Geospatial Methods for Monitoring Alternative Control Technology Sites

This is not a peer-reviewed article. International Symposium on Air Quality and Waste Management for Agriculture. CD-Rom Proceedings of the 16-19 September 2007 Conference (Broomfield, Colorado) Publication date 16, September 2007 ASABE Publication Number 701P0907cd

Geospatial Methods for Monitoring Alternative Control Technology Sites R.A. Eigenberg, B.L. Woodbury, J.A. Nienaber USDA-ARS U.S. Meat Animal Research Center, Clay Center, Nebraska Abstract. Monitoring alternative feedlot runoff control technology effectiveness, especially vegetative treatment systems (VTS), is of interest to both cattlemen and regulatory agencies. Producers have constructed VTS in several mid-western states under an agreement with the Iowa Cattlemen Association and the Natural Resource Conservation Service. This paper describes a method of monitoring VTS performance as applied to a system constructed at the U.S. Meat Animal Research Center (USMARC), Clay Center, Nebraska. Soil conductivity maps were generated on two separate dates separated by approximately 14 months. A program, ESAP, developed by the Soil Salinity Lab at Riverside, CA was used to: 1) determine soil core locations, 2) generate nutrient specific predictive maps based on combined soil core data and a soil conductivity map using multiple linear regression methods. The ESAP program suite provided estimates of the primary variable distribution across the VTS based on twelve core sites and high density soil conductivity (ECa) data for both sample dates. Prediction data for the two dates were compared by differencing gridded data of each date. The difference map illustrates management changes that occurred during the interval between the two survey dates. Methodologies described here demonstrate the capabilities of this analysis approach as applied to VTS in operation for eight years. The methods are sensitive enough to measure subtle management changes but robust enough to evaluate overall performance. These same methods have been applied to multiple sites in Iowa and are reported in a companion paper (Woodbury et al., 2007). Keywords. Chloride, Electromagnetic induction, Feedlot, Multiple linear regression, Nutrients

Introduction Open-lot cattle feeding operations traditionally have collected precipitation runoff in holding ponds where the accumulated liquid would be used for irrigation. Many such systems have been constructed and are managed to minimize negative environmental impact. However, these systems have the potential to contaminate groundwater by liquid seepage from the holding pond (Parker et al., 1999 and Ham, 2002). Additionally, holding ponds can be a source of odor (Chastain and Jacobson, 2006), may be perceived as unsightly by persons who pass by and are costly to remediate (Jones, et al., 2006). Recently, the US EPA has indicated that alternative control technologies (ACT) could be used for precipitation runoff control (Koelsch et al., 2006) if the ACT demonstrated equivalent or superior environmental protection compared to the traditional holding pond. Producers have constructed vegetative treatment systems (VTS) in several mid-western states under an agreement with the Iowa Cattlemen Association and the Natural Resource Conservation Service. The specific designs of each system varies; however, each site requires performance monitoring as they approach steady-state operation. Computer models have been used to compare traditional and ACT technologies, but system performance and sustainability measures require on-site evaluations. Alternative systems often include a vegetative treatment area (VTA) for nutrient utilization. Monitoring a VTA requires spatial methods to identify nutrient distribution over the surface, as well as movement within the soil. A multiple linear regression (MLR) approach (ESAP) was developed at the USDA-ARS, GEBJ Salinity Laboratory, Riverside, CA (Lesch et al., 1995) and has demonstrated applicability for monitoring salinity buildup in Western soils subjected to high

salt content irrigation waters. The ESAP is a MLR model that offered advantages of a reduced calibration set of primary variate data to generate estimates based on the MLR model. The greatest value of this method rests in the ability to digitally difference subsequent time periods. This difference map illustrates nutrient concentration and spatial distribution changes from one season to the next and can be a powerful tool for monitoring system performance at start-up and after it reaches a semi-steady-state operation. The objective of this work is to describe the methodology of applying the ESAP program to a liquid runoff control system from a beef cattle feedlot. This preliminary work demonstrates the methodology on a relatively stable VTS (eight years of operation) in use at the USDA-ARS Meat Animal Research Center (USMARC) near Clay Center, NE as it performed over an approximate 14-month interval.

Materials and Methods A vegetative treatment area located adjacent to a feedlot at USMARC was chosen as a site for evaluation of temporal VTS changes as determined by soil conductivity, soil core analysis, and modeling techniques. The VTS has been described by Woodbury et al., 2005. The VTS receives precipitation runoff from the eight pens via a grass approach and a terrace with a settling basin (fig. 1). The settling basin and terrace provide a five to eight minute settling time before being distributed relatively evenly across the VTA as the liquid flows through 13 iso-elevation discharge tubes (fig 1). The downslope 4.5 ha of brome grass VTA provides nutrient removal through hay harvest. The USMARC system has operated continuously since the Fall of 1997. Apparent soil electrical conductivity (ECa) measurements were collected using a Dualem-1S (Dualem Inc., Milton, ON, Canada). The Dualem-1S operates in the horizontal and vertical dipole modes simultaneously, but only the horizontal mode (with measurement depth centered at about 0.75 m) is reported in this study. The Dualem-1S was mounted on a non-metallic trailer and pulled by an all-terrain vehicle at about 10 km/hr, with passes made every 6 m. Apparent soil electrical conductivity was recorded and stored every second, with corresponding GPS coordinates provided by a Trimble EZ-Guide GPS/Guidance system (Trimble Navigation Limited, Sunnyvale, CA). The surveys for this test were conducted on Aug. 30, 2005 and Nov. 6, 2006. ESAP Program The EASP was originally developed by Lesch et al., 1995 to serve as the core salinity assessment software package for the Soil Chemistry and Assessment Research Program at the USDA-ARS Salinity Laboratory at Riverside, CA. This program is a combination of programs that perform specific functions. The programs are available on the World Wide Web: http://www.ars.usda.gov/Services/docs.htm?docid=8918. The primary functions for our application include: ESAP-RSSD generates optimal soil sampling designs from bulk ECa conductivity survey information, ESAP-Calibrate estimates both, stochastic (regression model) and deterministic (soil theory based) calibration equations, i.e., equations which will be used to predict spatial values of one or more soil variables from electromagnetic survey data. Specific Procedures The ECa data from the VTA were input to a BASIC computer program written in-house, GenEM, which formatted and edited the ECa data according to ESAP specifications producing a grid file. ESAP-RSSD optimized soil sampling locations by selecting sites based on high density ECa data, which were used to develop the prediction model. The output of ESAP-RSSD is a set of optimal calibration sample sites (the number is set by the operator, 12 for this test) that are co-located with the ECa dataset. The 12 sample sites were uploaded as a text file for GPS navigation. Those sites were located and soil cores obtained using a hand probe to a depth of 30 cm, on the same day of ECa data collection. Soil cores were then analyzed for Cl- content. Chloride was chosen as the ion of interest since being closely associated with liquid runoff and can serve as an indicator for NO3--N transport in overland flow and soil profile. The ESAP-Calibrate program used a stochastic model to estimate the theoretical strength of correlations between ECa and Cl- as well as Cl--X and Cl--Y trend data. It automatically developed the regression modeling, produced an output file predicting field loading estimates, and generated a map of the primary parameter, Cl-.

Results and Discussion Maps showing the results of the soil ECa surveys of the VTAs are shown in figs. 2 and 3. The lightest areas (higher soil conductivity), in general, are located near the discharge tubes of the VTA and represent the largest salt loads. The soil conductivity surveys and the associated ESAP-RSSD soil coring locations are also shown in figs. 2 and 3. The ESAP-RSSD located 12 soil collection sites for each of the survey dates. Soil cores were collected as described previously and analyzed for Cl-. The data were formatted and

processed for input to the ESAP-Calibrate program.The The ESAP-Calibrate program evaluated multiple models for fitting the data and ranked the models for application to the field under study. The best fit models for both dates were of the form: (1) Cl- = bo + b1(z) + b2(x) + b3(y) Where: bo is the intercept, b1 is determined by ESAP based on the association between ECa and Cl- data; b2 is determined by ESAP based on Cl trends in the east/west direction; and b3 is determined by ESAP based on Cl- trends in the north/south direction. The model R-square values are shown in Table 1. The best fit models were used to estimate the Clvalues at each survey location by the ESAP-Calibrate program. That dataset can be plotted by ESAPSaltMapper or by other graphics plotting programs such as Surfer®. An output map, showing estimated Clvalues across the VTA, is shown in fig. 4 and 5 for the two sampling dates. One consequence of linear prediction models is the possibility of negative prediction values; negative values occur in the figs. 4 and 5 but represent a small part of the total area. Comparisons of fig. 2 and fig. 4 as well as fig. 3 and fig. 5 show an expected strong resemblance; regions of high conductivity in figs. 2 and 3 result in regions of high estimated Cl- loading in figs. 4 and 5. Additionally, the program provided range interval estimates and the percent area associated with these Cl- levels. Table 1 lists the estimates generated by ESAP-Calibrate for four ranges of Cl- values, the overall field mean and the R2 of the model for both the Aug. 30, 2005 and the Nov. 6, 2006 data. In general, (when comparing 2006 with 2005) the amount of Cl- was reduced in the “less than 40 ppm range”, increased in the “40 to 80 ppm range”, increased in the “80 to 160 ppm range”, remained the same in the “160 to 320 ppm range” and decreased in the “greater than 320 ppm range”. The shifts in Table 1 values can be interpreted in light of the prediction maps and management changes over the sampling period. The prediction maps from each of the surveys provide insight into the absolute values of Cl- and the distribution as the liquid runoff is discharged into the VTA. The maps show higher Cl- around the discharge tubes as shown in the figs. 4 and 5. Figure 4 displays the predominate discharge (Aug. 30, 2005) as occurring to the west end of the settling basin and minimal outputs at outlets 8,6,4,2, and 1. Woodbury et al., 2005 indicated the inlet elevations of these tubes relative to the inlet at the west end were higher thereby allowing minimal discharge. However, fig. 5 (Nov. 6, 2006) shows a more even distribution across the entire system. Changes were made to the system in the intervening period between 2005 and 2006; tube number 13 (not shown in the figures) was capped late in 2005, since it had developed preferential flow patterns that allowed more liquid to be discharged through that outlet. Additionally, perforated elbows were installed on all outlets to allow a slow, even discharge for small to moderate rainfall events. The top of the elbow remained open to provide basin over-flow protection in the event of large runoff events. The effect of these management changes can be seen more dramatically in a difference map (fig. 6) made by subtracting gridded data of the 2005 image from gridded data of 2006. The majority of the map (fig. 6) shows little change as seen in the grayscale values of the map; the management impact can be seen as light or high Cl- values near the discharge tubes on the east side of the VTS. The increased Cl- loading on the east side of the VTS resulted from the improved distribution system using elbows on the discharge tubes. The west side of the VTS shows Cl- levels that are moderating as the evenly distributed flow reduced the relative loading on this side of the VTS.

Conclusions The ESAP program suite provided a straightforward method for analyzing and estimating a primary dataset based on a secondary high-density dataset and a small number of primary sample points. The vegetative treatment system examined in this paper was analyzed on two separate dates using high density ECa data processed by ESAP-RSSD to locate optimal calibration soil sample locations. The soil conductivity data and soil sample Cl- data from the identified sites were combined by the ESAP program to generate a best-fit model to predict Cl- data across the entire VTA for each sampling date. The model for each date estimated Cl- concentrations across the VTA as well as five categories of Cl- concentrations. The Cl- concentrations indicated a trend toward more even flow occurring over the field in 2006 than in 2005. The concentration shift likely resulted from installation of a more uniform distribution system on the discharge tube’s inlets. This result was supported by prediction maps that showed flow being more evenly divided among the discharge tubes in 2006 compared to 2005. A difference map reinforced that conclusion by showing flows increasing in regions with previously low flow and decreased flows where comparatively high flows had historically occurred. Validation data have been collected to evaluate the methods used in this paper. The results of the validation study will be the focus of a subsequent paper.

References Chastain, J.P, L.D. Jacobson. 2006. Site selection for animal housing and waste storage facilities. Univ. of Minn. Extension Program AEU-6. Ham, J.M., 2002. Seepage losses from animal waste lagoons: a summary of a four-year investigation in Kansas. Trans. ASAE. 45(4): 983-992. Jones, D.D, R.K. Koelsch, S. Mukhtar, R.E. Sheffield, J.W. Worley. 2006. Closure of earthen structures (including bsins, holding ponds and lagoons). Animal Agriculture and the evnironemtn: National Center for Manure and Animal Waste Management White Papers. Pp. 263-282. Koelsch, R.K., J.C. Lorimor, and K.R. Mankin. 2006. Vegetative treatment systems for management of open lot runoff: Review of literature. Appl. Eng. in Agriculture. Vo. 22(1): 141-153. Lesch, S.M., D.J. Strauss, J.D. Rhoades. 1995. Spatial prediction of soil salinity using electromagnetic induction techniques 1. Statistical prediction models: A comparison of multiple linear regression and cokriging. Water Resources Research, Vol. 31, No. 2, Pages 373-386, Feb. 1995. Lesch, S.M., J.D. Rhoades, D.L. Corwin, D.A. Robinson, and D.L. Suarez. 2002. ESAP-SaltMapper Version 2.30R User Manual and Tutorial Guide. Research Report No. 149. USDA-ARS, GEBJ Salinity Laboratory, Riverside, CA. Parker, D.B., D.D. Schulte, and D.E. Eisenhauer. 1999. Seepage from earthen animal waste ponds and lagoons – an overview of research results and state regulations. Trans. ASAE. 42(2): 485-493, Woodbury, B.L., J.A. Nienaber, and R.A. Eigenberg. 2005. Effectiveness of a passive runoff control system using a vegetative treatment area for nitrogen control, Applied Engineering in Ag. 21(4):581588 Woodbury, B.L., R.A. Eigenberg, and J.A. Nienaber. 2007. Spatial nutrient distribution of VTA pilot sites in Nebraska and Iowa. Proc., Table 1. Output data from ESAP-Calibrate model fit, percent area contribution for each range of Clconcentration and overall predicted mean of the model. The data is shown for the two survey dates that were separated by approximately 14 months.

Survey Date

Model 2

R

320

Mean

ppm

ppm

ppm

ppm

ppm

ppm

%

%

%

%

%

8/30/05

0.81

46.7

8.2

14.6

18.8

11.7

68.6

11/6/06

0.75

38.8

13.9

22.3

18.6

6.4

88.1

Figure 1. The VTA receives precipitation runoff from the eight pens; the liquid flows through iso-elevation discharge tubes into a 4.5 ha field of brome grass where nutrients are removed during harvest.

40.5585 11.0 to 18.5 18.5 to 26.0 26.0 to 33.5 33.5 to 41.0 41.0 to 48.5 48.5 to 56.0 56.0 to 63.5 63.5 to 71.0 71.0 to 78.5 78.5 to 86.0 86.0 to 93.5 93.5 to 101.0 101.0 to 108.5

40.558

40.5575

-98.171

-98.1705

-98.17

-98.1695

-98.169

-98.1685

Figure 2. Soil conductivity (mS/m) survey results for August 30, 2005. Soil core locations (solid circles) based on ESAP-RSSD are shown overlaid on the survey map.

40.5585 0.2 to 9.0 9.0 to 17.7 17.7 to 26.4 26.4 to 35.1 35.1 to 43.8 43.8 to 52.6 52.6 to 61.3 61.3 to 70.0 70.0 to 78.7 78.7 to 87.5 87.5 to 96.2 96.2 to 104.9 104.9 to 113.7

40.558

40.5575

-98.171

-98.1705

-98.17

-98.1695

-98.169

-98.1685

Figure 3. Soil conductivity (mS/m) survey results for Nov. 6, 2006. Soil core locations (solid circles) based on ESAP-RSSD are shown overlaid on the survey map.

400 350 300 250

40.558

200 150 40.5575

40.557

12 -98.171

11

10

9

8

7

6

5

4

3

2

1

100 50 0 -50

-98.1705

-98.17

-98.1695

-98.169

-98.1685

-100 -150

Figure 4. Predictive map (ppm) for Cl- on Aug. 30, 2005. Also shown are geo-referenced points showing inlet and outlet of twelve discharge tubes.

420 400 380 360 340 320 300 280 260 240 220 200 180 160 140 120 100 80 60 40 20 0 -20 -40

40.5585

40.558

40.5575

40.557

12

11

-98.171

10

9

-98.1705

8

5

6

7

-98.17

-98.1695

1

2

3

4

-98.169

-98.1685

Figure 5. Predictive map (ppm) for Cl- on Nov. 6, 2006. Also shown are geo-referenced points showing inlet and outlet of twelve discharge tubes.

350 40.5585

300 250 200 150 100

40.558

50 0 -50 40.5575

40.557

12

11

10

9

8

7

6

5

4

3

2

1

-100 -150 -200 -250 -300 -350

-98.171

-98.1705

-98.17

-98.1695

-98.169

-98.1685

Figure 6. Difference of map (ppm) in fig. 5 subtracted from map in fig. 6. This map illustrates changes that occurred Cl- levels during the period of Aug. 30, 2005 to Nov. 6, 2006. Also shown are geo-referenced points showing inlet and outlet of twelve discharge tubes.

-400