ASAE Conference Proceeding
This is not a peer-reviewed article.
Development and Testing of Dynamic Fertilizer Model to Assess the Effect of CNMPs in the North Bosque Watershed
J. B. Houser, A. Saleh, and L. Hauck
Pp. 019-026 in Total Maximum Daily Load (TMDL) Environmental Regulations II, Conference Proceedings, 8-12 November 2003 (Albuquerque, New Mexico, USA), ed. Ali Saleh. ,8 November 2003 . ASAE Pub #701P1503
Manure and litter management currently uses some form of phosphorus indicator to determine the correct amount of fertilizer and/or manure to be applied to a field. Therefore, the manure application rate to a field can be variable from year to year, and the maximum allowable application rate is a function of some phosphorus indicator.
The watershed model SWAT and the field-level model APEX allow the user to specify a priori multi-year management by application field; however, interaction of manure management in a dynamic manner with phosphorus indicators is limited. Though APEX has been recently modified to adjust manure and fertilizer application based on soil test P levels, management adjustments do not include accounting for changing application area requirements with changing application rates and a fixed quantity of manure for disposal. The watershed model, SWAT, was altered using algorithms already developed for APEX to allow manure application rates to change dynamically within the simulation as a function of soil test P. In addition, ARCVIEW SWAT (AVSWAT) output and SWAT were modified to adjust the area receiving fertilizer in response to dynamically changing application rates, and the amount of manure requiring land application each year based on simple constraints (such as mandated manure haul-off).
Initial testing of the modified SWAT gave indications that the modifications were performing as expected. This modified SWAT will be used to assess the effect of comprehensive nutrient management plans (CNMPs) in a watershed.KEYWORDS. Comprehensive Nutrient Management Plans, Manure management, SWAT , Phosphorus, ARCVIEW SWAT
Because of a recent environmental focus on animal feeding operations there has been an increased effort towards enhancing the management efficiency of manures and organic by-products (USDA and EPA, 1999). Current regulations mandating comprehensive nutrient management plans (CNMPs) for animal feeding operations, focus on manure management on the field level (NRCS, 2000; EPA 2003). Manure and litter management currently uses some form of phosphorus indicator to determine the correct amount of fertilizer and/or manure to be applied to a field (McFarland et al., 2000; USDA and NRCS, 2000). Therefore, the manure application rate to a field can be variable from year to year, and the maximum allowable application rate is a function of a phosphorus (P) indicator. In order to determine the potential impact of CNMPs it is necessary to be able to model the effects of dynamic manure management based on some form of P indicator at the watershed level. Our current research utilizes soil soluble P as the P indicator.
The watershed model SWAT and the field-level model APEX allow the user to specify a priori multi-year management by application field, however interaction of manure management in a dynamic manner with P indicators is limited (Gassman et al., 2002; Saleh et al., 2000). APEX has been recently modified to adjust manure and fertilizer application based on soil test P levels as represented in the model by the soil soluble P variable (Williams, 2002). Management adjustments in APEX, however, do not include accounting for changing application area requirements with changing application rates and a fixed quantity of manure for disposal. SWAT was altered using algorithms already developed for APEX to allow manure application rates to change dynamically within the simulation as a function of soil test phosphorus. In addition, the output of the ARCVIEW SWAT interface that delineates subbasins and creates hydrologic response units (HRUs) was modified to create a number of small equally sized identical HRUs (subHRUs) to permit quasi field level management changes. The SWAT program has being modified to adjust the management of the subHRUs in response to dynamically changing application rates based on soil soluble P levels at user specified depths and subbasin manure quantities.
Our modifications of SWAT are being developed initially for dairy cows and their wastes. The modified SWAT will be tested in the 921 km 2 upper North Bosque River watershed (UNBRW), an intensive dairy producing region in central Texas. This watershed is located primarily in Erath County, the top milk producing county in the state of Texas (USDA, 1997). We are currently developing this modified SWAT to allow assessment of alternative manure management schemes that key off of soil test P, a criterion that may be found in CNMPs dependent upon individual state specifications for CNMPs. In the future we plan to expand the capability of SWAT to handle other potential components of a CNMP.
SWAT Soil P and Fertilizer Modifications
Modifications to SWAT were made primarily in the fertilization subroutine and in the subroutine that reads soil profile information. Modifications were made to the soil profile so that the original layers of the soil profile determined by the soil database were subdivided into 5-cm layers with soil characteristics of the layer from which they were derived. These new layers were then used by the normal functioning of SWAT to simulate soil chemical and hydrological processes.
A new fertilization subroutine was created for SWAT that is called whenever manure is applied as the fertilizer. In all other instances of fertilization SWAT’s original fertilization subroutine is used. For manure applications the SWAT management file was altered to permit the inclusion of both a P and an N agronomic rate based on the crop which is being grown with manure fertilization. User input specifies the manure application rates appropriate based on annual soil soluble P concentrations within a user specified soil depth. The subroutine automatically insures that these rates are applied. If the specified rate is met by manure application before sufficient N has been applied for crop requirements, than the subroutine automatically applies the sufficient quantity of elemental N.
Modification of ARCVIEW SWAT (AVSWAT) Output to Allow for Dynamic Management Simulation.
A stand-alone program that separates the normally created HRUs into a series of subHRUs of equal size based on the parameters of the original land uses and HRUs has been written and preliminarily validated. Figure 1 shows the logic of the new stand-alone program.
Hrulandusesoilrepswat.txt and landusesoilrepswat.txt are two files generated by the ARCVIEW SWAT interface. As indicated in Figure 1, landusesoilrepswat.txt provides the area of each land use in the subbasin, and hrulandusesoilrepswat.txt provides the data on the soils associated with each land use and the percentage of a particular land use associated with a given soil for each subbasin in the watershed. The program reads landusesoilrepswat.txt and finds the land uses that are either waste application fields (WAFs) or likely to be WAFs and divides those land uses into a number of equally sized units (subHRUs) that are added to the SWAT SUB file. The number of subHRUs created is based on how many times the predetermined land area for a subHRU can be divided into the given land area of the original land use.
Figure 1. Logic flow chart of stand-alone program that creates subHRUs
Since the land use, such as WAF or pasture, determines the nature of the management file associated with the subHRU, the program can determine the number of subHRUs that are to be assigned a specific management file from a list of potential management files in the Management List. The program then reads a file created by ARCVIEW SWAT ( hrulandusesoilrepswat.txt ) that gives the association of soils with land uses (HRUs) and the area of each unique association. From this, proportions of a particular land use with a particular soil are determined. Using this proportion, the proper number of subHRUs with the appropriate MGT (management) files are assigned the correct SWAT SOL (soil) file. If there is a WAF land use that does not have an HRU created for it by the ARCVIEW SWAT interface, then the proportion of soils in the pasture land use is used to assign soils to that particular WAF.
The program assures that every significant land use is represented by an HRU. Under the usual HRU delineation in SWAT, land uses smaller in area than a user specified threshold are not assigned an HRU. Since we are interested in modeling a variety of WAFs with various combinations of liquid and solid manures and crops, we wanted to insure that each land use was represented by subHRUs. The structure of the program preserves very closely the actual land use distribution and land use soil combinations of the subbasin. The program was tested on a single subbasin extracted from the UNBRW and found to function properly.
New subroutine (CHNGMGT) to make run-time modification of subfiles
A new SWAT subroutine (CHNGMGT) alters the management files in the subfile to create the proper number of new WAF at the beginning of a new year of simulation based on the amount of manure in the subbasin. CHNGMGT has been tested on a single subbasin extracted from the UNBRW. By creating many subHRUs the management of small areas can be altered during the course of the simulation by simply assigning a new management file to a particular subHRU. Based on the number of cows a given amount of manure is produced per year in the subbasin and has to be disposed on WAF subHRUs within the subbasin. The amount of manure available is checked against the amount of land available for manure application and the rate at which manure is permitted to be applied based on the soil soluble P concentrations. The management of HRUs is adjusted by either increasing or decreasing the number of subHRUs that are assigned WAF management ( MGT) files in the subfile ( SUB) file at the beginning of each new year of simulation.
Validation and Calibration of SWAT Soil P Modifications
The Texas Institute for Applied Environmental Research (TIAER) has monitored field plots that have been subjected to different manure application rates based on soil test P concentrations in the top 15 cm soil layer. Each plot was about 0.3 ha and surrounded by an earthen berm 0.15 m high (McFarland et al., 2000).
The baseline (control) treatment was manure applied at the N agronomic rate when soil test P in the 0-15 cm soil layer was below 200 ppm, and at the P agronomic rate when 200 ppm was exceeded in 0-15 cm soil layer with commercial fertilizer added to meet the N requirements of the crop (McFarland et al., 2000). The nutrient management for the best management practices (BMP) treatment plots was based on the following guidelines defined for that project: if soil test P within the first 0-15 cm of soil was between 0-62 ppm, manure was applied at twice the agronomic P rate; if soil test P within the first 0-15 cm of soil was between 63-120 ppm then manure application rates were 1.5 times the P rate; if soil soluble P within the first 0-15 cm of soil was between 121-200 ppm then manure was applied at the P rate; and if soil soluble P within the first 0-15 cm of soil was greater than 200 ppm, manure was not applied (McFarland et al., 2000)
Management practices on field plot PH002 were simulated for October 1997 through December 1999 to calibrate the APEX model (APEX version 8190) to conditions in the UNBRW. Local rainfall data from the National Weather Service in Stephenville, TX was used (McFarland et al., 2000).
The input of the modified SWAT model was designed to mimic the APEX simulations for treatment plot PH002. Monthly total runoff (m 3 /ha) and monthly soluble P runoff (g/ha) output of the modified SWAT model was compared to the measured data for field plot PH0002. Comparisons were made using the R 2 value, the P statistic and the goodness-of-prediction (E) value. An E value equal to one (1) indicates a perfect prediction, while increasingly negative values indicate predictions increasingly less reliable. The E value is similar to a correlation coefficient obtained from linear regressions; however, E compares the measured values to the 1:1 line of measured-equals-predicted (perfect fit) rather than to the best fit regression line. After calibration, long-term simulations of SWAT of the field plot were compared to long-term simulations of APEX.
Testing of Dynamic Management Modifications (Subroutine CHNGMGT)
The modified SWAT with the new subroutine CHNGMGT was run on a single subbasin extracted from the UNBRW to determine if management files were being changed correctly and if manure accounting was functioning correctly. Simulations were performed using the baseline and BMP guidelines outlined above.
Modified SWAT validation and calibration results.
The modified SWAT model did a good job of predicting the observed data from the PH002 field plot. Predictions of monthly runoff matched very closely the observed runoff with an excellent R 2 value of 0.92 and a goodness-of-fit (E) value of 0.79 (Figure 2 and Table 1). The modified SWAT’s prediction of soluble P in runoff was even better (Figure 3 and Table 1), with an E value of 0.88.
Table 1. Measured vs. predicted monthly values.
Measured vs. Predicted
Runoff (m 3 /ha)
Sol P runoff (g/ha)
Figure 2. Measured vs. predicted (SWAT) monthly runoff
Figure 3. Measured vs. predicted (SWAT) soluble P monthly runoff
Comparison to long-term APEX simulations
Based on the baseline and BMP criteria of the field plots (baseline treatment - N agronomic rate when soil soluble P in the 0-15 cm soil layer was below 200 ppm, P agronomic rate when 200 ppm was exceeded; BMP treatment - twice the agronomic P rate when soil soluble P was between 0-62 ppm, 1.5 times the P rate between 63-120 ppm, 1 times the P rate between 121-200 ppm, and no manure applied when greater than 200 ppm), the long-term simulations showed soil soluble P changes very similar to the APEX long-term simulations (Figure 4). The APEX soil soluble P increased slightly faster in the baseline treatment than did SWAT soil soluble P; however, both models arrived at the 200 ppm threshold in almost the exact same year (year 28). Both models then proceeded to simulate values that fluctuated around the 200 ppm level as would be expected. In the BMP simulation SWAT increased soil soluble P slightly faster than APEX. However, both models reached the first limit (62 ppm) at approximately the same time (around year 10). Both models then showed simulated values that fluctuated around the first limit of 62 ppm; however, shortly after year 25 the SWAT version simulated values that were beginning to increase to the second limit of 120 ppm, while the values for APEX continued to fluctuate around the 62 ppm limit. Longer simulations of the SWAT model showed BMP soil soluble P concentrations rising to the second limit of 120 ppm as would be expected. Longer simulations of the APEX model were not available.
Figure 4. APEX and modified SWAT soil soluble P in the top 15 cm of soil
Preliminary Testing of Dynamic Management Subroutine CHNGMGT
Figure 5. Baseline soil soluble P in the top 15 cm of soil for subHRUs receiving manure
Preliminary testing of the modified SWAT with the new subroutine CHNGMGT, using the same baseline and BMP criteria as above, appeared to change the subfiles in an appropriate manner and accounted for changes in manure quantities based on application rates and available land. Figure 5 shows the change in soil soluble P for individual subHRUs for the baseline simulation.
Figure 6. BMP scenario soil soluble P in the top 15 cm of soil for subHRUs receiving manure
New subHRUs were added around year 20, as initial WAF subHRUs reached the 200 ppm limit. At the 200 ppm limit those subHRUs were managed at a lower manure application rate (P agronomic rate) and therefore, new WAF subHRUs were added. The difference in the slope of the curves beneath 200 ppm reflected the difference in the manure application rate. Below 200 ppm, manure was applied at an N agronomic rate and hence the increase in soil soluble P on a particular subHRU was more rapid than after 200 ppm. With the use of the BMP, it took much longer for the first WAF subHRUs to reach 200 ppm, and there was no increase above 200 ppm because all manure application on that subHRU stopped at that point (Figure 6). New fields were added in Figure 6 when the first limit of 62 ppm was reached by a number of subHRUs at about year 15. The change in the slope of the lines at 62 ppm and 120 ppm reflects the decreasing application rates implemented at those limits (Figure 6). The initial area of WAFs for both simulations was 430 ha. The total area of WAFs after 50 years with the baseline simulation was 1350 ha, while the BMP simulation showed a need for 2510 ha after 50 years (the total area of the subbasin was 4103 ha).
The comparison of the modified SWAT 30 year simulations of soil soluble P with the APEX results, demonstrated that the modified SWAT is able to adequately model user defined divisions of the soil layer, as well as changes in soil P, and to use those changes in soil soluble P to dynamically change manure application rates during the simulation based on user defined criteria.
The program that divides HRUs into subHRUs makes a virtual grid within SWAT that allows the dynamic changing of management within parts of the HRU. Simulations in a single subbasin have demonstrated the impact that CNMPs can make in soil soluble P levels and the amount of area required for manure application. The next step in the research is to use the modified SWAT program and the subHRU creation program to simulate the entire UNBRW and assess the impact of various CNMP methodologies, especially those associated with managing manure based on soil test P concentrations.
EPA, Environmental Protection Agency. 2003. National Pollutant Discharge Elimination System Permit Regulation and Effluent Limitation Guidelines and Standards for Concentrated Animal Feeding Operations (CAFOs); Final Rule. Federal Register. Vol. 68, No. 29. pp. 7176-7274
Gassman, P. W., E. Osei, A. Saleh, and L. M. Hauck. 2002. Application of an Environmental and Economic Modeling System for Watershed Assessments. Journal of the American Water Resources Association. 38:2 (423-438)
McFarland, A., A. Saleh, and L. Hauck. 2000. Section 319: Nonpoint Source Pollution Control Program Demonstration Project: Demonstration of Phosphorus Best Management Practices in the North Bosque River Basin. Technical Report TR0002, September 2000. Texas Institute for Applied Environmental Research, Tarleton State Univ., Stephenville, Texas.
NRCS, Natural Resources Conservation Service. 2000. Conservation Practice Standard: Nutrient Management Code 590. NRCS, Field Office Technical Guide, Section 4, All Field Offices, TX, March, 2000 (590-1)
NRCS, Natural Resources Conservation Service. 2000. Comprehensive Nutrient Management Planning: Technical Guidance. December 1, 2000. pp. 66.
Saleh, A., J.G. Arnold, P.W. Gassman, L.M. Hauck, W.D. Rosenthal, J.R. Williams, and A.M.S McFarland. 2000. Application of SWAT for the Upper North Bosque River Watershed. Transactions of the ASAE. 43(5):1077-1087
USDA, United States Department of 1997. The Market Administrator’s Report: New Mexico-West Texas Marketing Area, Vol. 23, no. 5.
USDA, United States Department of Agriculture and EPA, Environmental Protection Agency. 1999. Unified National Strategy for Animal Feeding Operations. March 9. 1999.
Williams, J. R. 2002. The APEX Manure Management Component. Proceedings of the Total Maximum Daily Load (TMDL) Environmental Regulations Conference, March 11-13, 2002, Fort Worth, Texas. The American Society of Agricultural Engineers. pp. 44-51.