OVERVIEW

This effort will emphasize deployment of a network of soil pylons and microclimate stations at the 30 acre experimental plot in Palmdale, CA, where reclaimed municipal wastewater is being used to water and fertilize forage crops.  Over the past year Contam efforts have focused on characterizing the geostatistics of the soil hydraulic properties at the experimental plot and gaining an understanding of the residual nitrate concentration profile in the soil using the potentiometric nitrate sensors created by the CENS Sensor Group.  These distributed soil properties are being used to parameterize 2D and 3D stochastic simulators for moisture, energy and nitrate propagation at the Palmdale site.  As we finalize the testing of the soil pylons and begin to install 12 to 15 of them in the field, the multidimensional models will be compared to the 1D simulators mentioned above in the context of sensor network calibration and adaptive management of this wastewater reuse project.

OBJECTIVES:

• Characterize the geospatial soil properties at the Palmdale test bed and develop algorithms for designing a 3D sensor network in soils
• Scale up soil pylon deployment to 3D heterogeneous array of sensors at the Palmdale test bed.
• Develop and test deterministic and nondeterministic algorithms for calibrating a 3D soil sensor network
• Demonstrate the effectiveness of project methods, technology, models and algorithms to understand nitrate propagation in the soil domain and manage a multi-objective water resources management problem.

APPROACH

Once the soil pylons are operational, we will scale-up to a sensor network comprising 12-15 soil pylons at the Palmdale test bed.  This project focuses on developing a link between the soil geospatial properties and the sensor network design.

SYSTEMS / EXPERIMENTS

Soil samples have been collected at the Palmdale site and analyzed for moisture retention characteristics and hydraulic conductivity.  The sequence of analyses and experiments in this project are as follows:

• Determining the geospatial distribution of soil characteristics (hydraulic conductivity) at the site
• Developing and testing the geospatial models to the design and calibration of the soil sensor network
• Deploying the full-scale Palmdale sensor network.

(i) Geospatial Distribution of Soil Hydraulic Conductivity. Roughly 94 soil samples have been collected and analyzed to date in an effort aimed at characterizing the geostatistics associated with the soil’s hydraulic conductivity (K_s).  An additional 30 samples were collected in March 2005, and are now being analyzed. K_s is a spatially distributed soil property. The relative error associated with these measurements was small relative to the spatial variability of this parameter at the site, and therefore the measurement errors are neglected in this analysis.

The EPS is a circular plot roughly 400 m in diameter. An aerial view of the sample collection locations is shown in Figure 1 below. 94 soil samples were collected at four or five depths from 23 borehole locations during several field trips to the Palmdale experimental pivot site between summer 2003 and winter 2004.  Sample locations were determined using a handheld GPS device, and these locations have been translated to the site-specific northing-easting coordinates as shown.  Sampling depths were generally 1, 3, 5, 7 and 9 ft below ground surface (bgs).  Occasional deviations from these depths were necessary when excessively dry soil conditions prevented sample collection at a particular depth.

Figure 1.  Mapping of the soil sampling locations at to date at the Palmdale 30 acre test bed.

Routines from the software package GSLIB were used to analyze the Palmdale K_s observations.  A histogram summarizing the measured K_s values in m/day is shown in Figure 2.

Figure 2.  Histogram for the 94 K_s value measurements made on Palmdale soils (K_s in m/day)

The measured values ranged from 0.018 to 12 m/day.  The lower end of this range is associated with a limited number of samples which exhibited a substantial fraction of fine silt or clay, which tends to lower the conductivity.  Statistical analysis of the measured values suggests that they are lognormally distributed (positive skewness = 1.74; kurtosis = 5.88), which is common in sandy soils.

(ii) Developing Geospatial Models Linked to Sensor Network Design and Calibration.  The plots in Figure 2 are kriging maps generated to estimate the conductivity distributions for 0, 1, 2, and 3 m depths.  The spatial distribution of measured values can be visualized in contour maps using ordinary kriging.

Figure 3. Kriging maps for the Palmdale soil at four sampling depths.  Maps are 2D slices generated from 3D model semivariogram (fitted to 94-point observed K_s data set).

A useful advantage of the kriging procedure compared with the other traditional linear interpolator schemes is that an error term, expressing the estimation variance or uncertainty in estimation, can be calculated for each interpolated value.  One algorithm of interest in this regard is known as indicator kriging, which enables one to map the probability of exceeding a given value. The plots in Figure 3 map the probability of K_s value exceeding 1.0 m/d, which is interpreted here as a relatively high hydraulic conductivity value.

Figure 4. Indicator kriging maps designating the probability of K_s values exceeding 1 m/d at the depths shown at Palmdale.

It is estimated that 50 to 100 more hydraulic conductivity measurements will be necessary to obtain a better spatial distribution of hydraulic conductivity at the Palmdale test bed.  This will also help resolve an apparent disagreement between the relatively higher hydraulic conductivity results found at the north-east corner of the site versus the lower conductivities shown in the soils survey.

(iii) Deployment of the Palmdale soil sensor network.  As the soil pylon testing is completed, we will begin to deploy pylons at the Palmdale test bed.  In a series of 4 or 5 site trips, we will sequentially deploy the sensor network totaling 12-15 pylons for a total of 84 to 105 sensors.

Major ACCOMPLISHMENTS and Research Partnerships

The key accomplishments pertain to site characterization completed at the Palmdale pivot site, including:

• Characterized the geostatistics of the soil hydraulic properties at the experimental plot
• Gaining an understanding of the residual nitrate concentration profile
• Our research partnership with the County Sanitation Districts of Los Angeles is continuing by allowing CENS investigators open access to the Palmdale test bed and generously allowing in-kind goods and services in support of field activities at the test bed site.

FUTURE Goals

• To develop and demonstrate the ENS design strategies supporting error resilient deployment in the soil environment.
• To use the knowledge about soil ENS systems gained from (1) create a link between these systems and (a) surface water and (b) habitat monitoring ENS and NIMS deployments.

These goals will be realized by achieving the following objectives/milestones:

• Develop  in situ and self-calibration methods and algorithms for differentiating between sensor signals and variations imposed by multidimensional environmental processes.
• Develop accurate and compact methods for representing noisy information from the sensors and sensor network, and for incorporating uncertainty about inter-nodal environmental media.
• Integrate the knowledge gained achieving the previous three objectives a fully integrated design protocol for creating error resilient ENS systems in soil systems.
• Expand the tools created in this effort, and combine them with those  to a large scale water quality and habitat monitoring network.

The general approaches and timelines associated with these objectives are as follows:

• A pilot deployment of multi-node sensor array will be tested in a long-term study at the Palmdale site. [June 2005 – January 2006] 
• Development of a stochastic data assimilation approach to network design and real-time state and parameter estimation to optimally interpolate between sensor readings and to quantitatively address sensor and model uncertainty.  [August 2005 – July 2006]
• A precision agriculture and water quality adaptive management strategy will be developed to control operations at the Palmdales site demonstrating an application of sensor networks in a real environmental systems. [July 2005 – July 2006]
• These tools will be expand to other applications through collaboration on the AMARSS and CLEANER projects. [July 2005-July 2007]

PEOPLE

FACULTY

Prof. Thomas Harmon, School of Engineering, UC Merced
Prof. Deborah Estrin, Computer Science, UCLA
Prof. Miodrag Potkonjak, Computer Science, UCLA
Prof. Jennifer Jay, Civil & Environmental Engineering, UCLA
Prof. Steve Margulis, Civil & Environmental Engineering, UCLA
Prof. Jose Saez, Civil & Environmental Engineering, LMU

STAFF

Dr. Juyoul Kim, Civil & Environmental Engineering, UCLA
Dr. J. Eric Haux, School of Engineering, UC Merced

GRADUATE STUDENTS

Ms. Yeonjeong Park, Civil & Environmental Engineering, UCLA
Mr. John Ewart, Environmental Systems, UC Merced
Ms. Nithya Ramanathan, Biology, UCLA
Mr. Tom Schoellhammer, Computer Science, UCLA
Mr. Naim Busek, Computer Science, UCLA

UNDERGRADUATE STUDENTS

Mr. Juan Soriano, Computer Science, Merced College
Ms. Mallory Davidson Chemical Engineering, U Washington (2004 REU)
Ms. Nicole Jurisch, Chemical Engineering, U Washington (2004 REU)
Mr. Kevan McLaughlan, Civil & Environmental Engineering, LMU
Mr. Mark Kang, Civil & Environmental Engineering, LMU

PARTNERS

County Sanitation Districts of Los Angeles County