Modeling pesticide diuron loading from the San Joaquin watershed into the Sacramento-San Joaquin Delta using SWAT

Authors: CHEN, HUAJIN - University Of California; LUO, YUZHOU - University Of California; POTTER, CHRISTOPHER - National Aeronautics And Space Administration (NASA); Moran, Patrick; GRIENEISEN, MICHAEL - University Of California; ZHANG, MINGHUA - University Of California

Submitted to: Water Research
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/15/2017
Publication Date: 5/16/2017
Citation: Chen, H., Luo, Y., Potter, C., Moran, P.J., Grieneisen, M.L., Zhang, M. 2017. Modeling pesticide diuron loading from the San Joaquin watershed into the Sacramento-San Joaquin Delta using SWAT. Water Research. 121:374-385. doi: 10.1016/j.watres.2017.05.032.

Interpretive Summary: Increasing pesticide contamination of surface waters has raised substantial concern, especially in the agriculturally- dominated San Joaquin River watershed in the Central Valley of California. Pesticides have the potential to enter the river and its tributaries as runoff from rainfall or agricultural irrigation. Pesticide contaminants in the river could reach the Sacramento-San Joaquin Delta ("the Delta"), the hub of California's water supply and home to numerous plant and animal species, including some rare, threatened, or endangered species. The Delta has large populations of non-native, invasive aquatic weeds, such as floating water hyacinth, that obstruct water flow and navigation. One method being implemented to reduce water hyacinth populations in the Delta is the release and evaluation of beneficial insects, a technique called biological weed control. The pesticide known as diuron, used commonly in the San Joaquin Valley for crop pest control, is an example of a potential contaminant pesticide that could negatively affect biological control of water hyacinth. A computer model known as the Soil-Water Assessment Tool, or SWAT, was used to predict water runoff, sediment movement, and the potential for diuron contamination into the Delta. The eastern part of the San Joaquin River watershed is controlled by reservoir releases from major rivers including the Stanislaus, Tuolumne, Merced, Chowchilla, and Fresno Rivers, and the main stem of the San Joaquin River which flows just north and east of Fresno, CA. These releases are in turn governed by snowpack melt from the Sierra Nevada. The inputs of water from the western side of the San Joaquin River are much smaller and flow only intermittently during winter rainstorms and as agricultural 'return' water after crops are irrigated. Thus, the San Joaquin River watershed is complex, and so the computer model had to take many weather, elevation soil type and land use factors into account, as well as predictions and observations of pesticide use and runoff. The SWAT model underestimated peak flows after wet winters and overestimated peak sediment flow during those times, but overall, performed well in estimating water flow and sediment flow from the eastern tributaries, while the western tributaries were more difficult to simulate due to their irregular flow patterns. Data from the California statewide Pesticide Use Records Database indicated that about 21,000 kilograms (47,000 lbs) of duiron are applied annually in the San Joaquin River watershed, mostly from November to February. The SWAT model predicted runoff of diuron within one month of its application to crops. Peak 'exportation' of duiron into the Delta was 239 kg (531 lbs) per month in the winter. Since the toxicity of duiron to the biological control agents of water hyacinth are not known, additional research would be needed to determine if this contamination is sufficient to negatively affect biological control of water hyacinth. However, the insect that provides biological control-a weevil species-and two others soon to be released-a closely-related weevil and a planthopper-are active and feeding in the spring, summer and fall months, when diuron contamination is low. This result suggests that duiron contamination is not likely to greatly reduce the ability of these insects to exert biological control. This paper provides valuable insight on the usefulness of computer simulation models to predict water flow in highly seasonal watersheds such as that surrounding the San Joaquin River.

Technical Abstract: Quantitative information on pesticide loading into the Sacramento-San Joaquin Delta waterways of northern California is critical for water resource management in the region, and potentially useful for biological weed control planning. The San Joaquin watershed, an agriculturally intensive area, is a major contributor to the elevated pesticide levels in the downstream Delta. In this study, the Soil and Water Assessment Tool (SWAT) was applied to model streamflow, sediment, and pesticide diuron in the San Joaquin watershed. The Sequential Uncertainty Fitting Version 2 (SUFI-2) algorithm was employed to perform multi-variable, multi-site calibration and uncertainty analysis. A combination of performance measures (PMs) and standardized performance evaluation criteria (PEC) was applied to evaluate model performance, while prediction uncertainty was quantified by 95% prediction uncertainty band (95PPU). Results showed that streamflow simulation was at least “satisfactory” at most stations, with more than 50% of the observed data bracketed by the 95PPU. Sediment simulation was rated as at least “satisfactory” based on two PMs, and diuron simulation was judged as “good” by all PMs. Slightly lower percentages of the observed data were bracketed by the 95PPU for sediment and diuron. Significant correlations were observed between the diuron loads, and precipitation, streamflow, and the current and antecedent pesticide use. According to model predictions, the majority (> 70%) of diuron was transported during winter months. This timing limits direct exposure of biocontrol agents to agricultural diuron runoff, whereas exposure in the dry season could be a concern. This study not only provides valuable information for the development of biological weed control plan in the Delta, but also serves as a foundation for the continued research on calibration, evaluation, and uncertainty analysis of spatially distributed, physically based hydrologic models.