H51D-0743
Comparison of Erosion Rates Estimated by Sediment Budget Techniques and Suspended Sediment Monitoring and Regulatory Implications
Watersheds in the northern California Coast Range have been designated as "impaired" with respect to water quality because of excessive sediment loads and/or high water temperature. Sediment budget techniques have typically been used by regulatory authorities to estimate current erosion rates and to develop targets for future desired erosion rates. This study examines erosion rates estimated by various methods for portions of the Gualala River watershed, designated as having water quality impaired by sediment under provisions of the Clean Water Act Section 303(d), located in northwest Sonoma County (~90 miles north of San Francisco). The watershed is underlain by Jurassic age sedimentary and meta-sedimentary rocks of the Franciscan formation. The San Andreas Fault passes through the western edge of watershed, and other active faults are present. A substantial portion of the watershed is mantled by rock slides and earth flows, many of which are considered dormant. The Coast Range is geologically young, and rapid rates of uplift are believed to have contributed to high erosion rates. This study compares quantitative erosion rate estimates developed at different spatial and temporal scales. It is motivated by a proposed vineyard development project in the watershed, and the need to document conditions in the project area, assess project environmental impacts and meet regulatory requirements pertaining to water quality. Erosion rate estimates were previously developed using sediment budget techniques for relatively large drainage areas (~100 to 1,000 km2) by the North Coast Regional Water Quality Control Board and US EPA and by the California Geological Survey. In this study, similar sediment budget techniques were used for smaller watersheds (~3 to 8 km2), and were supplemented by a suspended sediment monitoring program utilizing Turbidity Threshold Sampling techniques (as described in a companion study in this session). The duration of the monitoring program to date spanned the winter runoff seasons of Water Years 2006 and 2007. These were unusually wet and dry years, respectively, providing perspective on the range of measured sediment yield in relation to sediment budget estimates. The measured suspended sediment yields were substantially lower than predicted by sediment budget methods. Variation in geomorphic processes over time and space and methodological problems of sediment budgets may be responsible for these apparent discrepancies. The implications for water quality policy are discussed.
H51D-0744
Suspended Sediment Monitoring Strategies Reduce Model Uncertainties
Regulatory agencies require development and implementation of Total Maximum Daily Loads for watersheds listed under section 303d of the Clean Water Act. For rivers identified as sediment impaired, methods are required to identify sediment sources and to estimate loading capacities, and models, such as sediment budgets, are often employed. Models can be tested and improved when stream monitoring provides accurate estimates of sediment loads. Motivated by proposed vineyard development on forested ridges in a sediment-listed watershed in the Coast Range of northern California, four tributaries to Gualala River ranging in size from 300 to 800 ha2 were monitored over two winter seasons. More than 1850 samples were analyzed for suspended sediment concentration. Inter- annual variability of sediment loads from wet and dry years is compared. Traditional methods for estimating suspended sediment loads often rely on measurements, such as water discharge, that are not well correlated to sediment concentration due to the highly variable routing of sediment to the channel from hillslopes, roads, and landslides. A method, such as Turbidity Threshold Sampling, that employs a parameter well correlated to suspended sediment concentration can improve sampling efficiency by collecting samples that are distributed over a range of rising and falling concentrations. The resulting set of samples can be used to estimate sediment loads by establishing a relationship between concentration and turbidity for any sampled period and applying it to the continuous turbidity record. http://rivermetrics.com
H51D-0745
A Long Island Sound-Specific Water Quality Index Based on Cluster Analysis and Discriminant Analysis
A new water quality index (WQI) is computed using multivariate cluster analysis and discriminant analysis of a set of individual water quality indicators (e.g., chlorophyll-a, total dissolved nitrogen, total dissolved phosphorus, and dissolved oxygen) collected at different sampling stations. A numerical value (-1 to 1) results, with a value close to 1 indicating good water quality (oligotrophic), a value close to -1 indicating poor water quality (eutrophic), and a slight negative value representing mesotrophic conditions. This new water quality index provides a simple framework that overcomes several shortcomings of the traditional water quality index rating. The method is applied to Long Island Sound (LIS) water quality data (past 15 years at about 20 stations) collected by the Connecticut Department of Environmental Protection. There is a general trend of increasing WQI (increasing water quality) from west-LIS to east-LIS. For instance, the WQI values for station A4 (the most west station in LIS) are typically between -0.5 and -1 (indicating eutrophic conditions), whereas the WQI values for stations J2, K2, M3, N3 (east-LIS) are mostly between 0.5 and 1 (indicating oligotrophic conditions). These WQI values are consistent with the fact that west-LIS is more polluted (water quality more degraded) than mid- and east-LIS. There are some seasonal trends in the averaged WQI values. For instance, the WQI values increase from April to September for the east LIS stations, indicating that the water quality at these stations improves over this period. In contrast, the WQI values for the west to mid LIS stations generally decrease from June to August, indicating that the water quality degrades during this period.
H51D-0746
Using NASA and Earth Science Products to Improve EPA Non-point Source Water Quality Modeling for the Chesapeake Bay
The Environmental Protection Agency (EPA) estimates that over 20,000 bodies of water throughout the country do not meet water quality standards. Nonpoint sources -- pollution from urban, agricultural, and forest land that is transported by runoff -- typically cause 90 percent of impairments. EPA has developed the BASINS (Better Assessment Science Integrating Point and Nonpoint Sources) modeling system for performing numerous water quality studies. The key to this suite of models is the Hydrological Simulation Program - Fortran (HSPF), which calculates daily stream flow rates and the corresponding pollutant concentrations at the watershed outlet. EPA has partnered with NASA to use high spatial and temporal hydrological variables (e.g., precipitation, evaporation, etc.) from the NASA Land Information System (LIS) and land cover/vegetative indices derived from primarily MODIS and Landsat satellite data non-point source water quality for the Chesapeake Bay Basin. For the precipitation and evaporation data, EPA-based BASINS-HSPF streamflow runs were conducted on seven study watersheds in the Chesapeake Bay Basin. Sets of runs using precipitation from default weather stations, the NASA LIS 1/8th degree precipitation, NOAA Stage IV precipitation, NASA LIS Noah land surface model evapotranspiration datasets were conducted for each watershed. The output statistics summarized reveal that for 74% of the runs, the NASA LIS 1/8th degree and Stage IV precipitation-based runs performed better than when using only the default EPA precipitation station data. In addition, an automatic calibration method (‘PEST') and Noah land surface model evapotranspiration (ET) being further incorporated. The empirical ability of generalized spectral indices and land cover derived from Landsat and MODIS was tested for predicting stream water nitrogen export from predominately forested watersheds undergoing disturbance. The disturbance index, a summary index that is easily computed from Landsat Tasseled Cap bands was found to have a good predictive power identifying spatial variability in N export caused by a severe gypsy moth defoliation event (R2 = 0.45, p = 0.002, N = 18 watersheds). A novel index that directly relates the magnitude of forest disturbance was derived to inter-annual changes in wetness. This more physically-based and generalized index also showed a good ability to predict N export (R2 = 0.38, p = 0.006, N = 18 watersheds). The MODIS sensor appears to perform better than the Landsat-based indices (R2 = 0.48, p = 0.001, N = 18 watersheds, a finding which may be attributed to the enhanced capacity of MODIS to match the date(s) of image collection to the peak of a disturbance event. In a modeling environment such as BASINS-HSPF, additional variables would be included in the modeling that may improve predictive ability. These preliminary results show that alternative remote sensing variables have the capacity to improve predictions of watershed N whereas standard LULC products by EPA and related groups may not.
H51D-0747
Probabilistic Evaluation of the Vegetated Filter Strip Efficiency With Respect to Pathogen Removal From Runoff
Vegetated filter strips (VFS) have become an important component of the water quality improvement in watersheds. Relatively little is known about the efficiency of VFSs in retention of manure-borne pathogens that can be important agricultural pollutants. The objective of this work was to evaluate the uncertainty in the VFS efficiency caused by variations in major pathogen retention factors, i.e. vegetation status, soil infiltration capacity, and rainfall intensity and duration. We have developed the model STIR to simulate the overland transport and loss to infiltration of manure-borne pathogens in VFS. This model was used in Latin Hypercube sampling-based simulations to evaluate the possible variability in pathogenic bacteria breakthrough in VFSs. The sampled model parameters were rainfall intensity, rainfall duration, initial soil water content, Manning roughness coefficient, saturated hydraulic conductivity, shape parameters of soil water retention curve, pathogen partitioning coefficient, pathogen attachment rate, pathogen straining rate, and the dispersivity in the overland flow. Probability distributions of the parameters were selected to represent the USDA-ARS experimental site in Beltsville, MD, where the STIR model was tested with data on E. coli bacteria. Relatively long high-intensity rainfalls, low hydraulic conductivities, high soil moisture contents before the rainfall, and high dispersivities were the main reasons of the strip partial failure. However, the unfavorable conditions and their combinations were relatively rare, and the removal efficiency of the 6-m long VFS at the edge of 200-m long field was less than 100 % for 5 % of simulation scenarios, and less than 75 % for 2.5 % of simulation scenarios. The probabilistic characterization of the VFS efficiency with site-specific soil and weather properties can be useful in making decisions on VFS placement with respect to manure-borne pathogens.
H51D-0748
Development and application of a coupled bio-geochmical and hydrological model for point and non-point source river water pollution
The aim of this paper is to present recent developments of an integrated water- and N-balance model for the assessment of land use changes on water and N-fluxes for meso-scale river catchments. The semi-distributed water-balance model SWAT was coupled with algorithms of the bio-geochemical model DNDC as well as the model CropSyst. The new model that is further denoted as SWAT-N was tested with leaching data from a long- term lysimeter experiment as well as results from a 5-years sampling campaign that was conducted at the outlet of the meso-scale catchment of the River Dill (Germany). The model efficiency for N-load as well as the spatial representation of N-load along the river channel that was tested with results taken from longitudinal profiles show that the accuracy of the model has improved due to the integration of the aforementioned process-oriented models. After model development and model testing, SWAT-N was then used for the assessment of the EU agricultural policy (CAP reform) on land use change and consequent changes on N-fluxes within the Dill Catchment. http://geb.uni- giessen.de/geb/volltexte/2007/4531/
H51D-0749
Characterization of hyporheic exchange, infiltration and exfiltration patterns at the catchment scale
Field experiments have highlighted runoff contribution and nutrient and pollutant mobilization from hillslopes during rainfall events. This mobilization reduces the amount of minerals availability for plant and tree growth and soil ecosystems in general or affects stream ecosystems by transporting pollutants to the stream network. The saturated zone surrounding the stream network, or hyporheic zone (HZ), has the ability to mix with stream water. This mixing plays a critical role in reducing the effects of the pollutant or nutrient mobilization occurring during infiltration and runoff processes and provides stream organisms with nutrients and minerals. The hyporheic exchanges have been broadly described, highlighting processes occurring from the pore scale to few tenths of kilometers in length. Investigating in the HZ usually involves tracer experiments that are time and labor intensive and the outcome might have limited value across variable systems due to site specificity of these experiments. We propose a systematic approach based on reach scale tracer experiments to assess the ability of a given hyporheic system to retain nutrients released by its surrounding hillslopes, using readily available DEM-derived topographic parameters. The stream network of a 3.3km2 snow-dominated watershed has been split into 20 reaches each roughly 200m in length to investigate short time-space scale subsurface flow-paths within the HZ. For every reach, NaCl was injected into the stream at the upper boundary, following the slug injection method and the resulting breakthrough curve was recorded at its lower boundary. Three runs each consisting of 20 reach scale experiments were carried out at various hydrological regimes in order to investigate the influence of stream discharge on hyporheic exchanges. The OTIS model, based on a first order mass transfer between the HZ and the stream water was applied and Monte Carlo simulations were carried out to derive the best set of parameters and their related uncertainty that would characterize the hyporheic exchanges for a given reach (i.e. cross section of the HZ and mass transfer coefficient). The first step of the analysis highlights the spatiotemporal variability of the hyporheic exchanges from the headwaters to the outlet of the whole basin. In a second step, a potential correlation is investigated between the hyporheic exchange parameters output of OTIS and the topographic parameters derived from a DEM.
H51D-0750
A Geochemical Reaction Model for Titration of Contaminated Soil and Groundwater at the Oak Ridge Reservation
This study investigates geochemical reactions during titration of contaminated soil and groundwater at the Oak Ridge Reservation in eastern Tennessee. The soils and groundwater exhibits low pH and high concentrations of aluminum, calcium, magnesium, manganese, various trace metals such as nickel and cobalt, and radionuclides such as uranium and technetium. The mobility of many of the contaminant species diminishes with increasing pH. However, base additions to increase pH are strongly buffered by various precipitation/dissolution and adsorption/desorption reactions. The ability to predict acid-base behavior and associated geochemical effects is thus critical to evaluate remediation performance of pH manipulation strategies. This study was undertaken to develop a practical but generally applicable geochemical model to predict aqueous and solid-phase speciation during soil and groundwater titration. To model titration in the presence of aquifer solids, an approach proposed by Spalding and Spalding (2001) was utilized, which treats aquifer solids as a polyprotic acid. Previous studies have shown that Fe and Al-oxyhydroxides strongly sorb dissolved Ni, U and Tc species. In this study, since the total Fe concentration is much smaller than that of Al, only ion exchange reactions associated with Al hydroxides are considered. An equilibrium reaction model that includes aqueous complexation, precipitation, ion exchange, and soil buffering reactions was developed and implemented in the code HydroGeoChem 5.0 (HGC5). Comparison of model results with experimental titration curves for contaminated groundwater alone and for soil- water systems indicated close agreement. This study is expected to facilitate field-scale modeling of geochemical processes under conditions with highly variable pH to develop practical methods to control contaminant mobility at geochemically complex sites.
H51D-0751
Modelling the Impact of River Morphology on Nitrogen Retention
Denitrification is an important sink of nitrogen in riverine systems and mainly linked to interstitial sediments and hence to river morphology. Studies which quantify this relationship are rare. In this paper we analyse the effect of river morphology on nitrogen load using the river water quality model WASP5. The description of the denitrification was modified to provide a dependency with a segment-specific sediment surface area. The revised model was applied to a 70.6 km river reach of the 4th order river Weisse Elster (Germany) and measurements of the relevant water constituents were carried out for summer low-flow conditions (the Lagrange approach and 24h diurnal measurements). The river bed morphology is characterised by 876 cross sections. The program UNCSIM was used to determine parameter identifiability, which is based on the sensitivity and compensation effect of parameters. The uncertainty analysis was carried out using a Monte-Carlo-Analysis with Latin Hypercube Sampling. Model validation was reasonable with Nash-Suttcliffe efficiencies of 0.89 for nitrate-nitrogen, 0.30 for ammonium-nitrogen and 0.86 for phytoplankton concentrations. It was shown that nitrogen retention amounted to 23.4% of the nitrogen load of the upper boundary and benthic denitrification was identified as the largest contributor to that sink. The retention varies significantly along the river section with amounts being almost 2.4 times higher in a natural reach compared with a heavy modified and channelized river section. The mean denitrification rate was 189 mg N/(m2d). A sensitivity analysis indicates that the sinuosity is the most sensitive morphological factor on nitrogen loading, a 10% increase of which causes a 2.4% decrease in inorganic nitrogen loading. The impact of river structure restoration on nitrogen retention is relatively low. The model scenario representing the implementation of the most feasible measures can lead to an additional nitrogen load reduction of 5.4%. There is a strong need to further develop the modelled denitrification processes. We emphasize the need to include variable denitrification rates which depend on the sediment characteristics and the hydraulic exchange. These site-specific characteristics depend mainly on the variation of the hydraulic radius, i.e. on river width and depth, flow velocity and channel slope.
H51D-0752
Data Collection for Dynamic Temperature Modeling in High-Gradient Mountain Streams
Neilson [2006] published results from a study in the Virgin River, UT where a data collection methodology was developed to assist in modeling the separate effects of hyporheic and dead zones on heat and solute transport. This study was unique in that temperature and tracer data were collected in the main channel, hyporheic zone, and dead zones to help estimate parameters associated with a two zone modeling approach rather than previous one zone modeling approaches that lump the effects of hyporheic and dead zones (transient storage). Research on a small section of Curtis Creek, UT, USA, has begun to investigate whether the data collection methodology and modeling approach developed by Neilson in the Virgin River, a desert river system with sand/gravel substrate and relatively low average channel slopes, could be implemented in a mountain stream with gravel/cobble substrate and higher average channel slopes that are highly influenced by groundwater. Initial data suggested that instream temperatures in this small high-gradient steam are significantly affected by riparian shading, deep groundwater, surface groundwater seeps, hyporheic exchange, and bed conduction. Therefore, new data types were collected in the current study in order to begin to separate out some of the complex and confounded sources and sinks of heat found in mountain streams that need to be quantified for temperature model population and testing. These included riparian shading mapping, detailed channel surveys, groundwater observation wells, and an increased number of tracer tests to assist in quantifying groundwater influx through dilution studies. Additionally, new installation techniques had to be applied for equipment inserted in the bed sediments. Preliminary data, analyses, and installation techniques will be presented. Neilson, B. T. (2006), Dynamic Stream Temperature Modeling: Understanding the Causes and Effects of Temperature Impairments and Uncertainty in Predictions, Dissertation thesis, Utah State University, Logan, UT.
H51D-0753
Uncertainty Of Stream Nutrient Transport Estimates Using Random Sampling Of Storm Events From High Resolution Water Quality And Discharge Data
The uncertainties associated with stream nutrient transport estimates are frequently overlooked and the sampling strategy is rarely if ever investigated. Indeed, the impact of sampling strategy and estimation method on the bias and precision of stream phosphorus (P) transport calculations is little understood despite the use of such values in the calibration and testing of models of phosphorus transport. The objectives of this research were to investigate the variability and uncertainty in the estimates of total phosphorus transfers at an intensively monitored agricultural catchment. The Oona Water which is located in the Irish border region, is part of a long term monitoring program focusing on water quality. The Oona Water is a rural river catchment with grassland agriculture and scattered dwelling houses and has been monitored for total phosphorus (TP) at 10 min resolution for several years (Jordan et al, 2007). Concurrent sensitive measurements of discharge are also collected. The water quality and discharge data were provided at 1 hour resolution (averaged) and this meant that a robust estimate of the annual flow weighted concentration could be obtained by simple interpolation between points. A two-strata approach (Kronvang and Bruhn, 1996) was used to estimate flow weighted concentrations using randomly sampled storm events from the 400 identified within the time series and also base flow concentrations. Using a random stratified sampling approach for the selection of events, a series ranging from 10 through to the full 400 were used, each time generating a flow weighted mean using a load-discharge relationship identified through log-log regression and monte-carlo simulation. These values were then compared to the observed total phosphorus concentration for the catchment. Analysis of these results show the impact of sampling strategy, the inherent bias in any estimate of phosphorus concentrations and the uncertainty associated with such estimates. The estimates generated using the full time series underestimate the flow weighted mean concentration of total phosphorus. This work compliments other contemporary work in the area of load estimation uncertainty in the UK (Johnes, 2007). Johnes P,J. 2007, Uncertainties in annual riverine phosphorus load estimation: Impact of load estimation methodology, sampling frequency, baseflow index and catchment population density, Journal of hydrology 332 (1- 2): 241-258 Jordan, P., Arnscheidt, J., McGrogan, H & McCormick, S., 2007. Characterising phosphorus transfers in rural transfers using a continuous bank-side analyser. Hydrology and Earth System Science 11, 372-381 Kronvang B & Bruhn, A. J, 1996. Choice of sampling strategy and estimation method for calculating nitrogen and phosphorus transport in small lowland streams , Hydrological processes 10 (11): 1483-1501