Recent Presentations

The development and application of models to predict in-river concentrations of ingredients used in personal care products are important to the environmental risk assessment process. While suitable models exist in the US and Europe, there are currently no available models to estimate exposure of personal care products in China. To this end, work was performed to create a GIS-based system that develops exposure estimates used to predict the emission of personal care products into freshwater ecosystems. Currently, many of the available inputs for the generation of environmental concentrations to be used in the exposure assessment are at a very coarse spatial scale, in some cases only a single value for the entire country is available. In this study, environmental concentrations in surface water were estimated at the province and county-levels (ca. 2,900 counties in China) based on both user-supplied product/ingredient information, as well as other geographically-linked socio-economic and environmental information from official census and other data sources.

Product information such as category (e.g., hair, skin, etc.), composition (e.g., ingredient fraction) and “take off” values (GDP threshold under which the product would not likely be purchased) were used to distribute total tons of individual ingredients based on product sales data in China. These data were combined with economic information, population density (including urban and rural separation), water use, disposal mechanism (e.g., STP, septic, direct discharge to river, etc.), and location-specific dilution factors to estimate ingredient-level concentrations in surface water at the county level.

Results show that estimated concentrations vary considerably across the country when using more refined spatial data, and that economic information (“take off” values and population GDP) can have a significant influence on the resulting ingredient emission distributions. The method presented incorporates the inherent spatial variability of the model inputs so that patterns can be identified and used in the risk assessment. In other words, the ability to identify areas where existing combinations of model inputs may yield greater exposure estimates, i.e., the identification of realistic “worst case” scenarios. In addition, an understanding of where these “worst case” scenarios fit within the overall China-wide distribution is achieved.

Terrain elevation data play an important role in characterizing natural and anthropogenic agricultural features affecting the transport of water, sediment, nutrient, and chemicals at the field and watershed scale. LiDAR (LIght Detection And Ranging) is emerging as a technology of choice due to its high resolution and accuracy and increasing public availability. This study was conducted to examine the utility of a LiDAR-derived DEM (digital elevation model) in assessing natural and managed agricultural features potentially affecting runoff covering a 60 square mile region in Nebraska. Processing was conducted using publically available LiDAR and Geographic Information System (GIS) data within standard GIS software and functions. Quantitative metrics describing the topography, hydrology, and natural vegetation elevation were calculated from the LiDAR data and summarized for all fields in the study area. Integrating high resolution aerial imagery with the LiDAR data enabled the identification of engineered features such as terracing, overhead irrigation, and intakes for tile terracing systems. Natural processes identified in the data include concentrated and diffuse flow patterns towards the edge of a field; in-field depressions with the potential for surface water storage; and vegetation diversity of riparian areas characterized by canopy height and density. This study is intended as an initial examination into the utility of LiDAR to assess field runoff vulnerability and assist with stewardship and best management practices.

Technological advances in remote sensing and geographical information system (GIS) have increased the spatial and temporal resolution of meteorological (climate and weather) data which is an integral part of in environmental and pesticide transport modeling. However, considerable computing resources may be required as the spatial resolution of the simulation increases. Techniques have been developed to coordinate enterprise-level data collection and dissemination of advanced weather and climate information in order to prepare agriculturally relevant climate datasets for environmental modeling. This presentation will describe data collection and integration methodologies for combining these climate surfaces with soil data and other environmental information. In addition, similar methodologies to employ near real-time meteorological information for predictive agricultural and environmental applications will be discussed. This presentation will demonstrate additional approaches for preparation of daily gridded climate data estimates, applicable to water quality monitoring, environmental fate and agronomic modeling, and development of regional climate change scenarios.

An Atrazine Ecological Exposure Monitoring Program (AEMP) was initiated in 2003 and continues in 2011. Monitoring in 72 corn and sorghum agricultural watersheds in nine Midwestern states produced over 150 “site years” of land use, residue, total suspended solids, meteorological, and stream flow data. Sampling locations were identified on streams at the outlets of watersheds based on defined criteria for potential atrazine use and watershed scale. Watershed sampling locations were equipped with an integrated system of weather stations, automatic samplers, and stream stage measurement stations. Monitoring in each watershed was designed to collect four-day grab samples during the five month growing season. Runoff event based and daily automatic sample collection programs were used during the study. Land use was characterized by customized satellite image classification or USDA NASS Cropland Data Layer data. Results show that the AEMP study design adequately captured atrazine runoff events following chemical applications for each growing season.

A decline in pelagic species in the San Francisco Bay Delta region has led to speculation as to whether contaminants may be playing a role. A weight-of-evidence analysis was conducted to rank the relative risk potential for pesticides to impact threatened or endangered species in the Sacramento River, San Joaquin River, and Bay-Delta estuary watersheds. The study utilized monitoring data, simulation modeling, and GIS to address the co-occurrence of40 widely used herbicides, fungicides, and insecticides to 12 aquatic and semi-aquatic species, including Chinook salmon (Oncorhynchus tshawytscha), Central Valley steelhead (O. mykiss), southern North American green sturgeon (Acipenser medirostris), Delta smelt (Hypomesus transpacificus), striped bass (Morone saxatilis), San Francisco longfin smelt (Spirinchus thaleichthys), threadfin shad (Dorosoma petenense), California red-legged frog (Rana draytonii), and the California freshwater shrimp. Pesticide application sites represented in model simulations included fruit, vegetable, grain, nuts, rice, landscape maintenance, and structural applications. Daily pesticide concentrations were predicted at the PLSS section level from runoff, erosion, and drift sources. An index method was developed to evaluate the spatial and temporal co-occurrence of pesticides and species. Areas of highest potential concern were located along the main branch of the Sacramento River, the northern part of the Delta region, and the southern part of the Delta in San Joaquin County. A few small clusters of high co-occurrence values were predicted along the Sacramento River Deep Water Ship Channel and in southern Butte County along the Feather River. Monitoring is sparse in a number of these and lesser areas of potential concern. The results of this study are being used to support current and future monitoring programs by strategic placement of sampling locations and frequency. Ultimately, it is hoped that this project will improve decision making and optimize resource spending of groups seeking to improve the long-term sustainability of these aquatic habitats.

Determining the EDWCs (Estimated Drinking Water Concentrations) in surface water requires the use of a standard modeling process that combines information about the environmental fate, anticipated use patterns, standard cropping scenarios, and regional percent crop area assumptions. Implementing the standard modeling tools within the set of shells, while adequate for many products, may lead to assumptions that do not match the reality of actual labeled use patterns or cropping patterns. Improvement of several types can be implemented within the Tier 2 framework, resulting in a better representation of actual use patterns. The key refinements to be discussed are:

• representing combinations of foliar and soil application techniques
• representing multiple crop seasons and crop rotations
• combining loadings from major and minor crops within a geographic area
• applying Percent Crop Area values for Hawaii and Puerto Rico

The discussion will be focused on refinements that can be implemented within the EPA Tier 2 modeling framework across all drinking water risk assessments and that do not rely on proprietary data sources or complex modeling techniques.

The Pesticide Root Zone Model is used in the FOCUS Surface Water and Groundwater model software (winPRZM). Outside of these model frameworks, winPRZM may be used as a stand-alone model in predicting fate and transport of mass from the field in higher tier scenarios or in situations that are site specific in agricultural practices. This presentation will show various enhancements to the winPRZM model to address agricultural practices or regional variances that are not in the version currently used for FOCUS. These improvements involve pesticide transport in furrow irrigation, initial abstraction (seasonal and urban), manure application, feedlot management practices, vegetative filter strips, and sublateral flow/tile drainage.

Additionally, the presentation will show the improvements that have been made to the version of winPRZM for harmonization of the FOCUS groundwater models. These enhancements include irrigation file (separate from weather file), FAO 55 single/dual crop evapotranspiration, and specification of hydrodynamic dispersion length.

A flexible modeling platform has been developed to evaluate the potential impact of crop protection chemicals on the environment throughout the world. The tool currently has been configured with scenarios containing crop, soil, and weather conditions for major agricultural areas in Canada, Colombia, the European Union, Norway, the People’s Republic of China, and the United States. Scenarios are simulated using fate and transport models that have been accepted for regulatory assessment in the US and the European Union, including the Pesticide Root Zone Model (versions 3.12.2 and winPRZM 4.51); EXAMS, RICEWQ, ADAM (groundwater dilution model), and TOXWA . A key strength of the tool is that scenarios can be added for additional geographical areas with relative ease and the appropriate regulatory endpoints.

LiDAR (Light Detecting and Ranging) data, covering a 60 square mile region in Nebraska, was inspected for its utility in assessing natural and managed agricultural features at the field and watershed scales. This approach was used to qualitatively assess landscape features affecting potential runoff at a sub-field scale. Quantitative metrics describing the topography, hydrology, and natural vegetation were calculated from the LiDAR data and summarized for all fields in the study area. Integrating high resolution aerial imagery with the LiDAR data enabled the identification of engineered features such as terracing, overhead irrigation, and intakes for tile drainage systems. Natural processes identified in the data include: concentrated and diffuse flow patterns towards the edge of a field; in-field depressions with the potential for surface water storage; and vegetation diversity of riparian areas characterized by canopy height and density. Processing was conducted using publically available LiDAR and Geographic Information System (GIS) data and typical GIS software and functions. This study is intended as an initial examination into the utility of LiDAR to assess field runoff vulnerability and assist stewardship.

Increasingly, the spatial and temporal co-occurrence of pesticide mass loadings and threatened and endangered species is discussed as part of agrochemical product registration and regulation in the United States. However, methodologies have not been developed to address this issue in a transparent and standardized manner. An index method was developed to evaluate the spatial and temporal co-occurrence of pesticides and species, and then applied in a large ecosystem assessment project in California. The co-occurrence index combines monthly species abundance with statistical distributions of pesticide exposure events exceeding toxicity benchmarks for 40 widely-used pesticides. The frequency of co-occurrence was determined for 12 aquatic and semi-aquatic species. The application of the methodology is presented for three species, including the delta smelt, the Central Valley fall-run Chinook salmon, and the California red-legged frog.

US EPA drinking water risk assessments utilize a percent crop area (PCA) factor derived from major crops to adjust the index reservoir modeling result and to estimate the concentration expected in surface water. Examples of improvements to the current regional PCA values used will be presented to include the use of modern watershed and cropping information for refined regional default PCAs for minor crops developed and for crop distributions within smaller watersheds. PCA factors for Hawaii and Puerto Rico will also be presented. Regional PCAs are an interim step in moving from national PCAs to PCAs developed from actual drinking watersheds. In the time since the development of regional PCAs, updates to watersheds and land cover data have become available at much higher resolution, and it is now possible to improve drinking water estimates using cropping data distributed through PCA adjusted EECs (Estimated Environmental Concentration) in actual drinking watersheds.

Detections of pyrethroids and other insecticides in aquatic systems around urban areas in California have raised concerns about pesticide runoff from outdoor applications in urban and residential settings. Modeling pesticide transport from urban areas is an uncertain endeavor due to a lack of detailed data and established modeling procedures. A modeling system was developed to estimate the runoff of five widely used pyrethroids across a 164,000 km2 study area in California. County level use data was spatially distributed across the study area using detailed land use data. Homeowner and professional applicator surveys were used to allocate pyrethroid applications to buildings, lawns, and other impermeable and permeable surfaces. Washoff studies were used to develop model input parameter values for hard surface applications. Pyrethroid runoff corresponding to rainfall and irrigation events were simulated to provide a better understanding of the spatial and temporal distribution of pyrethroid mass loadings to aquatic systems in the study area.

The degradation of chemicals in soil resulting in complex metabolite pathways have been encountered in recent studies. Representation of extended pathways with up to five metabolites in a series and more than 10 metabolites in total present numerous challenges. These may be increased because of constraints driven by minor occurrence/transience or emergence later in studies with consequently limited decline phases. Additional challenges include cases when metabolites transform into their pre-cursors. Under such circumstances the estimation of reliable or statistically significant degradation rates and formation fractions of metabolites for such complex degradation pathways, using the EU FOCUS guidelines has proved challenging. A step-wise degradation kinetics approach using mini pathway studies for a metabolite and metabolite applied as parent studies have been used for estimation of reliable metabolite parameters as means of addressing kinetic uncertainties. Other suggested approaches that are being tested include the Markov Chain Monte Carlo (MCMC) method and the Iteratively Reweighted Least Squares (IRLS) method.

Over 150 site years of herbicide concentration data have been collected from second and third order streams draining from more than 65 small, Midwest watersheds that represent corn and sorghum agriculture with a high potential for agrochemical runoff. To complement these monitoring data, very detailed assessments of watershed soils, slopes, and cropping have been developed using GIS tools. These data have been analyzed using a combination of statistical approaches including logistic regression, discriminant analysis, and principal components analysis, among other multivariate techniques, in order to identify key drivers of runoff that differentiate between sites with varied vulnerability characteristics. Automated analyses were conducted across many variables and, after simplification, a three factor model was found to successfully distinguish between classes of vulnerable sites. One key factor is the fraction of the watershed where cropped lands have shallow depths to restrictive soil layers co-occurring with slopes; this offers potential for field-scale application.

With increasingly faster computers and more detailed data, it becomes easier to model pesticide mass loadings into aquatic systems at a high resolution for large geographical areas. A large-scale ecological risk assessment was recently completed to quantify spatial and temporal mass loadings of pesticides into tributaries to the Sacramento River, San Joaquin River, and Bay-Delta estuary for the purpose of guiding future risk assessments for sensitive and endangered species. Ten years of daily mass loadings were simulated for over nine million pesticide applications for 40 chemicals in a 164,000 km2 area of California’s Central Valley. Model input included historical pesticide use data from the California Department of Pesticide Regulation’s Pesticide Use and Registration (PUR) database, daily weather data from 19 stations in the California Irrigation Management Information System, detailed soils information (SSURGO) from the NRCS, and high resolution land use data from the Farmland Mapping & Monitoring Program (FMMP). Pesticide application sites represented in the simulations included fruit, vegetable, grain, nuts, rice, landscape maintenance, and structural applications. Environmental fate and transport models used for the analysis included the Pesticide Root Zone Model (PRZM), modified to simulate pesticide losses in irrigation tail water, and the rice water quality model (RICEWQ). This presentation details methods and results of the model simulations, including discussions on those factors contributing to the highest contributions of mass loadings into aquatic systems.

Evaluation of crop protection products for US registration requires the use of a standard modeling process that combines information about the environmental fate, anticipated use patterns, and toxicological data of a product. Implementing the modeling tools within the standard set of shells, while adequate for many products, may lead to assumptions that do not match the reality of actual use patterns for some products. Common shortcomings can be overcome using the core models in their native forms and custom tools that allow refinements within the US EPA Tier 2 modeling framework. Methods to address multiple crop seasons and mixed application methods currently not available in the standard tools will be reviewed. In drinking water assessments, methods to address mixed major and minor crop can be shown to represent the mixed application dates and methods associated with minor crops in the United States.

An ecological risk assessment was performed for California’s Central Valley to rank the relative risk potential for pesticides to impact threatened or endangered species. The study utilized monitoring data, simulation modeling, and GIS to address the temporal and spatial co-occurrence of 40 widely used herbicides, fungicides, and insecticides to 12 aquatic and semi-aquatic species, including several species of Chinook salmon (Oncorhynchus tshawytscha), Central Valley steelhead (O. mykiss) , southern North American green sturgeon (Acipenser medirostris), Delta smelt (Hypomesus transpacificus), striped bass (Morone saxatilis), San Francisco longfin smelt (Spirinchus thaleichthys),threadfin shad (Dorosoma petenense), California red-legged frog (Rana draytonii), and the California freshwater shrimp. Results are being used to improve decision making and optimize resource spending across a number of federal, state, and regional water quality programs. The results can be used to identify and rank areas of highest risk, aid in placement of BMPs, and support current and future monitoring programs' specifically strategic placement of sampling locations and frequency.

Gaining a better understanding of the presence of chemical mixtures in the aquatic environment is critical to refined risk assessment. While many methods for estimating the presence and concentration of single chemicals in surface water are available, few approaches attempt to quantify multiple chemicals and/or multiple stressors found in aquatic systems. This can be especially challenging when chemical co-occurrence in space and time is considered. However, it is possible to estimate the presence of multiple chemicals/stressors in spatially- and temporally-explicit assessments. This talk will present a number of chemical types and associated sources/routes of entry into surface water. Selected exposure models used to estimate the presence or concentration of chemicals will be introduced. Of particular interest will be approaches which incorporate multiple chemicals, sources and/or stressors as part of the aggregate exposure to aquatic environments, specifically highlighting the spatial and temporal dimensions. Case studies from the US (Ohio and California) and the UK will be presented.

The implementation of no-spray buffers on pesticide labels is a commonly used approach for mitigation. These mandatory distances between crop and nontarget areas reduce the amount of active substance deposition and are usually determined using models or regressions based on spray drift field trials. This presentation will introduce common models used in the US and Europe to assess drift deposition and present several case studies on the potential impact these buffers may have on crop production. The first example will illustrate the impact of no-spray buffers applied to endangered salmonid habitat in the Pacific Northwest, quantifying the amount of agriculture present within these buffers for several Evolutionarily Significant Units (ESUs). The second example examines the use of agricultural no-spray buffers in relation to residential nontarget areas in California, in which the area of crops potentially taken out of production was quantified along with estimates of the economic impact this might have from lost production. The final example will highlight the use of spatial technologies in a national analysis which examined four different no-spray buffers around flowing and static water bodies in 60,000 watersheds across the US. The goal of this presentation will be to introduce the audience to the use of spatial approaches in order to better understand the potential impact specific no-spray zones may have at both at a local and national scale.

The development and application of models to predict in-river concentrations of down-the-drain chemicals (i.e. those used in household and personal care products and pharmaceuticals) are important components of the environmental risk assessment process. While such models (e.g., EUSES and GREAT-ER) exist in Europe, there are currently no available models to estimate exposure of personal care products many developing countries. To this end, work was performed to create a GIS-based system that develops scenarios used to predict the fate of down-the-drain chemicals into freshwater and marine ecosystems.

Predicted Environmental Concentrations in surface water (PECsw) were generated for China at the county level (sub-province) and for India at the district level (sub-province). These geographic units were considered the smallest appropriate spatial analysis unit for each country. PECsw were based on both user supplied product information, as well as other geographically-linked socio-economic and environmental information from official census and other data sources.

Product information such as category (e.g., hair, skin, etc.), composition (e.g., ingredient fraction) and “take off” values (GDP threshold under which the product would not be purchased) were used to distribute total tons of individual ingredients used in each country. Product use information was available at a regional level (six regions in China and four regions in India). These data were combined with county-level economic information, population density (including urban and rural separation), dilution factors, and disposal mechanism (e.g., STP, septic, direct discharge to river, etc.) to estimate ingredient-level PECs in surface water.

The method presented incorporates the inherent spatial variability of the model inputs so that patterns can be identified and used in the risk assessment. Results identify combinations of model inputs resulting in PECsw distributions, with the ability to identify realistic high-exposure scenarios in the context of the overall country-wide distribution. The results from this modeling are at a more spatially detailed level than previously possible in thes countries. The approach was standardized so it can be applied with very few changes to generate refined datasets for other developing countries with similar data availability challenges.

The development and application of models to predict in-river concentrations of down-the-drain chemicals (i.e. those used in household and personal care products and pharmaceuticals) are important components of the environmental risk assessment process. While such models (e.g., EUSES and GREAT-ER) exist in Europe, there are currently no available models to estimate exposure of personal care products in China. To this end, work was performed to create a GIS-based system that develops scenarios used to predict the fate of down-the-drain chemicals into freshwater ecosystems. Currently, many of the available inputs for the generation of environmental concentrations to be used in the exposure assessment are at a very coarse spatial scale, in some cases only a single value for the entire country is available. In this study, Predicted Environmental Concentrations in surface water (PECsw) were generated at the county-level (ca. 3,000 counties in China) based on both user supplied product information, as well as other geographically-linked socio-economic and environmental information from official census and other data sources.

Product information such as category (e.g., hair, skin, etc.), composition (e.g., ingredient fraction) and “take off” values (GDP threshold under which the product would not be purchased) were used to distribute total tons of individual ingredients used in China. These data were combined with county-level economic information, population density (including urban and rural separation), dilution factors, and disposal mechanism (e.g., STP, septic, direct discharge to river, etc.) to estimate ingredient-level PECs in surface water.

Results show that local PECsw varies considerably across the country and that economic information (“take off” values and population GDP) can have a significant influence on the resulting ingredient distributions. The method presented incorporates the inherent spatial variability of the model inputs so that patterns can be identified and used in the risk assessment. In other words, the ability to identify areas where existing combinations of model inputs may yield greater exposure estimates, i.e., the identification of realistic “worst case” scenarios. In addition, an understanding of where these worst case scenarios fit within the overall country-wide distribution (i.e., 90th percentile) is achieved.