Assessing uncertainty in annual nitrogen,phosphorus, and suspended sediment load estimates in three agricultural streams using a 21-year dataset

Accurate estimation of constituent loads is important for studies of ecosystem mass balance or total maximum daily loads. In response, there has been an effort to develop methods to increase both accuracy and precision of constituent load estimates. The relationship between constituent concentration and stream discharge is often complicated, potentially leading to high uncertainty in load estimates for certain constituents, especially at longer-term (annual) scales. We used the loadflex R package to compare uncertainty in annual load estimates from concentration vs. discharge relationships in constituents of interest in agricultural systems, including ammonium as nitrogen (NH₄-N), nitrate as nitrogen (NO₃-N), soluble reactive phosphorus (SRP), and suspended sediments (SS). We predicted that uncertainty would be greatest in NO₃-N and SS due to complex relationships between constituent concentration and discharge. We also predicted lower uncertainty with a composite method compared to regression or interpolation methods. Contrary to predictions, we observed the lowest uncertainty in annual NO₃-N load estimates (relative error 1.5–23%); however, uncertainty was greatest in SS load estimates, consistent with predictions (relative error 19–96%). For all constituents, we also generally observed reductions in uncertainty by up to 34% using the composite method compared to regression and interpolation approaches, as predicted. These results highlight differences in uncertainty among different constituents and will aid in model selection for future studies requiring accurate and precise estimates of constituent load.     

Stream sediment and nutrient loads in the Tahoe Basin—estimated vs monitored loads for TMDL “crediting”

Total maximum daily load (TMDL) programs utilize pollutant load reductions as the primary strategy to restore adversely affected waters of the USA. Accurate and defensible “crediting” for TMDL reductions of sediment and nutrients requires stream monitoring programs capable of quantitative assessment of soil erosivity and the “connectivity” between erosive areas and stream channels across the watershed. Using continuous (15-min) stream monitoring information from typical alpine, snowmelt-driven watersheds [Ward (2,521 ha), Blackwood (2,886 ha), and Homewood (260 ha, Homewood Mountain Resort—HMR) Creeks] on the west shore of the Lake Tahoe Basin, daily sediment (and nutrient for HMR) loads are determined and compared with those developed from estimated load–flow relationships developed from grab sampling data. Compared to the previously estimated sediment load–discharge relationships, measured curves were slightly below those estimated, though not significantly so at Blackwood and Ward Creeks in the period 1997–2002. Based on average daily flowrates determined from calibrated hydrologic modeling during the period 1994–2004, average daily flowrate frequency distributions per year are determined from which load reduction “crediting” towards TMDL targets can be evaluated. Despite seemingly similar estimated and measured sediment load–flow relationships, annual “estimated” loads exceeded those “measured” by about 40 % for Ward and Blackwood Creeks and over 300 times for HMR Creek. Similarly, though less dramatic, estimated annual nutrient loads at HMR Creek exceeded those measured by 1.7 and 6 times for total nitrogen and total phosphorus, respectively. Such results indicate that actual measured load–flow relationships are likely necessary for realistic quantitative and defensible TMDL crediting.     

Soil disturbance/restoration effects on stream sediment loading in the Tahoe Basin—detection monitoring

Quantifying the relative impacts of soil restoration or disturbance on watershed daily sediment and nutrients loads is essential towards assessing the actual costs/benefits of the land management. Such quantification requires stream monitoring programs capable of detecting changes in land-use or soil functional and erosive area “connectivity” conditions across the watershed. Previously, use of a local-scale, field-data based runoff and erosion model for three Lake Tahoe west-shore watersheds as a detection monitoring “proof of concept” suggested that analyses of midrange average daily flows can reveal sediment load reductions of relatively small watershed fractional areas (～5 %) of restored soil function within a few years of treatment. Developing such an effective stream monitoring program is considered for tributaries on the west shore of the Lake Tahoe Basin using continuous (15-min) stream monitoring information from Ward (2,521 ha), Blackwood (2,886 ha), and the Homewood (260 ha, HMR) Creek watersheds. The continuous total suspended sediment (TSS) and discharge monitoring confirmed the hysteretic TSS concentration—flowrate relationship associated with the daily and seasonal spring snowmelt hydrographs at all three creeks. Using the complete dataset, daily loads estimated from 1-h sampling periods during the day indicated that the optimal sampling hours were in the afternoon during the rising limb of the spring snowmelt hydrograph, an observation likely to apply across the Sierra Nevada and other snowmelt driven watersheds. Measured rising limb sediment loads were used to determine if soils restoration efforts (e.g., dirt road removal, ski run rehabilitation) at the HMR creek watershed reduced sediment loads between 2010 and 2011. A nearly 1.5-fold decrease in sediment yields (kg/ha per m³/s flow) was found suggesting that this focused monitoring approach may be useful towards development of TMDL “crediting” tools. Further monitoring is needed to verify these observations and confirm the value of this approach.     

Use of fish functional traits to associate in-stream suspended sediment transport metrics with biological impairment

Loss of ecological integrity due to excessive suspended sediment in rivers and streams is a major cause of water quality impairment in the USA. Current assessment protocols for development of sediment total maximum daily loads (TMDLs) lack a means to link temporally variable sediment transport rates with specific losses of ecological functions as loads increase. In order to accomplish this linkage assessment, a functional traits-based approach was used to correlate site occurrences of 17 fish species traits in three main groups (preferred rearing habitat, trophic feeding guild, and spawning behavior) with suspended sediment transport metrics. The sediment transport metrics included concentrations, durations, and dosages for a range of exceedance frequencies; and mean annual suspended sediment yields (SSY). In addition, this study in the Northwestern Great Plains Ecoregion examined trait relationships with three environmental gradients: channel stability, drainage area, and elevation. Potential stressor responses due to elevated suspended sediment concentration (SSC) levels were correlated with occurrences of five traits: preferred pool habitat; feeding generalists, omnivores, piscivores, and nest-building spawners; and development of ecologically based TMDL targets were demonstrated for specific SSC exceedance frequencies. In addition, reduced site occurrences for preferred pool habitat and nest-building spawners traits were associated with unstable channels and higher SSY. At an ecoregion scale, a functional traits assessment approach provided a means to quantify relations between biological impairment and episodically elevated levels of suspended sediment, supporting efforts to develop ecologically based sediment TMDLs.     

The magnitude of variability produced by methods used to estimate annual stormwater contaminant loads for highly urbanised catchments

Stormwater contaminant loading estimates using event mean concentration (EMC), rainfall/runoff relationship calculations and computer modelling (Model of Urban Stormwater Infrastructure Conceptualisation—MUSIC) demonstrated high variability in common methods of water quality assessment. Predictions of metal, nutrient and total suspended solid loadings for three highly urbanised catchments in Sydney estuary, Australia, varied greatly within and amongst methods tested. EMC and rainfall/runoff relationship calculations produced similar estimates (within 1 SD) in a statistically significant number of trials; however, considerable variability within estimates (～50 and ～25 % relative standard deviation, respectively) questions the reliability of these methods. Likewise, upper and lower default inputs in a commonly used loading model (MUSIC) produced an extensive range of loading estimates (3.8–8.3 times above and 2.6–4.1 times below typical default inputs, respectively). Default and calibrated MUSIC simulations produced loading estimates that agreed with EMC and rainfall/runoff calculations in some trials (4–10 from 18); however, they were not frequent enough to statistically infer that these methods produced the same results. Great variance within and amongst mean annual loads estimated by common methods of water quality assessment has important ramifications for water quality managers requiring accurate estimates of the quantities and nature of contaminants requiring treatment.     

Impact of sampling strategy on stream load estimates in till landscape of the Midwest

Accurately estimating various solute loads in streams during storms is critical to accurately determine maximum daily loads for regulatory purposes. This study investigates the impact of sampling strategy on solute load estimates in streams in the US Midwest. Three different solute types (nitrate, magnesium, and dissolved organic carbon (DOC)) and three sampling strategies are assessed. Regardless of the method, the average error on nitrate loads is higher than for magnesium or DOC loads, and all three methods generally underestimate DOC loads and overestimate magnesium loads. Increasing sampling frequency only slightly improves the accuracy of solute load estimates but generally improves the precision of load calculations. This type of investigation is critical for water management and environmental assessment so error on solute load calculations can be taken into account by landscape managers, and sampling strategies optimized as a function of monitoring objectives.     

A low cost method to estimate dissolved reactive phosphorus loads of rivers and streams

Concentration of dissolved reactive phosphorus ([DRP]) in rivers changes periodically (daily, weekly, seasonally) and is dependent on the weather and discharge Q. Accordingly, accurate estimation of the annual DRP load requires intensive sampling if not even continuous monitoring, which is laborious and expensive. We present a new, elaborated low cost technique based on passive samplers (P-traps), describing their design and chemical analysis. P-traps use iron(oxy)hydroxide as a sorbent, are inexpensive, easy to handle, and can be exposed for several months. We compare average DRP concentrations obtained from spot samples and P-traps and discuss the applicability and accuracy of the suggested method to measure annual P loads of rivers characterized by highly variable DRP concentrations.     

Evaluation of phosphorus loads from Ontario municipal wastewater treatment plants

Results are presented from an evaluation of the nature and accuracy of phosphorus loads discharged by Ontario municipal wastewater treatment plants into the Great Lakes. Data were examined for the 96 plants treating flows in excess of 4546 m³d^-1 for the period 1981 to 1985. For the Lake Erie, Lake Ontario/St. Lawrence River and Upper Great Lakes basins, total basin phosphorus loads were classified according to type of wastewater treatment system, the type of chemical added for phosphorus removal, plant capacity, and sampling frequency. Load estimation techniques were compared using the 1985 daily data from three representative treatment plants. Annual phosphorus loads were compared using the complete data records for the plants and using Monte Carlo techniques to simulate an incomplete data record typical of once per week effluent phosphorus sampling. Potential sources of bias were identified in annual phosphorus load estimates from municipal treatment plants.     

Improving suspended sediment measurements by automatic samplers

Suspended solids either as total suspended solids (TSS) or suspended sediment concentration (SSC) is an integral particulate water quality parameter that is important in assessing particle-bound contaminants. At present, nearly all stormwater runoff quality monitoring is performed with automatic samplers in which the sampling intake is typically installed at the bottom of a storm sewer or channel. This method of sampling often results in a less accurate measurement of suspended sediment and associated pollutants due to the vertical variation in particle concentration caused by particle settling. In this study, the inaccuracies associated with sampling by conventional intakes for automatic samplers have been verified by testing with known suspended sediment concentrations and known particle sizes ranging from approximately 20 μm to 355 μm under various flow rates. Experimental results show that, for samples collected at a typical automatic sampler intake position, the ratio of sampled to feed suspended sediment concentration is up to 6600% without an intake strainer and up to 300% with a strainer. When the sampling intake is modified with multiple sampling tubes and fitted with a wing to provide lift (winged arm sampler intake), the accuracy of sampling improves substantially. With this modification, the differences between sampled and feed suspended sediment concentration were more consistent and the sampled to feed concentration ratio was accurate to within 10% for particle sizes up to 250 μm.     

Field assessment of flood event suspended sediment transport from ephemeral streams in the tropical semi-arid catchments

An assessment of suspended sediment transport was carried out in a number of semiarid catchments during flood events in order to quantify the degradation rates. In order to quantify these, a systematic sampling procedure of the episodic flood events was proposed for representative catchments. The procedure allows for an integration over the whole run-off episode using both the rising and falling limbs of the run-off hydrograph to compute the sediment quantities for each individual flood event. Higher sediment concentrations occurred in the rising limb than those at the recession for any stage of flow. The maximum suspended sediment concentration was observed at the peak of the flood hydrograph. An integration of the sediment concentration over its duration gave the total sediment yield from the flood event. For the ephemeral channels, only a small number of flood events were observed over a three-year experimental period each with a duration of the order of 3–6 h. It is notable that high sediment loads were associated with high flow volumes which were effectively the result of the catchment characteristics and incident rainfall causing the flood events in the respective catchments. A large percentage of the annual sediment yield from a catchment is transported by the ephemeral streams during a small number of flood events. The correct determination of the total sediment yield from any of the flood events depends entirely on the accuracy of the measurements. The understanding of run-off and sediment loss for the representative catchments aims at assisting planning, management and control of water and land resources for sustainable development in the semi-arid parts of the tropics. The sediment rates reveal the degradation of catchments which have repercussions on the crop and pasture production and this has a bearing on the soil and water conservation programmes in the delicate ecological balance of the semi-arid areas. Further, these rates will determine the lifespan of the reservoirs planned for the dry river valleys (ephemeral streams) and existing ones which serve livestock and domestic needs. These occasionally will require costly rehabilitation and scooping to increase effective storage unless conservation measures are taken, and these measures are bound to vary from place to place as per the representative catchments output.     

Spatial and temporal variation in suspended sediment, organic matter, and turbidity in a Minnesota prairie river: implications for TMDLs

The Minnesota River Basin (MRB), situated in the prairie pothole region of the Upper Midwest, contributes excessive sediment and nutrient loads to the Upper Mississippi River. Over 330 stream channels in the MRB are listed as impaired by the Minnesota Pollution Control Agency, with turbidity levels exceeding water quality standards in much of the basin. Addressing turbidity impairment requires an understanding of pollutant sources that drive turbidity, which was the focus of this study. Suspended volatile solids (SVS), total suspended solids (TSS), and turbidity were measured over two sampling seasons at ten monitoring stations in Elm Creek, a turbidity impaired tributary in the MRB. Turbidity levels exceeded the Minnesota standard of 25 nephelometric units in 73% of Elm Creek samples. Turbidity and TSS were correlated (r ²?=?0.76), yet they varied with discharge and season. High levels of turbidity occurred during periods of high stream flow (May–June) because of excessive suspended inorganic sediment from watershed runoff, stream bank, and channel contributions. Both turbidity and TSS increased exponentially downstream with increasing stream power, bank height, and bluff erosion. However, organic matter discharged from wetlands and eutrophic lakes elevated SVS levels and stream turbidity in late summer when flows were low. SVS concentrations reached maxima at lake outlets (50 mg/l) in August. Relying on turbidity measurements alone fails to identify the cause of water quality impairment whether from suspended inorganic sediment or organic matter. Therefore, developing mitigation measures requires monitoring of both TSS and SVS from upstream to downstream reaches.     

Determining storm sampling requirements for improving precision of annual load estimates of nutrients from a small forested watershed

This study sought to determine the lowest number of storm events required for adequate estimation of annual nutrient loads from a forested watershed using the regression equation between cumulative load (∑L) and cumulative stream discharge (∑Q). Hydrological surveys were conducted for 4 years, and stream water was sampled sequentially at 15-60-min intervals during 24 h in 20 events, as well as weekly in a small forested watershed. The bootstrap sampling technique was used to determine the regression (∑L-∑Q) equations of dissolved nitrogen (DN) and phosphorus (DP), particulate nitrogen (PN) and phosphorus (PP), dissolved inorganic nitrogen (DIN), and suspended solid (SS) for each dataset of ∑L and ∑Q. For dissolved nutrients (DN, DP, DIN), the coefficient of variance (CV) in 100 replicates of 4-year average annual load estimates was below 20% with datasets composed of five storm events. For particulate nutrients (PN, PP, SS), the CV exceeded 20%, even with datasets composed of more than ten storm events. The differences in the number of storm events required for precise load estimates between dissolved and particulate nutrients were attributed to the goodness of fit of the ∑L-∑Q equations. Bootstrap simulation based on flow-stratified sampling resulted in fewer storm events than the simulation based on random sampling and showed that only three storm events were required to give a CV below 20% for dissolved nutrients. These results indicate that a sampling design considering discharge levels reduces the frequency of laborious chemical analyses of water samples required throughout the year.     

A simplified sampling procedure for the estimation of methane emission in rice fields

Manual closed chamber methods are widely used for CH₄ measurement from rice paddies. Despite diurnal and seasonal variations in CH₄ emissions, fixed sampling times, usually during the day, are used. Here, we monitored CH₄ emission from rice paddies for one complete rice-growing season. Daytime CH₄ emission increased from 0800 h, and maximal emission was observed at 1200 h. Daily averaged CH₄ flux increased during plant growth or fertilizer application and decreased upon drainage of plants. CH₄ measurement results were linearly interpolated and matched with the daily averaged CH₄ emission calculated from the measured results. The time when daily averaged emission and the interpolated CH₄ curve coincided during the daytime was largely invariant within each of the five distinctive periods. One-hourly sampling during each of these five periods was utilized to estimate the emission during each period, and we found that five one-hourly samples during the season accurately reflected the CH₄ emission calculated based on all 136 hourly samples. This new sampling scheme is simple and more efficient than current sampling practices. Previously reported sampling schemes yielded estimates 9 to 32% higher than the measured CH₄ emission, while our suggested scheme yielded an estimate that was only 5% different from that based on all 136-h samples. The sampling scheme proposed in this study can be used in rice paddy fields in Korea and extended worldwide to countries that use similar farming practices. This sampling scheme will help in producing more accurate global methane budget from rice paddy fields.     

Influence of different nitrate–N monitoring strategies on load estimation as a base for model calibration and evaluation

Model-based predictions of the impact of land management practices on nutrient loading require measured nutrient flux data for model calibration and evaluation. Consequently, uncertainties in the monitoring data resulting from sample collection and load estimation methods influence the calibration, and thus, the parameter settings that affect the modeling results. To investigate this influence, we compared three different time-based sampling strategies and four different load estimation methods for model calibration and compared the results. For our study, we used the river basin model Soil and Water Assessment Tool on the intensively managed loess-dominated Parthe watershed (315 km²) in Central Germany. The results show that nitrate–N load estimations differ considerably depending on sampling strategy, load estimation method, and period of interest. Within our study period, the annual nitrate–N load estimation values for the daily composite data set have the lowest ranges (between 9.8% and 15.7% maximum deviations related to the mean value of all applied methods). By contrast, annual estimation results for the submonthly and the monthly data set vary in greater ranges (between 24.9% and 67.7%). To show differences between the sampling strategies, we calculated the percentage deviation of mean load estimations of submonthly and monthly data sets as related to the mean estimation value of the composite data set. For nitrate–N, the maximum deviation is 64.5% for the submonthly data set in the year 2000. We used average monthly nitrate–N loads of the daily composite data set to calibrate the model to achieve satisfactory simulation results [Nash–Sutcliffe efficiency (NSE) 0.52]. Using the same parameter settings with submonthly and monthly data set, the NSE dropped to 0.42 and 0.31, respectively. Considering the different results from the monitoring strategy and the load estimation method, we recommend both the implementation of optimized monitoring programs and the use of multiple load estimation methods to improve water quality characterization and provide appropriate model calibration and evaluation data.     

Siting analyses for water quality sampling in a catchment

Pollution loads discharged from upstream development or human activities significantly degrade the water quality of a reservoir. The design of an appropriate water quality sampling network is therefore important for detecting potential pollution events and monitoring pollution trends. However, under a limited budgetary constraint, how to site an appropriate number of sampling stations is a challenging task. A previous study proposed a method applying the simulated annealing algorithm to design the sampling network based on three cost factors including the number of reaches, bank length, and subcatchment area. However, these factors are not directly related to the distribution of possible pollution. Thus, this study modified the method by considering three additional factors, i.e. total phosphorus, nitrogen, and sediment loads. The larger the possible load, the higher the probability of a pollution event may occur. The study area was the Derchi reservoir catchment in Taiwan. Pollution loads were derived from the AGNPS model with rainfall intensity estimated using the Thiessen method. Analyses for a network with various numbers of sampling sites were implemented. The results obtained based on varied cost factors were compared and discussed. With the three additional factors, the chosen sampling network is expected to properly detect pollution events and monitor pollution distribution and temporal trends.     

AgInput: An Agricultural Nutrient and Pesticide Source Model

Agriculture can be a major nonpoint source (NPS) of nutrient and pesticide contamination in the environment. Available databases do not provide accurate and dynamic data on fertilizer and pesticide application, which limits the ability of complex watershed models to simulate contaminant loads into impaired water bodies. A model for estimating agricultural nutrient and pesticide input for watershed modeling has been developed. Climate, soils, and major agricultural operations are considered within the model, so that it can be adapted to any watershed or subregion within a watershed. The timing of the agricultural operations is a function of the weather data, providing realistic results at daily, monthly, or annual application rates. The model also predicts irrigation demand and biomass production, which can be used to calibrate the model. Model output can be used in any watershed model that considers agricultural land uses. Two case studies were evaluated, using grape vineyards in the Napa River and strawberry production in Newport Bay as examples. The predicted time to maturity corresponded well with actual data. Irrigation and fertilizer needs were very sensitive to weather input. Although the model can generate weather from long-term averages, the simulated results are best when at least observed precipitation and temperature are provided, to capture extreme events. The model has data for 98 crops and 126 pesticides, based on the California Department of Pesticide Regulation database. The databases are easily modifiable by the user to adapt them to local conditions. The output from AgInput is much needed for watershed modeling and for development of total maximum daily loads (TMDLs), based on realistic targets of irrigation, nutrient, and pesticide inputs. The model is available for free download at .     

Estimation of the Human Impact on Nutrient Loads Carried by the Elbe River

The reunification of Germany led to dramatically reduced emissions of nitrogen (N) and phosphorus (P) to the environment. The aim of the present study was to examine how these exceptional decreases influenced the amounts of nutrients carried by the Elbe River to the North Sea. In particular, we attempted to extract anthropogenic signals from time series of riverine loads of nitrogen and phosphorus by developing a normalization technique that enabled removal of natural fluctuations caused by several weather-dependent variables. This analysis revealed several notable downward trends. The normalized loads of total-N and NO3-N exhibited an almost linear trend, even though the nitrogen surplus in agriculture dropped dramatically in 1990 and then slowly increased. Furthermore, the decrease in total-P loads was found to be considerably smaller close to the mouth of the river than further upstream. Studying the predictive ability of different normalization models showed the following: (i) nutrient loads were influenced primarily by water discharge; (ii) models taking into account water temperature, load of suspended particulate matter, and salinity were superior for some combinations of sampling sites and nutrient species; semiparametric normalization models were almost invariably better than ordinary regression models.     

Influence of High Levels of Total Suspended Solids on Measurement of Cod and Bod in the Yellow River, China

Using the Yellow River, China, the study explores the problem of the use of COD and BOD₅ as water quality management parameters in the presence of very high levels of suspended sediment (TSS) that characterize this river. Although the amount of natural organic matter per unit of suspended sediment of the Yellow River is not high, the very high concentration of mineral sediment in the Yellow River results in a large concentration of organic matter, which artificially inflates the laboratory values of COD and, as a consequent, leads to greatly exaggerated reports of pollution of the Yellow River. BOD₅ can more accurately reflect the pollution of the Yellow River than COD; however, measured values of BOD under-report the actual values due to settling of the sediment in the incubation chamber resulting in values that are 21.6--38.3% less than the actual values. Therefore corrections are required for laboratory COD and BOD values so that the values are not artifacts of the sediment regime. Our work provides new insight into this phenomenon and demonstrates how correction factors may be determined and used with pollution data. Our work also suggests that the actual pollution levels of the Yellow River are probably not as high as reported by monitoring agencies.     

Analysis of the air pollution climate at a background site in the Po valley

The Po valley in northern Italy is renowned for its high air pollutant concentrations. Measurements of air pollutants from a background site in Modena, a town of 200 thousand inhabitants within the Po valley, are analysed. These comprise hourly data for CO, NO, NO(2), NO(x), and O(3), and daily gravimetric equivalent data for PM(10) from 1998-2010. The data are analysed in terms of long-term trends, annual, weekly and diurnal cycles, and auto-correlation and cross-correlation functions. CO, NO and NO(2) exhibit a strongly traffic-related pattern, with daily peaks at morning and evening rush hour and lower concentrations over the weekend. Ozone shows an annual cycle with a peak in July due to local production; notwithstanding the diurnal cycle dominated by titration by nitrogen oxide, the decreasing long term trend in NO concentration did not affect the long term trend in O(3), whose mean concentration remained steady over the sampling period. PM(10) shows a strong seasonality with higher concentration in winter and lower concentration in summer and spring. Both PM(10) and ozone show a marked weekly cycle in summer and winter respectively. Regressions of PM(10) upon NO(x) show a consistently greater intercept in winter, representing higher secondary PM(10) in the cooler months of the year. There is a seasonal pattern in primary PM(10) to NO(x) ratios, with lower values in winter and higher values in summer, but the reasons are unclear.     

Phosphorus Load Reduction Goals for Feitsui Reservoir Watershed, Taiwan

The present paper describes an effort for developing the total maximum daily load (TMDL) for phosphorus and a load reduction strategy for the Feitsui Reservoir in Northern Taiwan. BASINS model was employed to estimate watershed pollutant loads from nonpoint sources (NPS) in the Feitsui Reservoir watershed. The BASINS model was calibrated using field data collected during a 2-year sampling period and then used to compute watershed pollutant loadings into the Feitsui Reservoir. The simulated results indicate that the average annual total phosphorus (TP) loading into the reservoir is 18,910 kg/year, which consists of non-point source loading of 16,003 kg/year, and point source loading of 2,907 kg/year. The Vollenweider mass balance model was used next to determine the degree of eutrophication under current pollutant loading and the load reduction needed to keep the reservoir from being eutrophic. It was estimated that Feitsui Reservoir can becoming of the oligotrophic state if the average annual TP loading is reduced by 37% or more. The results provide the basis on which an integrated control action plan for both point and nonpoint sources of pollution in the watershed can be developed.