Projects / A Wetland Mixing Sensor to Support Water Quality Management

A Wetland Mixing Sensor to Support Water Quality Management

Intelligent Infrastructure

Creating intelligent infrastructures to manage scarce resources and to help realize the social, cultural and economic potential of communities.

A Wetland Mixing Sensor to Support Water Quality Management

Synopsis
Constructed wetlands are a traditional tool of environmental engineers, used for water treatment, flood control, and habitat preservation. The two most powerful options for wetland management are depth adjustment and mixing adjustment. Depth measurements are straightforward to collect, but the tools currently available for mixing measurements are severely lacking. Point-measurement devices that are used to determine mixing in rivers and reservoirs suffer from significant limitations and data biases when deployed in wetlands. Furthermore, their spatially narrow perspective cannot easily be scaled up to inform full-wetland management decisions. Large-scale measurements that reveal the integrated mixing dynamics of a wetland can be very informative. Unfortunately, the significant manpower required by these cannot be repeated with enough frequency to guide management through the seasonal and inter-annual changed observed in wetlands. The plan enables such high-frequency measurements and thus more agile management of wetland biogeochemistry.

The year-long research project has developed a sensor that measures wetland mixing conditions at the time and spatial scales needed to inform management decisions. To ensure applicability, the plan began with management applications and worked from there towards sensor design.

Details
The first stage of this work was to demonstrate the potential impact of wetland sensors in the context of water quality management. This stage began with the construction of a numerical model that added a more accurate description of surface water dynamics to a standard wetland biogeochemical model. These models often neglect the surface water entirely, regardless of the key role that it plays in wetland processes. This simplification is acceptable for some purposes (e.g. prediction of net fluxes over multi-year timescales) but cannot inform management. Our model is the first (to our knowledge) that explicitly includes the effects of surface water depth and mixing. This is important for two reasons. First, the improved model will be a contribution to the scientific literature. Second, depth and mixing are the two features that managers can most directly control in an engineered wetland.

We used this model to explore a range of hypothetical wetland conditions, identifying in each case the impacts that surface water management can achieve. This supported the development of the wetland mixing sensor by suggesting the conditions in which measurements will be most valuable to managers. This will also support the procurement of follow-up funding by demonstrating the potential impacts of wetland sensors.

The second stage of this work was to determine the design parameters for a wetland mixing sensor focused on the highest-impact management scenarios. This was accomplished by an assessment of the physical conditions expected in the target wetlands. Many of these can be determined a priori, such as temperature range, solar exposure, and vegetation density. Many others, however, require direct field measurement, such as the scale-dependence of dispersion, distribution of velocity fluctuations, and rate and type of biofouling. We determined these through a series of field experiments in northern California wetlands.

Three field sites were studied, and these span a wide range of wetland conditions. One site is an experimental rice field, the second is a restored wetland managed for maximum vegetation growth, and the third is a restored wetland managed to provide bird habitat. Each is located in the Sacramento-San Joaquin Delta, and the PI has ongoing research access to these sites through a partnership with the California Department of Water Resources (DWR). Velocity  measurements were collected at these sites from a micro-scale Lagrangian tracker and tracer releases to determine volume-averaged transport. The scaling of both over a range of spatial scales were a major focus of these experiments. These measurements were used along with other recent field measurements by the PI (surface influences of wind shear, variance of interstem spacing in wetland vegetation) to support sensor design in the third project phase.

The third project phase was to design, build, and test wetland mixing sensor prototypes. The goal was to produce and test mixing sensor versions v0.1 through v0.9, which (a) allow invention disclosures to be filed and (b) set the stage for funded follow-up projects that produce and install sensors for specific wetland management applications.

Work began by considering several options for the sensor’s fundamental operating principle. One option is a velocity-based sensor, which records fluid displacement at a number of locations and determines a mixing rate from velocity gradients and variances. Another option is a tracer-based sensor, which determines the mixing rate from the spread of a conservative solute. The tracer can be intentionally injected, or a naturally occurring tracer may be leveraged.

The sensor that was developed in this project is, by fluid dynamic necessity, an array of single sensor elements. The array geometry was determined by using the measurements collected in the field studies. A range of candidate geometries were determined based on wetland fluid flow patterns. In doing this, a key challenge was the fact that wetland mixing is inherently a multi-scale process, depending both on individual vegetation stems and vegetation patches. This challenge was addressed by using the results of the field measurements. For each candidate geometry, we conducted a priori models of sensor accuracy, repeatability, and detection limits, using inputs from the field measurements. Other inputs came from the PI’s ongoing laboratory measurements of wetland flow patterns. Because mixing in wetlands is neither dominated by turbulence nor by molecular diffusivity, we used a Lagrangian stochastic model (and not a Fickian model) as the method by which the sensor calculates mixing rate.

The sensor enables use of the biogeochemical model developed in stage 1 to directly guide management decisions. After assessing current conditions, managers can simulate the effect of changes in depth and mixing with the biogeochemical model. Adjustments to depth and mixing can then be made by changing volume flowrate, inflow and outflow gates, and (in some cases) vegetation types..

Published Material
Tse I.C. and E.A. Variano, “Lagrangian Measurement of Fluid and Particle Motion Using a Field-Deployable Volumetric Particle Imager (VoPI),” Limnology and Oceanography: Methods, 2013, Vol. 11, pp. 225-238, doi: 10.4319/lom.2013.11.225.