Lead Investigators:

Thomas Harmon (UC Merced), William Kaiser (UCLA), Mark Hansen (UCLA)

Overview

The main goal of this project is to develop a multiscale approach integrating ENS (including NIMS) designs in the observation of a large-scale, dynamic environmental system.  The test bed in this case is the hydrodynamic mixing zone of the San Joaquin River downstream of its confluence with the Merced River in Central California.  Specific objectives include:

• To develop an end-to-end system for characterizing coupled velocity and water quality parameter distributions in a large-scale river.
• To demonstrate the multiscale ENS approach within an impacted riparian zone over a range of conditions (flows, seasons, etc)
• Collaborate with agency partners and other researchers to create hardware and software architecture conceptual design for ENS and NIMS deployments at local, watershed, and regional/basin scales.

Approach

This project focuses on developing and demonstrating the CENS multiscale ENS approach using high resolution synoptic sampling (NIMS RD) of steady velocity and water quality distributions across the San Joaquin River, an agricultural drainage-impacted river in Central California.  The San Joaquin River basin for example, is already equipped with synoptically sparse, networked gauging stations (separated by 10s of km) which provide time series data on river stage, flow, salinity and temperature data in an existing CI framework called the California Data Exchange Center.  These data enable simple river network routing model calibration and basic flow and water quality forecasting in the confluence zone.  For example, we can predict how a sudden decrease in flow by reservoir operators can be accompanied by an increase in nitrate, salinity,  and/or temp in the river if nonpoint source pollution inputs remain reasonably steady.  However, this forecasting capability is not adequate to say whether the water quality changes across the confluence zone will impact salmon migration. But the low resolution knowledge has proven to greatly inform mobile (e.g., kayak-deployed) and higher resolution stationary (river javelin) sensor deployments enabling more detailed characterization of the confluence zone and identification of optimal locations for NIMS RD super-resolution efforts.  Optimal NIMS RD timing is then identified using the ensemble of networked sensors and simulators.

Systems/Experiments

A major field campaign took place on the San Joaquin River in August 2006 (Figure 1).  The first transect is located approximately 290 m downstream of the San Joaquin River and Merced River confluence.  The sensor node payload consists of (1) a Sontek Argonaut-ADV; used for flow monitoring, and (2) a Hach Environmental Hydrolab Minisonde 4a; sensing ammonium, total ammonium, nitrate, pH, specific conductivity (SC), and temperature.  The raster scan for the first transect was performed on 08/22/06 form 18:29 to 20:56.

The second transect is located approximately 135 m downstream of the confluence (upstream of the first transect).  The sensor node payload consists of (1) a Sontek Argonaut-ADV and (2) a Hach Environmental Hydrolab DS5; sensing luminescent dissolved oxygen (LDO), oxidation reduction potential (ORP), pH, specific conductivity (SC), temperature, turbidity, and depth.

Figure 1:  Bathymetry and NIMS-RD transect locations.

Accomplishments

A series of contour plots are presented (Figures 3-19) showing parameter distributions generated from NIMS-RD raster scans at the first and second transects.  The spatial domain for each distribution is that area bounded by the bathymetric profile within the transect and the water surface elevation.  All contour plots are presented with upstream views.  Therefore, mixing occurs with Merced River water entering from the left and San Joaquin River water entering from the right.  Flow in the San Joaquin River is dominated by agricultural drainage.  The Merced River, in contrast, is relatively cold and less saline as it enters the San Joaquin River.

First Transect.  Figure 2 shows the velocity distribution for the first transect, where magnitude velocities are based on the length of the resultant vector, the sum of velocity vector components within the principle directions (i.e. easting, northing, elevation).  NIMS-RD sampling locations are represented as black point symbols.  Sampling of velocities goes no deeper than 1 m due to operational error.  A volumetric flow rate of 23.17 m3 sec-1 is calculated for the distribution.  A USGS and DWR gauging station (Newman), located approximately halfway between the first and second transects, allows for a comparison of volumetric flow rates.  The average volumetric flow rate reported at the gauging station during the duration of the raster scan is 29.73 m3 sec-1; resulting in a percent difference of 22.1%.  The high percent difference is attributed to a lack of sampling points at lower elevations.

 

Figure 2:  Distribution of the resultant velocity magnitude (top) longitudinal and transverse velocity components (bot) for the downstream transect.

Velocity vector components within the longitudinal and transverse directions are also shown in Figure 2.  The contour plot gives the longitudinal velocity distribution with arrows symbolizing the magnitude and direction of transverse velocities within the profile.  A volumetric flow rate of 21.69 m3 sec-1 is calculated for the distribution with a 27.1% percent difference when compared to the gauging station.  The large percent difference is again attributed to the absence of sampling points below 1 m.

Figure 3: A subset of the water quality distributions acquired with the velocity distributions is shown looking upstream for ammonium, nitrate and specific conductance (EC).  These cross-sectional distributions exhibit the prevalence of dissolved salts on the San Joaquin (west) side of the confluence zone.

Second Transect.Figure 4 shows the velocity magnitudes and vector components for the second transect.  Sampling locations are represented as black point symbols.  The high flows of the Merced River are clearly seen within the distribution.  Recall, that the Merced River enters from the left and the San Joaquin River from the right.  A volumetric flow rate of 30.20 m3 sec-1 is calculated for the distribution.  The USGS and DWR gauging station (Newman) reports an average volumetric flow rate of 31.15 m3 sec-1; resulting in a percent difference of 1.56%.  Therefore, for the given spatial resolution of sampling there is a minimal difference between the volumetric flow rate determined with NIMS-RD and the value reported at the gauging station.

Figure 4:  Distribution of the resultant velocity magnitude (top) longitudinal and transverse velocity components (bot) for the second (upstream) transect.

 

Velocity vector components within the longitudinal and transverse directions of the second transect are also shown in Figure 4.  The contour plot gives the longitudinal velocity distribution with arrows symbolizing the magnitude and direction of transverse velocities within the profile.  A volumetric flow rate of 23.97 m3 sec-1 is calculated for the distribution with a 19.4% percent difference. 

Figure 5 shows an array of water quality distributions generated simultaneously with the velocity distribution in Figure 4.  The distributions show sharper gradients than the downstream transect due to the fact that the upstream transect has had less opportunity to mix.

Figure 5:  Water quality distributions for the second (upstream) transect (top to bottom):  dissolved oxygen, oxidation-reduction potential, pH, specific conductance, turbidity, and temperature.

Total mass loading estimates can be obtained by integrating the product of local concentrations and velocities over a cross-section.  For example, the specific conductivity (SC) distribution within the second transect can be used to estimate a total salt load for the river.  A conversion from SC in mS cm-1 to total dissolved solids (TDS) in g m3 is made by multiplying SC by 0.65.  A TDS flux rate of 7.5 kg sec-1 is determined by multiplying TDS by longitudinal velocity values.

Future Directions

We have succeeded in creating a robust system for characterizing mixing and mass fluxes in a real river system.  Our next steps include:

• Completing upstream and downstream NIMS transects to develop a mass and energy balance over the confluence zone, and using this mass balance to test whether or not there is a significant groundwater contribution to this reach of the river
• Executing a multiscale mixing and reaeration experiment by introducing tracer dye and deoxygenated water in a controlled manner at the confluence.

People

Faculty:

• Thomas C. Harmon (School of Engineering, UC Merced)
• William Kaiser (Electrical Engineering, UCLA)
• Mark Hansen (Statistics UCLA)

Researchers:

• Jason C. Fisher (UC Merced)
• Alexander A. Rat'ko (UC Merced)

Graduate Students:

• Victor Chen (Electrical Engineering, UCLA)
• Amarjeet Singh (Computer Science UCLA)
• Michael Stealey (Electrical Engineering UCLA)
• Sandra Villamizar (Environmental Systems, UC Merced)
• Henry Pai (Environmental Systems, UC Merced)

Undergraduates:

• Chris Butler (Environmental Engineering, UC Merced)
• Brett Jordan (NIMS, UCLA)