and micropollutant concentrations increased from up- to downstream reaches, but their magnitudes were not distinctly different to conditions at medium discharges.

Insights into Phytoplankton Dynamics and Water Quality Monitoring with the BIOFISH at the Elbe River, Germany

Understanding the key factors influencing the water quality of large river systems forms an important basis for the assessment and protection of cross-regional ecosystems and the implementation of adapted water management concepts. However, identifying these factors requires in-depth comprehension of the unique environmental systems, which can only be achieved by detailed water quality monitoring. Within the scope of the joint science and sports event “Elbschwimmstaffel” (swimming relay on the river Elbe) in June/July 2017 organized by the German Ministry of Education and Research, water quality data were acquired along a 550 km long stretch of the Elbe River in Germany. During the survey, eight physiochemical water quality parameters were recorded in high spatial and temporal resolution with the BIOFISH multisensor system. Multivariate statistical methods were applied to identify and delineate processes influencing the water quality. The BIOFISH dataset revealed that phytoplankton activity has a major impact on the water quality of the Elbe River in the summer months. The results suggest that phytoplankton biomass constitutes a substantial proportion of the suspended particles and that photosynthetic activity of phytoplankton is closely related to significant temporal changes in pH and oxygen saturation. An evaluation of the BIOFISH data based on the combination of statistical analysis with weather and discharge data shows that the hydrological and meteorological history of the sampled water body was the main driver of phytoplankton dynamics. This study demonstrates the capacity of longitudinal river surveys with the BIOFISH or similar systems for water quality assessment, the identification of pollution sources and their utilization for online in situ monitoring of rivers.

Assessment of Automatically Monitored Water Levels and Water Quality Indicators in Rivers with Different Hydromorphological Conditions and Pollution Levels in Greece

Water resources, especially riverine ecosystems, are globally under qualitative and quantitative degradation due to human-imposed pressures. High-temporal-resolution data obtained from automatic stations can provide insights into the processes that link catchment hydrology and streamwater chemistry. The scope of this paper was to investigate the statistical behavior of high-frequency measurements at sites with known hydromorphological and pollution pressures. For this purpose, hourly time series of water levels and key water quality indicators (temperature, electric conductivity, and dissolved oxygen concentrations) collected from four automatic monitoring stations under different hydromorphological conditions and pollution pressures were statistically elaborated. Based on the results, the hydromorphological conditions and pollution pressures of each station were confirmed to be reflected in the results of the statistical analysis performed. It was proven that the comparative use of the statistics and patterns of the water level and quality high-frequency time series could be used in the interpretation of the current site status as well as allowing the detection of possible changes. This approach can be used as a tool for the definition of thresholds, and will contribute to the design of management and restoration measures for the most impacted areas.

Stochastic modelling for predicting COVID-19 prevalence in East Africa Countries

Coronavirus (COVID-19) has continued to be a global threat to public health. As the matter of fact, it needs unreserved effort to monitor the prevalence of the virus. However, applying an effective prediction of the prevalence is thought to be the fundamental requirement to effectively control the spreading rate. Time series models have extensively been considered as the convenient methods to predict the prevalence or spreading rate of the disease. This study, therefore, aimed to apply the Autoregressive Integrated Moving Average (ARIMA) modeling approach for projecting coronavirus (COVID-19) prevalence patterns in East Africa Countries, mainly Ethiopia, Djibouti, Sudan and Somalia. The data for the study were obtained from the reports of confirmed COVID-19 cases by the official website of Johns Hopkins University from 13th March, 2020 to 30th June, 2020.The results of the study, then, showed that in the coming four month, the number of COVID-19 positive people in Ethiopia may reach up to 56,610 from 5,846 on June 30, 2020 in average-rate scenario. However, in worst case scenario forecast, the model showed that the cases will be around 84,497. The analysis further depicted that with average interventions and control scenario, cumulative number of infected persons in Djibouti, Somalia and Sudan will increase from 4,656, 2,904 and 9,258 respectively at the end of June to 8,336, 3,961 and 21,388, which is by the end of October, 2020, after four-months. But, with insufficient intervention, the number of infected persons may grow quickly and reach up to 14,072, 10,037 and 38,174 in Djibouti, Somalia and Sudan respectively. Generally, the extent of the coronavirus spreading was increased from time to time in the past four month, until 30 th June, 2020, and it is expected to continue quicker than before for the coming 4-month, until the end of October, 2020, in Ethiopia, Djibouti, Somalia, and Sudan and more rapidly than before in Sudan and Ethiopia, while the peak will remain unknown yet. Therefore, an effective implementation of the preventive measures and a rigorous compliance by avoiding negligence with the rules such as prohibiting public gatherings, travel restrictions, personal protection measures, and social distancing may alleviate the spreading rates of the virus, particularly, Sudan and Ethiopia. Moreover, more efforts should be exerted on Ethiopian side to control the population movement across all the border areas and to strengthen border quarantining. Further, through updating more new data with continuous reconsideration of predictive model, provide useful and more precise prediction. Applying, ARIMAX-Transfer Function model in region-wise by take in to consideration of climatic data like temperature and humidity in each countries looking spatial pattern for reliable measure of COVID-19 prevalence.

Towards the comprehensive water quality control in Lake Taihu: Correlating chlorphyll a and water quality parameters with generalized additive model

In this study, the generalized additive model (GAM) was used to analyze seasonal monitoring data from Lake Taihu, collected from 2010 to 2014, with the aim to explore the correlation between chlorophyll a (Chla) and other water quality parameters. The selected optimal multivariable GAM could effectively explain the concentration variation of Chla occurring during each season, and the interpretation degree followed the order: summer > autumn > spring > winter. The fitting results indicated that the concentration variation of Chla could reflect that of biochemical oxygen demand and chemical oxygen demand in all seasons. In addition, the total phosphorus showed strong ability to explain the concentration change of Chla in spring and summer, as the growth of algae would be affected when the concentration of phosphorus shifted high or low. Nitrogen showed strong ability to explain the variations in Chla concentration in autumn. The conclusions of the optimal multivariable GAM could provide decision basis for the eutrophication control. In other words, the prevention of eutrophication outbreaks could be carried out via the targeted control of key water pollutants. According to these results, the concentration of Chla was higher in northern and western lake during summer and autumn, the management should focus on nutrient input of adjacent rivers.

Sedimentary Archives Reveal the Concealed History of Micropollutant Contamination in the Seine River Basin

Sedimentary archives provide long-term records of particulate-bound pollutants (e.g. trace metal elements, PAHs). We present the results obtained on a set of selected cores from alluvial deposits within the Seine River basin, integrating the entire area’s land uses upstream of the core location, collected upstream and downstream of Paris megacity and in the estuary. Some of these cores go back to the 1910s. These records are complemented by in-depth studies of the related pollution emissions, their regulation and other environmental regulations, thereby establishing contaminant trajectories. They are representative of a wide range of contamination intensities resulting from industrial, urban and agricultural activities and their temporal evolution over a 75,000 km2 territory. A wide set of contaminants, including metals, radionuclides, pharmaceuticals and up to 50 persistent organic pollutants, have been analysed based on the Seine River sediment archives. Altogether, more than 70 particulate contaminants, most of them regulated or banned (OSPAR convention, European Water Framework Directive (WFD 2000/60/EC)), were measured in dated cores collected at 7 sites, resulting in a large data set.

After drawing a picture of the literature devoted to sedimentary archives, the findings resulting from several decades of research devoted to the Seine River basin will be used, together with other studies on other French and foreign rivers, to illustrate the outstanding potential of sedimentary archives. The limitations of using sedimentary archives for inter-site comparison and the approaches developed in the PIREN-Seine to overcome such limitations such as selecting pertinent indicators (specific fluxes, per capita release, leakage rate, etc.) will be described. The very complex interactions between humans and their environment will be addressed through questions such as the impact on the spatial and temporal trajectories of contaminants of factors such as wastewater management, deindustrialisation within the Seine River basin, implementation of national and EU environmental regulations, etc. This chapter will show how such studies can reveal the persistence of the contamination and the emergence of new pollutants, e.g. antibiotics. It will propose indicators for the evaluation of the environment resilience and the efficiency of environmental policies.

Detection, provenance and associated environmental risks of water quality pollutants during anomaly events in River Atoyac, Central Mexico: A real-time monitoring approach

River Atoyac is considered to be one of the most polluted rivers in Mexico due to the discharges of untreated or partially treated wastewater from industrial and municipal activities. In order to improve the river water quality, it is obligatory to identify the possible contaminant sources for upholding a well-balanced ecosystem. Henceforth, the present study incorporates the application of a continuous real-time monitoring system to identify the provenance of pollutants of the river mainly from anomaly events. Four monitoring stations were installed all along the River Atoyac in the State of Puebla, Central Mexico. The real-time monitoring systems have an ability to measure various water quality parameters for every 15 minutes such as Temperature (T), pH, Conductivity (EC), turbidity (TURB), Dissolved Oxygen (DO), Oxidation Reduction Potential (ORP) and Spectral Absorption Coefficient (SAC). In total, eight water samples of anomaly events (i.e.) 2 per monitoring station during rainy (August-September) and winter seasons (November-December), that were detected using the parameters previously mentioned were procured and also analyzed in the laboratory for evaluating almost 54 physicochemical, inorganic and organic characteristics. Statistical results of factorial analysis explained that 30% of the total variance corresponded to textile effluents, 23% related to discharges produced by automobile and petrochemical industries, and 18% of the total variance defined the agricultural activities. Additionally, indices like Overall Index Pollution, Heavy Metal Evaluation Index, Screening Quick Reference Table and Molecular ratios of hydrocarbons for PAH sources was also calculated to estimate the grade of pollution and associated ecotoxicological risks. The present study also enlightens the fact that the assessed results will definitely provide valuable information for the management of river water quality by developing stringent public policies by governmental agencies for the sustainable conservation of Atoyac River.

Seawater quality conditions of the south Andaman Sea (Bay of Bengal, Indian Ocean) in lustrum during 2010s decade

Andaman and Nicobar islands is one of the major tourism hubs of the World. Most travelers visit these islands for historical attractions, beaches, snorkeling, scuba diving, coral reefs, adventure and recreation. Port Blair is the capital and sole entry/exit point of these islands. The coasts of Port Blair Bay (PBB) and Wandoor Creek (WC) are largely populated due to its services offered to different public/private sectors and for the economic significance. Nevertheless, the global recognition of these islands relies on its healthy ecosystem. Effective management of beaches, bays and their environmental services requires knowledge of coastal water quality. This study assesses the datasets of twenty seawater quality parameters of PBB and WC generated during five years (2011-2015) at eight sites. Multivariate statistical techniques were used for (i) analysis and interpretation of water quality parameters (ii) identification of pollution factors/sources and (iii) understanding spatio-temporal variations valuable for coastal water quality management.

Driver detection of water quality trends in three large European river basins

This study analyses how indicators of water quality (thirteen physico-chemical variables) and drivers of change (i.e., monthly aggregated air temperature and streamflow, population density, and percentage of agricultural land use) coevolve in three large European river basins (i.e., Adige, Ebro, Sava) with different climatic, soil and water use conditions. Spearman rank correlation, Principal Component Analysis, and Mann-Kendall trend tests were applied to long-term time series of water quality data during the period 1990-2015 in order to investigate the relationships between water quality parameters and the main factors controlling them. Results show that air temperature, considered as a proxy of climatic change, has a significant impact, in particular in the Adige and Ebro: positive trends of water temperature and negative of dissolved oxygen are correlated with upward trends of air temperatures. The aquatic ecosystems of these rivers are, therefore, experiencing a reduction in oxygen, which may exacerbate in the future given the projected further increase in temperature. Furthermore, monthly streamflow has been shown to reduce in the Ebro, thereby reducing the beneficial effect of dilution, which appears evident from the observed upward patterns of chloride concentrations and electrical conductivity. Upward trends of chloride and biological oxygen demand in the Adige and Sava, and of phosphate in the Adige appears to be related to increasing human population density, whereas phosphates in the Sava and biological oxygen demand in the Ebro are highly correlated with agricultural land use, considered as a proxy of the impact of agricultural practises. The present study shows the complex relationships between drivers and observed changes in water quality parameters. Such analysis can represent, complementary to a deep knowledge of the investigated systems, a reliable tool for decision makers in river basin planning by providing an overview of the potential impacts on the aquatic ecosystem of the three basins.

Disentangling the influence of hydroclimatic patterns and agricultural management on river nitrate dynamics from sub-hourly to decadal time scales

Despite extensive efforts to reduce nitrate transfer in agricultural areas, limited response is often observed in the nitrate concentration in rivers. To investigate the reasons for this limited response, nitrate dynamics in a 100km(2) agricultural catchment in eastern Germany was analysed from sub-hourly to decadal time-scales. Sub-hourly analysis of storm event dynamics during a typical hydrological year (2005-2006) was performed to identify periods of the year with high leaching risk and to link the latter to agricultural management practices in the catchment. Dynamic Harmonic Regression analysis of a 32-year (1982-2014) record of nitrate and discharge revealed that i) the long-term trend in nitrate concentration was closely related to that in discharge, suggesting that large-scale weather and climate patterns were masking the effect of improved nitrogen management on nitrate trends; ii) a persistent seasonal pattern with winter concentration maxima and summer minima could be observed, which was interpreted in terms of a dynamic nitrate concentration profile in the soil and subsoil; and iii) the catchment progressively changed from chemodynamic to more chemostatic behaviour over the three decades of study, which is a sign of long-term homogenisation of nitrate concentrations distribution over depth. This study shows that detailed physical understanding of nitrate dynamics across time scales can be obtained only through combined analysis of long-term records and high-resolution sensor data. Hence, a joint effort is advocated between environmental authorities, who usually perform long-term monitoring, and scientific programmes, which usually perform high-resolution monitoring.

Using recent high-frequency surveys to reconstitute 35 years of organic carbon variations in a eutrophic lowland river

Concentrations of dissolved and particulate organic carbon (DOC and POC), total suspended solids (TSS), were measured daily, and phytoplankton pigments (chlorophyll-a and pheopigments) were measured every 3 days at three strategic stations along the eutrophic Loire River between November 2011 and November 2013 marked by a high annual and seasonal variability in hydrological regimes. This unique high-frequency dataset allowed to determine the POC origin (autochthonous or allochthonous). Some strong relationships were evidenced between POC, total pigments and TSS and were tested on a long-term database with a lower frequency (monthly data) to reconstitute unmeasured algal and detrital POC concentrations and estimate annual total organic carbon (TOC) fluxes from 1980 onwards. The results were subjected to only ≈25 % uncertainty and showed that the annual TOC fluxes at the outlet of the Loire River decreased from 520 10(3) tC year(-1) (i.e. 4.7 t km(-2) year(-1)) in the early 1990s to 150 10(3) tC year(-1) (i.e. 1.4 t km(-2) year(-1)) in 2012. Although DOC always dominates, the autochthonous POC represented 35 % of the TOC load at the basin outlet by the end of the 1980s and declined to finally represent 15 % only of the TOC. The control of phosphorus direct inputs and the invasion by Corbicula clams spp. which both occurred since the early 1990s probably highly reduced the development of phytoplankton. Consequently, the autochthonous POC contribution declined and TSS concentrations in summertime significantly decreased as well as a result of both less phytoplankton and less calcite precipitation. At the present time, at least 75 % of the POC has allochthonous origins in the upper Middle Loire but downstream, autochthonous POC dominates during summer phytoplanktonic blooms when total pigments concentrations reach up to 70 μg L(-1) (equivalent to 75 % of the total POC).

Biochemical markers for the assessment of aquatic environment contamination

The need for assessment of aquatic ecosystem contamination and of its impact on water dwelling organisms was developed in response to rising aquatic environmental pollution. In this field study, liver enzymes of phase I and phase II of xenobiotic transformation, namely cytochrome P450, ethoxyresorufin-O-deethylase, glutathione-S-transferase and tripeptide glutathione were used to assess the contamination of the aquatic environment at different rivers in the Czech Republic. The indicator species selected was the male chub (Leuciscus cephalus L.) and male brown trout (Salmo trutta fario). Chemical analyses included also the assessment of the most important inductors of previously mentioned biochemical markers. The major inductors of monitored biomarkers are industrial contaminants which belong to a large group of organic pollutants (PCB, PAH, PCDD/F, DDT, HCH, HCB and OCS), persistent in the environment. Four different groups of river basins were assessed: the River Tichá Orlice and its tributary the Kralický brook; important tributaries of the River Elbe (the rivers Orlice, Chrudimka, Cidlina, Jizera, Vltava, Ohře and Bílina); major rivers in the Czech Republic (the rivers Lužnice, Otava, Sázava, Berounka, Vltava, Labe, Ohře, Svratka, Dyje, Morava and Odra) and the River Vltava. The use of the biochemical markers together with chemical analyses seems to be an effective way to monitor the quality of aquatic environment.

Biomarkers of Contaminant Exposure in Chub (Leuciscus cephalus L.) - Biomonitoring of Major Rivers in the Czech Republic

Biochemical analysis of organisms to assess exposure to environmental contaminants is of great potential use. Biochemical markers, specifically liver enzymes of the first and the second phase of xenobiotic transformation - cytochrome P₄₅₀ (CYP 450), ethoxyresorufin-O-deethylase (EROD), glutathione-S-transferase (GST) and tripeptide reduced glutathione (GSH) - were used to assess contamination of the aquatic environment at 12 locations near the mouths of major rivers in the Czech Republic. These rivers were the Lužnice, Otava, Sázava, Berounka, Vltava, Labe, Ohře, Svratka, Dyje, Morava and Odra. The indicator species selected was the Chub (Leuciscus cephalus L.). The highest levels of CYP 450 and EROD catalytic activity were found in livers of fish from the Labe (Obříství; (0.32±0.10 nmol mg⁻¹ protein and 1061.38±545.51 pmol min⁻¹ mg⁻¹ protein, respectively). The highest levels of GST catalytic activity and GSH content were found in fish from the Otava (35.39±13.35 nmol min⁻¹ mg⁻¹ protein and 4.29±2.10 nmol GSH mg⁻¹ protein, respectively). They were compared with levels of specific inductors of these biochemical markers in muscle. The results confirmed contamination of some river locations (Labe Obříství, Svratka;.

Spatial and temporal variability of sediment deposition on artificial-lawn traps in a floodplain of the River Elbe

Artificial-lawn mats were used as sediment traps in floodplains to measure sediment input and composition during flood events. To estimate the natural variability, 10 traps were installed during two flood waves at three different morphological units in a meander loop of the River Elbe. The geochemical composition of deposited and suspended matter was compared. The sediment input showed weak correlations with concentration and composition of river water. It also correlated poorly with flood duration and level as well as distance of trap position from the main river. This is due to the high variability of the inundation, different morphological conditions and the variability of sources. The composition of the deposits and the suspended matter in the river water was comparable. Hence, for the investigated river reach, the expected pollution of the floodplain sediments can be derived from the pollution of the suspended matter in the river during the flood wave.

Long-term variations in dissolved trace elements in the Sagami River and its tributaries (upstream area), Japan

We investigated long-term variations in dissolved chemical species in water sampled from the Sagami River and its tributaries, Japan. The samples were taken monthly from May 1993 to April 2000 at 28 sampling sites in the Sagami River system. In this paper, we concentrate on 17 sites in the upper catchment. Twenty-four major to trace elements (Li, Mg, Al, Ca, V, Mn, Fe, Co, Ni, Cu, Zn, As, Rb, Sr, Mo, Ag, Cd, Sb, Cs, Ba, W, Tl, Pb, U) were determined by using conventional Q-pole ICP-MS with direct nebulizalion. Water flux was measured from 1995. The data for most species were subjected to fast Fourier transformation (FFT) to extract dominant periodicities, their magnitudes, and their phases. Clear seasonal variations were observed for Al, V, As, Rb and Cs at some sampling points, especially in the tributaries. The seasonal variations may be due to either anthropogenic causes, such as irrigation or wastewater discharge, or natural causes, such as water temperature, pH, redox condition, water flux, or activity of microorganisms. We found no correlation with pH. Water temperature may not be a main controlling factor, although the seasonal variability would be correlated with it. Hydrologic factors may have only minor effects. We suggest that most of the seasonal variation might be correlated with the irrigation of rice paddies. It was difficult to identify all the causes of the seasonal variability. To investigate interannual trends, we used centered 12-month moving averages to eliminate seasonal variations. The water quality of the uppermost streams was generally constant, being controlled by springs on Mt. Fuji. However, in 1993 and 1998, V, As, Rb, Cs and U were depleted briefly, possibly by dilution by overflow from Lakes Kawaguchi and Yamanaka. Hydrologic factors may be more important for interannual variability than seasonal variability. Some heavy trace elements (Zn, Ni, Cd and Sb) showed very irregular variations with high concentration peaks. These elements were discharged from either factories or abandoned mine tailings.