Developing and Applying Next Generation Tools for Farm and Watershed Nutrient. to Protect Water Quality

Transcription

1 College of Agriculture and Life Sciences at Cornell Integrated Nutrient Management Program Work Team Modeling Workshop Proceedings December 19-20, 2001 Developing and Applying Next Generation Tools for Farm and Watershed Nutrient Management to Protect Water Quality These proceedings summarize models and research presented at a workshop held December 19-20, The objective of the workshop was to communicate current program efforts related to agricultural nutrient management and to develop a cohesive program strategy for future collaborative efforts and funding opportunities. Soil Resource and Landscape Characteristics Water Quality Soil Fertility Management State and Federal Policy Integrated Nutrient Management Program Animal Nutrient Management Farm Business Records Cover Photograp h by Q.M. Ketterings Developing and Applying Next Hydrology Generation Precision Tools for and Farm Soil Erosion and Agriculture Watershed Nutrient Management GIS and Information Management Integrated Crop Management to Protect Water Quality Cornell Animal Science Department Mimeo 220 i Cornell Crop and Soil Sciences Research Series E-02-1

2 Contents Introduction to CALS Integrated Nutrient Management Program Work Team Modeling Workshop...1 Danny G. Fox and Caroline N. Rasmussen Using The Net Carbohydrate and Protein System (CNCPS) for Evaluating Herd Nutrition and Nutrient Excretion...4 Danny G. Fox and Tom Tylutki Cornell Cropware Version 1.0, a cunmps Software Program...13 Caroline Rasmussen, Quirine Ketterings and Greg Albrecht The Dairy Forage System Model (DAFOSYM)...30 C.A. Rotz The Soil Moisture Routing Model (SMR)...43 Pierre Gérard-Marchant and T.S. Steenhuis Modeling the Cannonsville Watershed for Water Quality Assessment and Management...51 Christine A. Shoemaker and Jennifer Benaman Modeling Phosphorus Movement from Agriculture to Surface Waters...61 Andrew Sharpley, Peter Kleinman, Margaret Gitau, Bill Gburek and Ray Bryant New Nutrient Management Approaches Using Precision and Information Technologies...77 Harold M. van Es ii

3 iii

4 Introduction Introduction to the CALS Integrated Nutrient Management Program Work Team Modeling Workshop Danny G. Fox 1 and Caroline N. Rasmussen 1 A workshop was organized by the Cornell College of Agriculture and Life Sciences Integrated Nutrient Management Extension Program Work Team (CALS INM PWT) as part of its goals and plan of work. A description of the vision and goals of the CALS INM PWT is summarized below. Vision: To improve profitability and competitiveness of New York farms while protecting the environment by assessing current knowledge, identifying research and educational needs, facilitating new research, technology and knowledge transfer and aiding in the on-farm implementation of strategies for managing nutrients. Justification: Agriculture is one of New York State s largest businesses, and keeping our farms sustainable is critical to the economy of the state, particularly in rural areas. Maintaining economic viability while ensuring environmental quality is key to their sustainability. Sustainability of New York State farms can be improved through more effective use of existing knowledge in creating comprehensive and integrated nutrient management plans for each farm. The ability to develop farm plans is, in part, limited by computer tools that can apply existing scientific knowledge, as well as incomplete knowledge of nutrient management. There are many individual program efforts addressing nutrient management issues in CALS departments, including Crop and Soil Sciences, Biological and Environmental Engineering, Horticulture, and Animal Science. Promoting inter-communication and co-operation through cross-departmental nutrient management activities will strengthen our research, extension, and teaching efforts, and streamline information development and delivery to stakeholders such as the state s AEM program. Goals: 1. Improve communication regarding nutrient management within the CALS community. 2. Enhance program planning and implementation of various CALS nutrient management research and educational efforts, including integration of knowledge across disciplines. 3. Identify gaps in knowledge and critical research important for improving nutrient management on farms. 4. Optimize application of research based understanding to farmers and other stakeholders. Proposed Steps: 1. Develop a nutrient management working group comprised of CALS faculty and staff that have major responsibilities for nutrient management programming (extension, research, and/or teaching). 1 CALS, Department of Animal Science, Cornell University, Ithaca, NY

5 Introduction 2. This working group will: a. Identify key investigators, collaborators within the broader AEM community (such as NYSSWCC, SWCD, CCE, NRCS, DEC, private consultants), farmers, and other stakeholders. b. Develop a forum (e.g. a website, newsletter and/or seminar/field day series) for information exchange and discussion among CALS faculty, staff and other stakeholders whose research, extension and/or teaching programs relate to nutrient management and those who are impacted by these efforts. c. Identify program (research, extension and teaching) priorities, potential collaborators and funding opportunities based on feedback from major stakeholders. d. Enhance opportunities for funding through improved coordination of research. Workshop Objectives One of the specific objectives of the CALS INM PWT is to develop forums for information exchange and discussion among CALS faculty, staff, and off campus collaborators, and other stakeholders whose research, extension and/or teaching programs relate to nutrient management, and those who are impacted by these efforts. This workshop was organized to communicate current program efforts related to agricultural nutrient management and develop a cohesive program strategy for future collaborative efforts and funding opportunities. Approximately 40 attended this workshop. Ultimately the goals of the workshop participants are to: Use models to provide information to answer the question, do we have a problem in a particular watershed and what are the sources of the problem? ; use models to develop strategies, and as tools for use in managing nutrients on the farm, and use models to monitor progress in reducing risk to water quality in a watershed. The questions that were addressed during the presentations, breakout session and final discussion were: 1. What models do we have to address these issues? 2. Are there missing links? 3. What linkages and collaborations are needed to improve accuracy and effectiveness and to provide creative solutions? 4. Can we develop a cohesive strategy to work together to improve our effectiveness? The presentations included the following; participants were provided with copies of each Powerpoint presentation, and papers summarizing each are included in this proceedings. 1. The Cornell Net Carbohydrate and Protein System: a model for developing herd nutrient management plans. Tom Tylutki, Animal Science Dept. Cornell University. 2. Cornell Cropware: a program for developing soil and manure nutrient management plans. Quirine Ketterings, Crops and Soil Sciences Dept. Cornell University. 3. DAFOSYM: a model for evaluating alternative whole farm plans. Al Rotz, Pasture Systems and Watershed Management Research Unit, USDA, ARS, Univ. Park, PA. 4. Soil Moisture Routing model (SMR). Tammo Steenhuis, Biological and Environmental Engineering Dept. Cornell University. 2

6 Introduction 5. Whole watershed modeling. Christine Shoemaker, School of Civil and Environmental Engineering, Cornell University. 6. Runoff P relationships from the field. Ray Bryant, Pasture Systems and Watershed Management Research Unit, USDA, ARS, Univ. Park, PA. 7. Precision Agriculture. Harold VanEs, Crops and Soil Sciences Dept. Cornell University. Workshop Conclusions: Critical Issues The CALS Director of Extension, Merrill Ewert, started the workshop by saying that if we don t get this (developing sustainable farming systems that protect the environment) right it won t matter what else we do for agriculture in New York State. Integrated nutrient management and water quality protection are critical issues. Communication Workshops such as this one and on-going CALS INM PWT meetings are a useful way for researchers to understand how their work fits in the complex series of processes from farm production to watershed environment. This communication needs to be broadened to include other stakeholders such as farmers, policy makers and regulating government agencies. Collaboration Nutrient loading and its effect on water quality are complex issues. A multidisciplinary team approach is necessary to create comprehensive nutrient management plans. Excellent nutrient management models and tools have been developed. The challenge we now have is to work together to integrate or facilitate data flow between models to account for nutrient flows from within farm to basin scale to protect water wality. One effective way to start this process is to track nutrients from farm to steam (or reservoir) with a multidisciplinary team on a joint case study farm(s). Funding and publication in peer review journals are the glue that will bond collaborative efforts. Costs When integrating tools, we must prioritize our efforts based on what work will provide the greatest economic and environmental benefit. The costs of nitrogen and phosphorus loss needs to be quantified to include direct costs (fertilizer, filtration), opportunity costs (cost of restrictions on business) and indirect costs (cost of damage to ecosystem). Conflict A major challenge is how to fit the components together to manage inherent conflicts between practices that control N, P and pathogen loss. Our current area of concern is water quality but how will our best management practices change when air quality is considered? Is it possible for us to deliver the unified recommendations that producers and policy makers want? Continuing Research Funding is needed to both continue refining models and to put needed tools in the field now. Implementation and research have to continue in tandem. Monitoring of BMP and models must be implemented to verify results. Long term research efforts are necessary to establish baseline analysis and evaluate models under varying spatial and temporal conditions. 3

7 CNCPS Herd Nutrition and Nutrient Excretion Using The Net Carbohydrate and Protein System (CNCPS) for Evaluating Herd Nutrition and Nutrient Excretion Danny G. Fox 1 and Tom Tylutki 1 Name of model: Model purpose: The Net Carbohydrate and Protein System (CNCPS) for evaluating herd nutrition and nutrient excretion To predict farm specific nutrient requirements, feed utilization and nutrient excretion for dairy and beef cattle to reduce excess nutrients fed and improve animal performance Model developers: D.G. Fox, T.P. Tylutki, M.E. Van Amburgh, L.E. Chase, A.N. Pell, T.R. Overton, L.O. Tedeschi, C.N. Rasmussen, and V.M. Durbal Contact person: D.G. Fox, Animal Science Dept., 130 Morrison Hall, dgf4@cornell.edu Description of model: The CNCPS is the herd nutrition component of the Cornell University Nutrient Management Planning System (cunmps) that was developed for use in designing whole farm nutrient management plans. The other major component is Cornell Cropware. In addition to evaluating rations for each group in the herd, the CNCPS is designed to predict whole herd annual feed requirements, nutrient excretion in total and from homegrown feeds. This information can be used to plan annual home grown crops and purchased feed requirements, and the impact of various combinations of home grown and purchased feeds, herd size, and milk production level on annual returns over feed costs and nutrient balances. The plan for the cunmps is for the CNCPS to be used in conjunction with Cornell Cropware to match homegrown feed available with alternative crop rotations to minimize nutrient excretion and nutrient loading/acre. The CNCPS 4.0 is a stand alone, windows based program written in visual basic, and is available from mlc44@cornell.edu. The CNCPS model integrates our knowledge of cattle requirements as influenced by breed type and body size, production level and environment with our knowledge about feed composition, digestion and metabolism in supplying nutrients to meet requirements and to predict nutrient excretion. The CNCPS model contains a biologically based structure to evaluate diets for all classes of cattle (i.e. beef, dairy, and dual purpose) based on consideration of the existing animals, feeds, management and environmental conditions. The approach taken and level of aggregation of variables are based on our experience in working with farmers and consultants to diagnose performance problems, and to develop more accurate feeding programs. Over 20 years, separate 1 CALS, Department of Animal Science, Cornell University. 4

8 CNCPS Herd Nutrition and Nutrient Excretion submodels that can be classified by physiological function have been developed and refined: (1) feed intake and composition, (2) rumen fermentation, (3) intestinal digestion, (4) metabolism, (5) maintenance, (6) growth, (7) pregnancy, (8) lactation, (9) reserves, and (10) nutrient excretion. New information has been periodically incorporated into these submodels. The user must have some nutritional knowledge to use the CNCPS because of the risks associated with not knowing how to choose inputs and interpret results. However, with training and experience, the CNCPS can be used to evaluate the interactions between animal type, production level, environment, feed composition and management. Changes in the ration needed to meet animal and rumen fermentation requirements under widely varying conditions can also be identified. The ability to predict requirements and supply the diet nutrients needed depend on the accuracy with which these components are predicted: 1. Maintenance requirement for energy and amino acids The maintenance energy requirement is determined by metabolic body size and rate with adjustments reflecting breed type, physiological state, previous nutritional treatment, activity, environment (temperature, wind velocity, and animal surface area and insulation) and heat gain or loss required to maintain normal body temperature. The proportion of the energy and protein intake needed for productive functions cannot be accurately determined until the proportion needed for maintenance is determined. The combined animal and diet heat production must thus be determined to assess energy balance in a particular environment, requiring the prediction of both Metabolizable and Net energy. The amino acid requirements for maintenance depend on the prediction of sloughed protein and net tissue turnover losses, as predicted from metabolic fecal nitrogen, urinary nitrogen loss, and scurf protein. 2. Energy and amino acid requirements for tissue deposition and milk synthesis Growth requirements are based on empty body tissue composition of the gain expected, based on expected mature size for breeding herd replacements or expected weight at a particular final composition, considering body size, effect of dietary ingredients, and anabolic implants. Prediction of amino acid requirements will not be accurate without accurate predictions of empty body gain or milk composition. Pregnancy requirements are predicted from uterine and conceptus demand with varying expected birth weights and day of gestation. These become critical in accuracy of ration formulation during the last 60 days of pregnancy. Lactation requirements are computed for varying day of lactation, levels and composition of milk. Reserves are used to meet requirements when nutrient intake is inadequate. Reserves must be taken into account when evaluating ability to meet requirements, especially under environmental stress, feed shortage or early lactation conditions. Visual appraisal is used to assign a body condition score, which in turn is used to predict body fat and energy reserves. The cycle of reserve depletion and replenishment during lactation and the dry period is reflected by predicted condition score change. 5

9 CNCPS Herd Nutrition and Nutrient Excretion 3. Prediction of intake and ruminal degradation of feed carbohydrate and protein fractions, and microbial growth The absorbed energy and amino acids available to meet requirements depend on accurate determination of dry matter intake (DMI), ingredient content of carbohydrate and protein fractions, microbial growth on the fiber and non fiber carbohydrates consumed, and the unique rates of digestion and passage of the individual feed carbohydrate and protein fractions that are being fed. First limiting in the CNCPS is accurate determination of DMI, and we typically insist on having actual DMI values to enter. Then we use predicted DMI as a benchmark for diagnostic purposes. The interactions of DMI, digestion and passage have several implications. First, the growth rate of each microbial pool that digests respective available carbohydrate fractions, and absorbable microbial amino acids produced, will depend on the special characteristics and intake of the feeds being fed, which in turn determines the demand for the nitrogen source required by each pool. Second, the percentage of cell wall that escapes digestion will change, depending on digestion and passage rates. Third, the site of digestion and, depending on the rate of whole tract passage, the extent of digestion will be altered. Variable rates of digestion and passage have similar implications for protein fractions in feeds. Those readily available will be degraded in the rumen, while those more slowly degraded will be partially degraded in the rumen and partially degraded postruminally, the proportion depending on rates of digestion and passage of the protein fractions in the feeds. 4. Prediction of intestinal digestion and excretion Coefficients empirically derived are used to predict intestinal digestibilities and fecal losses based on summaries of data in the literature. A more mechanistic approach is needed to incorporate the integration of digestion and passage to predict intestinal digestion. However, the accuracy of prediction of pool sizes digested depends on the accuracy of prediction of ruminal flows, and therefore has second priority to the prediction of ruminal fermentation, particularly since, with most feeds, over 75% of total tract digestion occurs in the rumen. Until routine predictions of feed content of carbohydrate and protein fractions are available with accurate predicted digestion and passage rates, the use of a more complex intestinal submodel could result in a multiplication of errors. Overall, mineral excretion predicions needed for manure nutrient management planning (N and P) were nearly identical to that measured in manure applied to fields in a 500 cow dairy, when volatilization losses of N and bedding contributions are considered. 5. Prediction of metabolism of absorbed energy and amino acids A metabolic submodel needs to be able to predict heat increment and efficiency of use of absorbed carbohydrate, volatile fatty acids, lipid and amino acids for various physiological functions with changes in productive states. However, we are currently limited to the use of transfer coefficients derived from equations for an application level model because of the limitations in predicting end products of ruminal fermentation, absorbed carbohydrate and amino acids, and the infinite metabolic routes connecting the numerous tissue and metabolic compartments, the multiple nutrient interactions, and the sophisticated metabolic regulations which drive the partioning of absorbed nutrients in various productive states. Pitt et al. (1996) 6

10 CNCPS Herd Nutrition and Nutrient Excretion have described the prediction of ruminal fermentation end products within the CNCPS structure as a first step. The equations used to predict Metabolizable Energy from Digestible Energy reflect the variation in methane produced across a wide range in diets. The equations used for lactating dairy cows to predict Net Energy for Lactation from Metabolizable Energy reflects the energetic efficiency associated with the typical mix of metabolites in the Metabolizable Energy, based on respiration chamber data (Moe, 1981), and validated on independent data (Roseler, 1994). The equations used for growing cattle to predict Net Energy for maintenance and Net Energy for gain reflect the wide variation in metabolites used in growing cattle and dry cows, and validated with little bias across a wide range of Metabolizable Energy contents (NRC, 1996). Model Applications: Applications of the CNCPS have been the following: as a teaching tool to improve skills in evaluating the interactions of feed composition, feeding management and animal requirements in varying farm conditions; to design and interpret experiments; to apply experimental results; to develop tables of feed net energy and metabolizable protein values and adjustment factors that can extend and refine the use of conventional diet formulation programs; as a structure to estimate feed utilization for which no values have been determined and on which to design experiments to quantify those values; to predict requirements and balances for nutrients for which more detailed systems of accounting are needed, such as peptides, total rumen nitrogen, and amino acid balances; as a tool for extending research results to varying farm conditions; as a diagnostic tool to evaluate feeding programs and to account for more of the variation in performance in a specific production setting, and to provide information that can be used to improve whole farm nutrient management planning. Included are reduction of imported nutrients and reduction of manure nutrients, and crop production that matches herd and soil productivity. Many graduate students have used the CNCPS to design experiments and to evaluate their results for their thesis research, and continue to use it in their work as nutritionists, primarily in North and Latin America. Many undergraduates at Cornell University and many feed consultants in New York have been trained to use the CNCPS, and they now use the model to improve feeding programs on NY dairy farms. In our first case study to evaluate the model on a New York dairy farm, changes made using the CNCPS were estimated to save $42,000 in feed costs the first year. Over three years the average milk production for this herd increased from 11,000 kg to over 12,000 kg per cow per year, while nitrogen excretion was reduced approximately 33% (Klausner et al., 1998). In a 500 cow dairy case study described by Tylutki and Fox (1997), this new version was evaluated for its ability to predict the amount of N, P and K by comparing predicted amounts to that accounted for by manure amount and composition applied to the fields. The predicted amount of manure excreted was 16% lower than the measured value; the difference can be accounted for by the bedding in the manure hauled to the fields. The N predicted was 10% higher 7

11 CNCPS Herd Nutrition and Nutrient Excretion than measured values; in another case study with a similar manure handling system, 10% was lost in the barn by volatilization, accounting for this difference (Hutson et al., 1998). The predicted P excretion was 2% higher than measured values, which is within measurement error for feed and manure analysis and load weights. The predicted K excreted was 16% higher; the reason for this discrepancy is not obvious unless the variation in K in the ration was greater than measured by the feed analysis used in the case study. Given this accuracy in prediction of excretion, we believe the CNCPS 4.0 can be used to compute nutrient balances on dairy farms and to evaluate alternatives that will reduce N, P and K balances. The most recent application of the CNCPS 4.0 has been in the Precision Feeding Project of Paul Cerosaletti. The objectives of the Phosphorus Reduction Through Precision Animal Feeding program underway in Delaware County are to investigate and implement strategies to improve the phosphorus (P) mass balance on dairy farms by reducing the imported and excreted (manure) P and improving P cycling within the farm. This will be accomplished by addressing purchased feed P and homegrown forage management. Four cooperating pilot dairy farms located in the Cannonsville Reservoir basin and ranging in size from 40 to 100 lactating cows per farm were enlisted in November Monthly monitoring of lactating cow diets, cow performance, and manure nutrient composition on the farms began in December Lactating cow diets were analyzed using the Cornell Net Carbohydrate and Protein System. Strategies to reduce phosphorus supplementation to requirement have been implemented on two pilot farms. On these same two farms, strategies to improve the quality and quantity of homegrown forages were implemented during the 2000 growing season. Initial evaluations indicated that in two herds (A and B), P intakes were 165% and 141% of requirement on average across all production levels. By intervening in the diets of these herds, P intakes have been lowered closer to requirement, resulting in reductions in predicted P intakes and excretions of 32 and 18 lbs per cow per year for herds A and B respectively. These reductions represent deceases in predicted P intakes of 28% and 21% and P manure excretions of 36% and 29% for herds A and B respectively. A review of forage quality analyses across two growing seasons (n=86) shows a trend towards increased P content as homegrown forage quality, and often quantity, are improved. These data suggest that substantial reductions in feed phosphorus imports and manure excretions on dairy farms typical of the Cannonsville Reservoir basin are possible, and that there exists the potential to increase forage yield and P content with improved management, thus increasing removal of soil P, reducing the need for supplemental feed P, and improving recycling of P within the farm. References for more information: Animal Science Department Mimeo 213 is a complete manual that contains sections on CNCPS 4.0 model development and evaluation with animal performance studies, a list and description of all model equations, the feed library, model tutorials, and a list of publications, including the nearly 50 publications in peer reviewed journals on the science underlying the model, evaluations, and applications. A CD containing the model, this manual, publications, and 8

12 CNCPS Herd Nutrition and Nutrient Excretion tutorials is available in 130 Morrison Hall from Michelle Cole at no charge for any academic user,.any New York resident, or Company doing business on farms in New York. A fee schedule is available for commercial users outside of New York. Model Inputs and Outputs: See Table 1.1, Cornell Net Carbohydrate and Protein System Example Input and Output Data. Cornell Net Carbohydrate and Protein System example reports are shown in Figures 1.1, 1.2 and

13 CNCPS Herd Nutrition and Nutrient Excretion Table 1.1 Cornell Net Carbohydrate and Protein System example input and output data CNCPS Inputs and outputs: Inputs for each animal group Animal Descriptions Environmental factors and descriptions Age, sex and breed Previous and current temperature Body Weight (average and mature) Previous and current rel. humidity Days Pregnant and Calving Interval Wind speed Exp. Calf Birth Weight Hair Depth and coat condition Age at first calving and lactation Number Housing/pasture descriptions Milk amount, and price, fat and protein % Body Condition Score Inputs for each feed in the diet Chemical composition Digestion rates Dry Matter (%) CHO-A (%/hr) Cost ($/100 lb.) CHO-B1 (%/hr) Amount fed/day CHO-B2 (%/hr) NDF (%DM) Protein-B1 (%/hr) endf (%NDF) Protein-B2 (%/hr) Lignin (%NDF) Protein-B3- (%/hr) Starch (%NSC) Intestinal digestibilities CP (%DM) Sol-P (%CP) NPN (%Sol-P) NDFIP (%CP) ADFIP (%CP) Fat (%DM) Ash (%DM) Amino acids (% of undegraded protein) Minerals (% of DM or mg/kg) Outputs in reports Animal performance factors Dry matter intake Milk Production Energy allowable milk Protein allowable milk Condition Score Change Rumen Nitrogen Balances Predicted Ruminal ph Nutrient balances Ration cost/day Target ADG ME Allowable Gain MP Allowable Gain Diet nutrient concentrations herd and nutrient management information Group and herd Ration Cost Income over feed costs Feed required for 365 days Annual manure output Annual fecal output urine output Total nitrogen excretion Total phosphorus excretion 10

14 CNCPS Herd Nutrition and Nutrient Excretion Figure 1.1 Whole herd group performance and nutrient balances example report 11

15 CNCPS Herd Nutrition and Nutrient Excretion Figure 1.2 Whole herd summary example report Herd Analysis Total milk production 5,597,538 kg per year Avg. milk production of lactating cattle 34.2 kg/cow/day Avg. gain of growing cattle.39 kg per day Avg. wt of all cattle 526 Kg Rations Avg. percent raised 78% Avg. percent purchased 22% Avg. ration cost per cwt milk $ 4.33 Avg. ration cost per kg. gain $ Total ration cost of herd $1, /day $653,877 /year Nutrient N P K Avg. percent purchased 33% 40% 9% Excreted kg/year 94,249 15,507 74,026 Urinary kg/year 37, ,059 Fecal kg/year 57,212 15,155 22,967 Product kg/year 32,791 6,068 11,314 Efficiency of nutrient use 35% 39% 15% Manure Kg/day Predicted Total Manure 42,999 Predicted Fecal 25,650 Predicted Urine 17,349 Figure 1.3 Whole herd feed requirement example report 12

16 Cropware 1.0 Cornell Cropware Version 1.0, a cunmps Software Program Caroline Rasmussen 1, Quirine Ketterings 2 and Greg Albrecht 2 Name of program: Cornell Cropware Version 1.0 (release August 2001) Program purpose: Cornell Cropware is a planning tool for the spatial and temporal allocation of manure and fertilizer. It is a computer program that integrates information on soils, crop nutrient requirements for each field, hydrological sensitivity, environmental risk factors including the New York P runoff and N leaching indices, crop rotations, and volume and nutrient content of manure. Cropware contains equations and coefficients needed to implement Cornell guidelines for meeting crop requirements with manure and fertilizer nutrients and was developed to assist New York nutrient management planners and livestock producers in generating nutrient management plans that meet NRCS standards. Program developers: Q.M. Ketterings, K.J. Czymmek, C.N. Rasmussen, G.L. Albrecht, V.M. Durbal, and S.D. Klausner Contact Person: Q.M. Ketterings Department of Crop and Soil Sciences Cornell University 817 Bradfield Hall Ithaca NY qmk2@cornell.edu Phone: Program Description: Cornell Cropware is the agronomic component of the Cornell University Nutrient Management Planning System (cunmps) that was developed for use in designing whole farm nutrient management plans for New York. The other major component is Cornell Net Carbohydrate and Protein System (CNCPS). Cropware is a tool that facilitates: 1) balancing farm manure nutrient supply throughout the year with crop nutrient demand; 2) allocating manure nutrients using best management practices to sites that are least hydrologically sensitive; and 3) determining the need for additional fertilizers to balance crop requirements with nutrient supply for optimum economic yields. 1 CALS, Department of Animal Science, Cornell University 2 CALS Department of Cropand Soil Sciences, Cornell University 13

17 Cropware 1.0 Cropware Planning Process The basic Cropware planning flow consists of several steps (see Figure 2.1): 1. Establish a library of farm information. 2. Determine nutrient supply from manure. 3. Determine crop nutrient requirements (N, P, K and lime). 4. Allocate manure and fertilizer to balance nutrient supply and demand. 5. Allocate annual manure and fertilizer applications over the 12 months in a year. 6. Evaluate the New York Phosphorus Run-off and Nitrogen Leaching Risk Indices and adjust manure allocations (rate, timing, and/or application method) if the risk indices are not acceptable. 7. Produce reports that will facilitate the nutrient management plan s tactical implementation. Each of the seven steps will be described in the following sections. Figure 2.1 Cornell Cropware planning process flow diagram Characterize the farm and nutrient sources Spatial allocation of manure and fertilizer in the Allocation Screen Temporal allocation of manure in the Calendar Screen If changes are necessary to lower the P index ratings or reduce the nitrate leaching risk, then reassess manure and fertilizer allocation (rate, timing, method) Update manure application timings in the Manure Use screen Assess P Index and Leaching Index ratings on Allocation Screen If ok, create reports to describe plan 1. Characterize the farm and nutrient sources Cropware produces Nutrient Management Plan (NMP) plan for the use of farm nutrients for one up to 12 plan years (Figure 2.2). Global information such as the farm and planner contact information that does not change from year to year is held at the Plan level. Plan data are organized into the following classes corresponding to different data entry screens: Contact Information, Options, Fertilizers Available, and Crop Rotations. The Plan level data provide the planner with a library of data to use when creating the plan. This information is constant across plan years. 14

18 Cropware 1.0 Figure 2.2: Plan object chart. Plan 1 st Plan Year 2 nd Plan Year 3 rd Plan Year x th Plan Year Each plan year may have unique field and manure source data sets (Figure 2.3). Information about total manure available for allocation and manure storage capacity are associated with each manure source. Multiple manure nutrient analyses can used to describe the manure composition from each source (see Figure 2.3). Figure 2.3: Plan year object chart. i th Plan Year Fields Manure Sources Field 1 Field X M. Source 1 M. Source Y M. Analysis 1 M. Analysis Z 15

19 Cropware Characterize Nutrient Sources - Determine nutrient supply from manure Essential to developing a nutrient management plan is an accurate estimate of the quantity of manure to be distributed to farm fields or exported off of the farm. The manure quantity must be entered or calculated for each manure source. In this context, a manure source is defined as a discrete manure handling system. The total annual manure quantity is the amount in the system at the start of the plan year (for storage systems) plus the amount added to the system annually less the amount exported from system annually. There may be a significant difference between the total quantity of manure produced from livestock and the manure considered in the nutrient management plan. The quantity of manure to be handled may be increased by additions of wastewater, clean water (to facilitate pumping), silage leachate, lot runoff and precipitation. A reduction can take place as a result of animals being on pasture, manure treatment such as solids separation and composting and manure or compost exports off the farm. In Cropware, the user has the option of choosing one of three ways to estimate the total quantity of manure that needs to be allocated from each source: 1. Estimate amount added using farm records. 2. Estimate using animal parameters. 3. Estimate using number and average weight of manure applications. The Estimate amount added using farm records is a single quantity value entered by the user. The Estimate using number and average weight of manure applications is a simple counting of manure loads hauled multiplied by the capacity of the manure spreader. The Estimate using animal parameters is the estimated quantity from the total manure excreted plus water, bedding and/or other wastewater added to the manure produced by the animals. The daily manure for dairy cattle is calculated based on the animal s body weight, average daily milk production and percent milk fat. The estimated daily manure excretion for all other species is based on the animal s weight as presented in table 2-1 of the second edition (1985) of the Livestock Waste Facilities Handbook (MWPS-18). Additions from precipitation include precipitation directly into uncovered storage plus run-off into storage from adjacent lots calculated as outlined in the Agricultural Waste Management Field Handbook (Part 651a, July 1996). The total amount of manure nutrients is calculated by multiplying the quantity of manure and manure nutrient content. Manure nutrient content is extremely variable. Periodic lab analysis of representative waste samples is critical to developing an accurate nutrient management plan. In Cropware, the total amount of N in the manure, ammonia N, organic N, P 2 O5 equivalent, K 2 O equivalent and total solids are entered as percents. Multiple manure analysis can be entered for each manure source. For each waste source, if there is storage, the storage capacity can be entered or calculated. This is not a required input but is necessary if the user wants to compare manure storage capacity to storage requirements and project months of storage duration. Solids accumulation and 25 year 24 hour storm precipitation and runoff are calculated following procedures outlined in the Liquid Manure Application Systems Design Manual (NRAES 89, page 46-47) and used to estimate the total waste volume required for 12 months of storage. 16

20 Cropware Determine crop nutrient requirements (N, P, K and lime) The second part of Cropware s balancing act is to determine the crop demand for nutrients. Cropware calculates N, P, K and lime requirements for each field and each crop in the rotation based on equations derived from decades of field research conducted in New York by members of the Department of Crop and Soil Sciences at Cornell University. The basis for these equations is formed by soil test results obtained using the Morgan extraction solution and method (sodium acetate buffered at ph 4.8). Compliance with Code 590 (Nutrient Management Standard) developed by US Department of Agriculture s Natural Resources Conservation Service (USDA-NRCS) requires that comprehensive nutrient management plans be based on land grant recommendations. In New York, this implies that Bray-1 (HCl and NH 4 F), Mehlich- III (an unbuffered solution of acetate, ammonium nitrate, ammonium fluoride, and ethylenediaminetetraacetic acid) and modified Morgan (ammonium acetate buffered at ph 4.8) soil test results for P and K need to be converted to Morgan equivalents prior to calculating the soil P contribution to the NY P index and P fertilizer recommendations. Conversion equations were developed for a number of commercial laboratories that serve New York producers. See Ketterings et al. (2001d) for conversion equations and a discussion on their use. A. Nitrogen There is currently no reliable soil test for N other than the Pre-Sidedress Nitrogen Test. Nitrogen requirements for specific crops are detailed in Ketterings et al. (2001a). Corn nitrogen requirements depend on the corn yield potential, nitrogen content of the soil and nitrogen content of sod crops on the field in the past three years adjusted for the soil s specific nitrogen uptake efficiency (ability of that soil to actually deliver N to the crop). Sunflowers, grain sorghum, sorghum forage, sudangrass, sorghum sudan hybrid, and millet are all calculated using a similar methodology. For corn the equation is: NetRequiredN = (YP_corngrain*1.2 SoilN - SodN)/(N_eff/100) Where: NetRequiredN is the total amount of N (lbs N/acre) from any source required for optimum crop production. The N requirement is increased by 20 lbs/acre for a no till crop production system due to slower soil warming in the spring. YP_corngr is the yield potential of corn grain in bushels (85% dry matter) per acre given field soil type and artificial drainage. SoilN is the soil s nitrogen supplying capacity. SoilN in lbs N/acre is a function of soil type and artificial drainage class. SodN is the amount of N (lbs N/acre) released from a plowed-down sod available to subsequent crop(s) through mineralization. The amount of N available from these crop residues is a function of the sod density and quality, the percent legume and time since the sod crop was plowed or killed. N_eff is the soil type and drainage dependent uptake efficiency (see Ketterings et al. 2001a). Plants are not able to take up 100% of the inorganic N supplied to the soil, although 100% efficiency for fertilizer additions and inorganic N from manure can be approached when small quantities are directly delivered to the growing crop (e.g. as 17

21 Cropware 1.0 sidedress). The percentage of applied fertilizer that does become part of the plant is called the uptake efficiency. The estimates for New York State soils range from 50 to 75 percent. In general, N uptake efficiencies (N_eff) are soil type and artificial drainage class specific. To establish a legume or legume-grass sod, no N is required. Nitrogen requirements for established legume and legume-grass stands (i.e. topdressing) depend on management intensity and on the percentage legume in the sod. Grass and pasture nitrogen requirements are constant values based on stand management classification as intensively or non-intensively managed. The nitrogen recommendations for wheat, wheat seeded with legume, barley-winter barley-winter with legume, oats, oats with legume, barley-spring, barley-spring with legume, and rye production depend on the number of years since sod was grown on the field and the soil management group. For further details on the N requirements for field crops, see Ketterings et al. (2001a). B. Phosphorus Cropware s P recommendations (expressed in lbs P 2 O 5 /acre) are based on soil P level extracted with the Morgan solution. P recommendations for grain corn and corn silage on soils with STP s <50 lbs P/acre are presented in Figure 2.4. The solid line is the average recommended fertilizer P application. The dashed lines imply that recommendations are ranges rather than absolute values. Thus, optimum economic recommendations fall with the dashed line for each soil test P level. P recommendations for other field crops are given in Ketterings et al. (2001b). Figure 2.4. Cornell recommendations for P application to silage and grain corn P recommendation (lbs P2O5/acre) Soil test P (lbs P/acre Morgan solution) 18

22 Cropware 1.0 C. Potassium Potassium requirements are expressed in lbs of K 2 O. The K recommendations for sod crops depend on yield potential, soil test K level and constants associated with the soil type. Non-sod crop K requirements depend on soil test K level and constants associated with the soil type. Potassium recommendation equations and constants are shown in Ketterings et al. (2001c). D. Lime For optimal crop production and to obtain expected yield potential, soil ph must be adjusted with lime to fit crop needs. The soil test lime requirement is determined for the crop requiring the highest ph within the rotation. Lime requirement are based on the difference between the greatest desired ph and the current soil test ph, the exchange acidity of the soil and adjustments are made depending on tillage depth and ENV (effective neutralizing value) of the liming material being used. 4. Allocate manure and fertilizer to balance nutrient supply and demand The Allocation Screen is a grid showing each farm field as a row and nutrient requirements (N, P 2 O 5, K 2 O in lbs/acre), manure source, manure and fertilizer application rates, nutrient balances (N, P 2 O 5, K 2 O in lbs/acre) and additional user selected data items in columns (Figure 2.5). A second grid, at the top of the screen, dynamically displays the manure inventory balance as manure is allocated to each field. Figure 2.5. Cropware allocation screen example. 19

23 Cropware 1.0 The basic goals in the Allocation Screen are to optimally: 1. Meet crop nutrient requirements on a field-by-field basis by allocating manure and/or fertilizer at achievable rates on the farm. 2. Allocate all of the farm s manure across the land base. 3. Minimize the risk of nutrient losses via runoff, erosion, and leaching, as indicated by the Dissolved Phosphorus Index, the Particulate Phosphorus Index, and the Nitrogen Leaching Index, respectively. The three Balance columns show the nutrients required less nutrients available to the plants in lbs per acre. Nitrogen available to the plant from manure is the sum of the inorganic fraction of the nitrogen multiplied by the ammonia N utilized by the crop (Table 2.1) and the organic fraction of the current year manure applied multiplied by the first year decay rate (Table 2.2). Table 2.1. Estimated ammonia-n losses as affected by application method. Manure Application Method Ammonia N Utilized by the Crop (%) Injected during growing season 100 Incorporated within 1 day 65 Incorporated within 2 days 53 Incorporated within 3 days 41 Incorporated within 4 days 29 Incorporated within 5 days 17 No conservation/injected in fall 0 Table 2.2. Decay series for stable organic N in manure by animal type. A last year release rate of 12% indicates that an estimated 12% of the organic N applied in the manure is expected to be utilized by the crop a year after application. Release rate for organic N in manure (%) Source Present Year Dry Matter Content Decay_current Last Year Decay_lastyr Two Years Ago Decay_2yrs Cows < Cows Poultry < Poultry Swine < Swine Horses < Horses Sheep < Sheep Temporal Nutrient Allocation 20

24 Cropware 1.0 An important consideration in the development of a nutrient management plan is determining whether the applications of manure planned on the Allocation Screen are feasible given temporal constraints. For example, the plan may call for the bulk of the manure to be spread on corn fields. But, it may not be possible to carry out the plan because there is not enough labor and machinery available to spread all the manure between corn harvest and planting. Or, the quantity of manure required by the plan may not be available when the field is accessible. To plan for these contingencies, CropWare provides a Calendar Screen with a running manure inventory to plan the timing of manure applications for each month of the year (see Figure 2.6). Figure 2.6. Cornell Cropware manure spreading calendar screen example 6. Evaluate The New York Phosphorus Run-off and Nitrogen Leaching Risk Indices After the initial manure and fertilizer allocation, the planner must consider the risk of water quality degradation from nutrient leaching and runoff. Cropware calculates three values that act as indicators of relative risk: Nitrogen Leaching Index (LI), Dissolved P Index (DP) and Particulate P Index (PP). Best management practices associated with varying index ranges are described in the Cropware Help. Additional environmental factors include inputs for Highly Erodible Land (HEL), soil erosion estimates (RUSLE), buffer widths and other hydrologic sensitivity comments. A. NY Nitrogen Leaching Index The New York Nitrate Leaching Index (LI) is an estimate of the average annual percolation expressed in inches for a particular location. The LI is based on the concept that a soil s leaching potential increases as rainfall increases. The extent of the increase depends on soil drainage characteristics. For a given annual rainfall amount, well drained and excessively well- 21

25 Cropware 1.0 drained soils have a significantly greater leaching potential than poorly drained soils. The current LI rates leaching potential based on soil hydrologic group and seasonal (October though March) and annual average county rainfall data. The Leaching Index equations used in Cropware are those derived by Williams and Kissel (1991). Leaching Index equations and county rainfall data for each county in New York are described in Czymmek et al. (2001a). The hydrologic codes for each soil type are detailed in Ketterings et al. (2001a). B. NY Phosphorus Run-off Index The NY-Phosphorus Index (NY-PI) is a rating system designed to assist producers and planners in identifying fields or portions of fields that are at highest risk of contributing phosphorus (P) to lakes and streams. The New York PI includes transport (soil drainage class, flooding frequency, distance to the stream and stream type, and concentrated flow presence) and source factors (soil test P, fertilizer and manure P application rate, timing and method). A full description of the NY-PI can be found in an article in What s Cropping Up? by Czymmek et al. (2001b). This article as well as a P index calculator (Excel file) can be downloaded from the Nutrient Management Spear Program website at index.html. The NY-PI assigns two scores to each field based upon its characteristics and the producer s intended management practices. Dissolved P Index (DP), addresses the risk of loss of water-soluble P from a field (flow across the field or through the soil profile) while Particulate P Index (PP) estimates the risk of loss of P attached to soil particles and manure. The NY-PI scores will rank a field to determine its susceptibility to P losses. Fields with high or very high site vulnerability should be managed with minimizing P losses in mind. A low or medium ranking implies management can be nitrogen based. The NY-PI score will also indicate whether other management changes such as winter spreading must be addressed. It is, however, important to note that the PI is not a measure of actual P loss, but rather an indicator of potential loss. A high or very high PI score is a warning to further examine the causes, and a low PI score means the risk of phosphorus loss is reduced, but perhaps not eliminated. 7. Produce reports that will facilitate the nutrient management plan s tactical implementation Finally, after the plan has been adjusted to satisfy feasibility constraints and environmental concerns, the user creates reports to describe and implement the plan. In Cropware, users can construct, print and save customized reports that can be exported to word processing, spreadsheet and mapping/gis software. The Work Order component of Cropware allows creation of a tactical plan, for the person(s) applying manure, of how many loads to apply per field per month per spreader. Cropware also provides a framework to collect and store manure application records. 22

26 Cropware 1.0 Program Applications: Cornell Cropware is as a decision aid for farmers and consultants to create a site-specific nutrient management plan which will promote nutrient recycling and limit environmental degradation and meets NRCS standards for nutrient management. Cornell Cropware can be used to generate supporting documentation for development of CAFO compliant Comprehensive Nutrient Management Plans in the following areas: General information o Farm and producers name and address, planner name and address, county, livestock enterprise, number of animal units and age classes. Field specific information o Field number, acreage, land use, RUSLE, and HSA. Soil Management/Erosion Control o Soil type, crop rotation, type timing, and depth of tillage. Fertility program information and environmental risk o N leaching index, P runoff index, soil tests results including the soil lab and extraction method, soil ph maintenance recommendations, fertilizer recommendations considering manure applications, and nutrients in sod. Manure/Waste Utilization o Bedding material and quantity, estimate of annual waste production, waste spreading schedule based on the priority nutrient, template available to record manure analysis and applications. Manure Transfer and Storage Existing Facilities o Capacity calculated and reported in terms of volume and time. The Cornell Cropware is currently being used to train extension agents, NRCS and SWCD employees, consultants and other CNMP planners. Cropware will continue to be used in a crosslisted Animal Science and Crop and Soil Science course on whole farm nutrient managment planning at Cornell University. In this course, senior dairy fellows and graduate students learn the principles of improving dairy farm sustainability and apply the knowledge using Cornell Cropware to develop a nutrient management plan for farm case studies. References For More Information: A CD with the software, tutorials, extensive help section and documentation can be obtained from Michelle Cole ( mlc44@cornell.edu; phone: (607) ; mailing address: 130 Morrison Hall, Cornell University, Ithaca NY 14853) or downloaded it directly from The software is available at no cost to New York users. Registered users are automatically subscribed to the Cropware listserve and will be kept updated on new releases through this listserve. Program Inputs and Outputs: Figure 2.7 shows all program inputs and outputs. Figures 2.8, 2.9 and 2.10 are examples of Cropware output reports. 23

27 Cropware 1.0 Figure 2.7 List of inputs required, and outputs obtained Input Plan (Global) Data: Producer Name Farm Name Farm Address Farm CityStateZip Farm Phone FarmFAX Farm_ Planner Name Planner Company Planner Address Planner CityStateZip PlannerPhone Planner FAX FarmWatershed Farm County First_NMP_year Annual Precipitation Winter precipitation 25 yr rainfull RunOffValuePaved RunOffValueUnpaved Input Data associated with Fields Field Name Acres Date Sampled SampleLab RotationName Crop array[1..20] of string StandingYear SoilName SoilTest_pH SoilTest_P SoilTest_K SoilTest_Zn SoilTest_Mg SoilTest_B SoilTest_Fe SoilTest_Mn TillageDepth Exchange_acidity ArtificialDrainage PSNT PercentLegumeInSod FieldCounty FieldAccess ManureRate ManureApplicationSource ManureSource ManureTest ManureTiming HydroSenComment User_ypc FertilizerName(x4) FertilizerRate(x4) FertilizerIncorp(x4) FertilizerTiming(x4) NutrientPriority AmmoniaConservation ManureTiming Flooding Frequency WaterbodyType Distance to Waterbody RUSLE Comments HEL Input Data Associated with Manure Analysis ManureAnalysis_N ManureAnalysis_NH 4 N ManureAnalysis_OrganicN ManureAnalysis_P 2 O 5 ManureAnalysis_K 2 O ManureAnalysis_DM ManureAnalysis_Date Input Data Associated with Manure Source ManureSystemIDName AnimalSpecies AnnualProduction Units Density AnimalUnits ManureSourceCapacity StorageDimensions Freeboard SolidsAccumulation 25yrStorm ManureProductionPlus ManureExportedOffFarm MilkHouse Waste SilageLeachate BedAnnual PercentManureToStore SourceArea SourceDrainage PavedLot Constant Data associated with Soils SoilName Soil_Group Lime_Index N_EFF_UD N_EFF_DR N_SUP_UD N_SUP_DR CORN_UD CORN_DR Flooding Frequency ALF_UD DrainageClass ALF_DR HydrologicGroup Constant Data associated with Crop Crop code Seeding_year Crop_Description Sodcrops A_P B_P C_P MIN_P MAX_P A_K B_K(1..5) C_K(1..5) MIN_K(1..5) MAX_K(1..5) 24

28 Cropware 1.0 Figure 2.7 List of inputs required, and outputs obtained (continued) Input Data associated with Fertilizers FertilizerName FertilizerDry or Liquid FertilizerDensity FertilizerCost FertilizerUnits FertilizerN FertilizerP 2 O 5 FertilizerK 2 O FertilizerB FertilizerFe FertilizerMg FertilizerMn FertilizerZn FertilizerS Output Data associated with each plan year: Total_Annual_Manure ManureCollected_TotalN ManureCollected_NH4N ManureCollected_OrganicN ManureCollected_P2O5 ManureCollected_K2O NH4N_supply OrganicN_supply CommercialFertCost FertilizerN PI_DP_Farm PI_PP_Farm FertilizerP2O5 FertilizerK2O ManureN ManureP2O5 ManureK2O NutrientBalanceN NutrientBalanceP NutrientBalanceK LI_Farm FarmCropAcres CornPercent HayPercent PasturePercent OtherPercent IdlePercent CornSilageAcres CornGrainAcres CornAcres Hay50+LegumeAcres Hay25-50LegumeAcres Hay1-25LegumeAcres Hay0LegumeAcres PastureAcres OtherCropAcres IdleCropAcres FarmAnimalUnits FarmAU_dairy FarmAU_beef FarmAU_poultry FarmAU_swine FarmAU_sheep FarmAU_horses Output Data Associated with Field LoadsPerField Nreq Preq Kreq ResidNlastyr ResidN2yrsago legumen Man_Recommend TotalLime yp_alfalfa yp_corn Soil_n N_eff ysp% SodN ResidualN_manure SodN NetRequired_N FieldNBal FieldPBal FieldKBal PI_DP_Field PI_PP_Field LI_Field 25

29 Cropware 1.0 Figure 2.8 Crop, livestock and nutrient index summary example report Crop, Livestock, and Nutrient Index Summary Crop Plan Acres Percent Acres Percent CORN 1st Year Corn % % 2nd Year Corn % 0.0-3rd+ Year Corn % % Total Corn Silage % % Total Corn Grain HAY 1st Year Hay % % 2nd Year Hay % % 3rd Year Hay % 4th+ Year Hay % % Total Hay % % PASTURE BEANS SMALL GRAINS SOYBEANS SORGHUM Idle & Other Livestock Animal Type Animal Units Cattle Poultry 0 0 Swine 0 0 Sheep 0 0 Horses 0 0 Total Animal Units Animal Units/Crop Acre Nutrient Index Summary Farm Weighted Phosphorus Index (DP/PP): / Farm Weighted Leaching Index:

30 Cropware 1.0 Figure 2.9 Field detail example report Field Detail Report (5), 17.9 acres Crop Rotation Plan Year Crop/Standing Year AGT3 AGT4 COS1 COS2 COS3 AGE1 Soil Soil Name: BATH Artificial Drainage: Adequate Soil Group: 3 Tillage Depth: 7-9 Inches Percent Sod: 1-25% Legume Risk Factor Highly Erodable: True Hydrologic Group: C Hydrologic Sensitivity: KEEP MANURE 100 FT FROM STREAM EDGE Value V. Low Low Medium High V. High PI-DP 66.9 PI-PP 66.9 LI 7.4 Soil Test Results (Lab: CNAL - Extraction Method: Morgan - Sample Date: 4/11/01) Value V. Low Low Medium High V. High ph 7.0 Phosphorus (lbs/acre) 11.0 Potassium (lbs/acre) Magnesium (lbs/acre) 0.0 Calcium (lbs/acre) 0.0 Ex. Acidity (ME/100g) 0.0 Nutrient and Lime Requirements For 2002 Plan Year Lime: 0.0 (tons 100% ENV Lime/acre) Nitrogen: 62 (lbs N/acre) Phosphate: 20 (lbs P2O5/acre) Potash: 0 (lbs K2O/acre) Nutrient Management Plan (2002) Manure Source Test Rate Application Method Timing Heifer Barn Heifer tons/acre Surface App on Frozen or Saturated Ground Feb-Apr Fertilizer Name Rate gal/acre 27

31 Figure 2.10 Nutrient management plan example report Cropware Nutrient Management Plan Manure Available For Application: 1,218,267 gal & tons Calculated Manure Application: 1,716,248 gal & tons Manure Allocated: 1,130,500 gal & tons Field ID Total Nutrients Required Nutrients From Applied Nutrients From Nutrient Balance (lb/a) Field 2002 Residual Gross Residual (lb/a) Manure (lb/a) Fertilizer (lb/a) PI Acres Name Crop Sod N N Req. Manure N P2O K2 (DP/PP) N P2O5 K2O N P2O5 K2O N N P2O5 K2O 5 O LI COS / ALE / ALT / ALT / COS / AGE / COS / GIT / GIT / COS /

32 Cropware 1.0 References: Agricultural Waste Management Handbook. Part 651(a). July Cornell Guide for Integrated Field Crop Management Cornell Cooperative Extension Publication. Cornell University, Ithaca N.Y. Czymmek, K.J., Q.M. Ketterings, H.van Es (2001a). The New York Nitrate Leaching Index. What s Cropping Up? Volume 11(5):1-3. Czymmek, K.J., Q.M. Ketterings, and L. Geohring. (2001b). Phosphorus and Agriculture VIII: The New Phosphorus Index for New York State. What's Cropping Up? 11(4): 1-3. Ketterings, Q.M., S.D. Klausner and K.J. Czymmek (2001a). Nitrogen recommendations for field crops in New York. Department of Crop and Soil Sciences Extension Series E Cornell University, Ithaca, NY. 45 pages. Ketterings, Q.M., S.D. Klausner and K.J. Czymmek (2001b). Phosphorus recommendations for field crops in New York. Department of Crop and Soil Sciences Extension Series E Cornell University, Ithaca, NY. 32 pages. Ketterings, Q.M., S.D. Klausner and K.J. Czymmek (2001c). Potassium recommendations for field crops in New York. Department of Crop and Soil Sciences Extension Series E01-6. Cornell University, Ithaca, NY. 39 pages. Ketterings, Q.M., B. Bellows, K.J. Czymmek, W.S. Reid, and R.F. Wildman (2001d). Do modified Morgan and Mehlich-III P have a Morgan equivalent? What's Cropping Up? 11(3): 2-3. Livestock Waste Facilities Handbook. MWPS-18. Third Edition, Midwest Plan Service, Iowa State University, Ames Iowa. NRAES 89. Liquid Manure Applications Systems Design Manual. Williams, J.R., and D.E. Kissel Water percolation: an indicator of nitrogen-leaching potential. In: R.F. Follet, D.R. Keeney, and R.M. Cruse (Eds.). Managing nitrogen for groundwater quality and farm profitability. Soil Science Society of America, Inc. Madison, Wisconsin. pp

33 DAFOSYM The Dairy Forage System Model (DAFOSYM) C.A. Rotz 1 Name of model: Model Purpose: The Dairy Forage System Model (DAFOSYM) To evaluate the whole farm impacts of alternative technology and management strategies on dairy farms Model developers: C.A. Rotz, U.S. Gupta, D.R. Buckmaster, T.M. Harrigan, L.R. Borton, R.E. Muck, D.R. Mertens, L.D. Parsch, P.H. Savoie and others Contact Person: C.A. Rotz, USDA/ARS, Building 3702, Curtin Road, University Park, PA 16801; alrotz@psu.edu Description of model: The DAFOSYM model is a whole farm simulation model of dairy production. Farm systems are simulated over many years of weather to determine long-term performance, environmental impact, and economics of the farm. As such, the model is a long-term or strategic planning tool. All of the major processes of crop production, harvest, storage, feeding, milk production, manure handling, and crop establishment are simulated, as well as the return of manure nutrients back to the land (Figure 3.1). By simulating various alternative technologies and/or management strategies on the same representative farms, the user can determine those alternatives that provide the desired level of farm production or profit. DAFOSYM is a structured program that uses numerous objects or subroutines to represent various processes on the farm. There are nine major submodels that represent these major component processes: 1) crop and soil, 2) grazing, 3) machinery, 4) tillage and planting, 5) crop harvest, 6) feed storage, 7) feed allocation and animal performance, 8) manure handling, and 9) economic analysis. 1. Crop and Soil Crops include alfalfa and grass forage crops and corn, small grain, and soybean grain crops. Alfalfa growth is simulated using ALSIM1 level 2 (Fick, 1977). Daily accumulation of both leaf and stem dry matter is predicted based upon soil-water availability and growing degree-days above 5 C. Crude protein (CP) and neutral detergent fiber (NDF) contents are modeled with empirical functions of the accumulated growing degree-days during growth (Fick and Onstad, 1988). Grass growth and development are predicted on a daily basis with a version of the GRASIM model (Mohtar et al., 1997). Growth is a function of photosynthesis and temperature as influenced by soil water and nitrogen availability. Crude protein concentration is related to nitrogen uptake and the total accumulation of dry matter. NDF is predicted from the developmental stage of the crop and the partitioning of carbohydrates among leaf and stem components. Corn and small grain growth and development are predicted using the CERES-maize (Jones and Kiniry, 1986), CERES-wheat (Ritchie and Otter, 1985), and CERES-barley models essentially as 1 USDA/ARS Pasture Systems and Watershed Management Research, University Park, PA. 30

34 DAFOSYM implemented in the DSSAT version 3 model (Tsuji et al., 1994). Developmental staging is primarily a function of the accumulation of thermal time. Grain and silage yields are predicted each day as functions of the available solar radiation, temperature, day length, available soil moisture, and available soil nitrogen. Grain and high-moisture grain are assigned values for CP and NDF. Silage quality is a function of the total nitrogen uptake and the partitioning of carbohydrates among plant components. Figure 3.1. DAFOSYM simulates material and nutrient flows for various dairy farm systems over many years of weather to determine the performance, nutrient losses, and economics of the farm. Fixed nitrogen Crop Harvest Storage Feed sold Volatile loss Establish Volatile loss Purchased fertilizer Soil Grazing Manure Animal Purchased feed, bedding, etc. Runoff & Leaching loss Volatile loss Milk sold Animals sold Soybean development is predicted using functions from SOYGRO as implemented in the DSSAT version 3 model. Stage of development is predicted from the accumulation of thermal time, photothermal time, and day length (Jones et al., 1991). Vegetative growth is predicted from photosynthetic fixation (Sinclair, 1986). The accumulation of grain dry matter is a function of the length of the stage for grain development, vegetative production, temperature and day length as influenced by moisture stress. Grain CP and NDF are assigned typical values. 2. Grazing Either the grass or alfalfa crop models can be used to predict pasture growth and yield. The model used is dependent upon the predominant pasture crop selected. With either model, pasture growth is simulated on a daily time step beginning about the end of March (Soder and Rotz, 2001). The quantity of forage available each month from April through October is that grown during the month. Sixty percent of the available forage is removed at the end of each month 31

35 DAFOSYM throughout the grazing season. This quantity of forage removed is the amount made available to the grazing animals. The remaining 40% provides the initial conditions for regrowth. Predicting the nutrient content of grazed forage is difficult since animals are selective in what they consume. Grazing animals tend to eat the plants and the parts of given plants that are highest in nutritive value. Therefore, the nutritive contents of pasture are assigned different values during monthly periods of the grazing season. 3. Machinery The machinery component is used to determine the performance and resource use rates for all machinery operations on the farm (Savoie et al., 1985). These rates include field capacity, throughput capacity, engine load, fuel consumption, electrical use, and labor requirement. Relationships are used to predict the performance and power requirements of each operation based upon the type and size of equipment used and the machinery parameters specified to describe each machine. With this information, engine load and the rates for the use of fuel, electricity, and labor are determined. Both parallel and sequential operations are modeled. Parallel operations are those in which two or more machinery components are performing their distinct functions simultaneously and interdependently. As an example, many harvest operations are parallel with harvest, transport and unloading occurring simultaneously. A delay in one component can affect the other two. Sequential operations are continuous and independent from other operations. This category includes most tillage, planting, and feeding operations where one machine is used to complete each operation. 4. Tillage and Planting Crops are established through a sequence of tillage and planting operations (Harrigan et al., 1996). These operations can only occur on days suitable for fieldwork. Moisture in the upper 15 cm (6 in) of the soil is tracked through time to predict suitable working days. Field operations are allowed on days when the soil moisture is below a critical level (a little less then field capacity). Soil moisture level is predicted with the same soil model used to predict crop growth. Soil moisture is increased by rainfall and decreased through evaporation and moisture flow to lower soil layers. Tillage and planting operations primarily occur in the spring and/or fall. Spring operations are simulated prior to crop growth based upon the fallow soil conditions of the spring. Fall operations are simulated following crop growth using the soil moisture conditions following crop production. Simulation of spring operations begins on the specified earliest starting date or when the soil thaws, whichever is later. Fall operations cannot begin until after a portion of the crops is harvested. For example, fall operations can be performed on land where corn silage has been harvested, but the fall operations cannot be completed until after the harvest of corn grain is completed. The sequence of operations is simulated through each day suitable for fieldwork until all are complete or all available time is used. A delay in planting due to untimely operations results in a decrease in yield for grain crops. Operations begin with manure application and proceed through the designated sequence of tillage operations ending with planting. More than one operation can occur simultaneously if enough machinery and labor is available. For any given block of land though, an operation must be completed before the next operation can occur. Machine hours, fuel and labor use are totaled over all tillage and planting operations to determine the resources required. 32

36 DAFOSYM 5. Crop Harvest Different models are used to harvest forage and grain crops. Forage harvest operations are simulated on a three-hour time step and grain crop harvest is simulated on a daily time step (Rotz et al., 1989). Harvest operations occur when weather and crop conditions are suitable. Crop yield is influenced by the timeliness of harvest to account for preharvest losses and the growth and quality changes occurring during the harvest period. Timeliness is a function of the suitable days available for fieldwork and the size of the machines used for these operations (Harrigan et al., 1996). Machine hours, fuel use, and labor requirements are totaled as each operation is completed. Field drying and rewetting processes that occur following mowing influence alfalfa and grass harvest rates. Drying rate is a function of the daily weather conditions, swath density, and the type of conditioning (Rotz and Chen, 1985). Rewetting from dew is a function of the crop moisture content before nightfall and the humidity and wind conditions over the night period. Rain induced rewetting is a function of the crop moisture content and the amount of rainfall. Forage dry matter and nutrient losses during field curing include respiration, rain and machine induced losses (Rotz, 1995). Dry matter loss from plant respiration is a function of crop moisture content, ambient air temperature, and curing time. Dry matter lost through respiration is assumed to be totally digestible, non-protein and non-ndf material (available carbohydrate). Dry matter and nutrient losses caused by rain consist of leaf shatter and leaching losses. Shatter losses due to machine operations are set according to the type of machine used with leaf and stem losses determined separately. Crop quality during harvest changes according to the change in leaf to stem ratio and the relative change of other plant constituents. 6. Feed Storage Harvested feeds are stored as either dried or ensiled material (Rotz et al., 1989). Following harvest, alfalfa and grass hay or silage can be separated between two levels of quality for storage and feed allocation. All forage harvested with an NDF content greater than an assigned value (normally about 42%) is considered low quality feed and the remaining material is considered high quality. Separation of feeds by quality level enables more efficient allocation of the feeds to animals at various stages of growth and lactation. The hay storage model includes dry matter and nutrient losses due to microbial activity on the hay and nutrient changes due to heating of the hay (Buckmaster et al., 1989b). Dry matter lost is non- NDF material, so NDF concentration increases as dry matter is lost. Crude protein loss is 40% of the loss of other dry matter, which causes a small decrease in the concentration of protein. A portion of the protein is bound to carbohydrate during the heating process so less is available for animal utilization. Losses and quality changes in large round bales are also a function of storage method, weather conditions and bale size (Harrigan et al, 1994). Silo losses and forage quality changes are modeled for alfalfa, grass, and grain crop silages. Both tower and bunker types of structures can be used as well as bagged and baled silage. Loss and quality of forage are modeled for each plot (material harvested in three hours) throughout the storage period (Buckmaster et al, 1989a). Losses occur in five phases: preseal, effluent, fermentation, infiltration, and feedout. Dry matter lost from each phase is primarily respirable substrate; i.e., not CP or NDF. Therefore, the concentration of CP increases with the loss of dry matter. The breakdown of hemicellulose partially offsets the gain in NDF concentration that occurs through the loss of non-ndf constituents. 33

37 DAFOSYM Silage handling and feeding losses are uniformly distributed across all plant constituents. Due to uniform distribution of this loss, feed quality is not affected. In the case of hay, animals can selectively reject lower quality hay increasing the quality of that consumed (Harrigan et al., 1994). 7. Feed Allocation and Animal Performance Farm produced feeds are allocated to the dairy herd according to animal requirements and feed availability (Rotz et al., 1999). Possible feeds include: 1) low-quality forage (hay and silage), 2) high quality forage (hay and silage), 3) grain crop silage (corn and small grain), 4) high-moisture grain (corn and small grain), and 5) dry grain (corn, small grain, and soybeans). When required, these feeds are supplemented with purchased feeds, which can include 1) a degradable protein supplement, 2) an undegradable protein supplement, 3) vegetable oil or animal fat, 4) hay, and 5) corn grain. The herd is split into six groups for feed allocation; a feeding order is strategically chosen to allocate feeds where they are best used by the animal. Dry cows are fed first, heifers greater than one year of age are fed second and heifers under one year old are fed third with a mix of low quality forages and grain crop silage as the preferred forage. Feed requirements for these groups are determined first so that if there is a shortage of forage, the higher quality hay purchased will be fed to lactating cows. The remaining three groups are lactating cows at three different stages of lactation. Highest producers are fed first and lowest producers last. The preferred forage is a mix of high quality forage and grain crop silage. The preferred forage mix for any group is used when available. If not, an alternative forage mix is established. The ratio of hay to silage and/or the amount of alfalfa, grass or corn silage in the forage mix is determined from the amount of each left in storage. The preferred grain in the ration is always high-moisture grain if it is produced on the farm. The first alternative is stored dry grain and the second is purchased grain. Rations are formulated for each of the six animal groups (Rotz et al., 1999). If the feeds available cannot provide the necessary nutrients for the given milk production level (yet satisfy intake and fiber limitations), the production level is decreased to that level which can be met with the given feeds. The following five criteria are used to determine rations: 1) animal intake is limited by physical fill or energy consumption, 2) adequate forage must be fed to maintain a roughage requirement in the rumen, 3) energy requirement must be met, 4) ammonia pool in the rumen must be adequate for microbial growth, and 5) substrate must be available in the rumen for microbial growth. Physical fill and roughage are functions of the NDF, digestibility of the NDF, and particle size distribution in the feeds. An absorbed protein system is used to determine protein requirements. Rations are determined with a linear programming algorithm. For high forage diets, forage use is maximized while using as little energy and protein supplements as necessary. To achieve this objective, ration costs are minimized using relative prices of forages, grain, and supplements with homegrown forages having no cost. For a high concentrate diet, the relative price of forage is set high for lactating cow diets forcing greater use of corn and protein supplements. 8. Manure Production and Handling Manure production is modeled as feed dry matter consumed minus the digestible dry matter extracted by the animals plus urine dry matter and feed lost into the manure (Borton et al., 1995). The total quantities of silage, hay, grain, and supplements consumed by each animal group are multiplied by the fraction of indigestible nutrients of each feed. The sum of indigestible dry matters for all animal groups gives the fecal dry matter. Urinary dry matter is set as 5.7% of total urine with a fecal/urine ratio of 1.2 for heifers and 2.2 for cows. Manure dry matter is increased an additional 3% of the feed intake to account for feed losses into the manure. The quantity of wet manure handled is a function of the type and amount of bedding used and the manure handling method (manure dry matter content). 34

38 DAFOSYM Nutrients in the fresh manure are determined through a mass balance of the six animal groups (Borton et al., 1995). Manure nutrients equal the nutrient intake minus nutrients contained in milk produced and in meat produced through animal growth. Nutrient losses are subtracted to determine that available for plant growth. Nitrogen (N) losses during collection, storage, and field application are each modeled as functions of temperature, storage method, and the time between spreading and incorporation. Phosphorus (P) and potassium (K) losses are restricted to that lost during manure handling or through runoff. Since good management is assumed, uncontrolled runoff is presumed to be small. Losses of P and K are set at 5% of that applied to crop fields as fertilizer and manure. Crop nutrient requirements are based on the nutrients removed by crops and their yield. Fertilizer nutrients are added to that available from legumes and manure to predict that available for crop uptake. 9. Economic Analysis A partial budget is used to account for all costs associated with growing, harvesting, storing, and feeding of crops to the milking herd and young stock and the collection, storage, and application of manure back to the crop land. A total feed and manure cost is determined as the sum of all costs associated with these processes. A net return over feed and manure costs is the difference between the income from milk sales and the net cost of feeding the animals and handling the manure. All production costs and net returns are determined for each simulated year of weather conditions. Production costs include capital investments in machinery and structures. Annual costs for capital investments are determined by amortizing the initial price over a given life with a given real interest rate. Annual operating costs include costs of labor, fuel and electricity, maintenance and repair of machinery, land, seed, fertilizer, chemicals, and supplemental feeds. Annual requirements for each are determined by the model and multiplied by a given price to determine annual costs. Whole farm profit is also estimated by including all other major costs. The costs for feed, manure handling, animal housing, animal care and milking are subtracted from the incomes of milk, excess feed and animal sales to obtain the overall return to management and unpaid factors. Model Applications: DAFOSYM has primarily been used as a research tool for evaluating alternative technology and management strategies for dairy farms across the northern U.S, Canada, and Northern Europe. Some of the major applications follow: Alternative alfalfa harvest strategies Timeliness cost in delayed alfalfa harvest Matching equipment and silo sizes to the farm Mechanical and chemical conditioning treatments to speed field curing of forage Swath manipulation treatments to improve field curing of forage Maceration and mat drying of alfalfa Chemical and biological preservation of high-moisture hay Ambient air drying of high moisture hay Alternative storage methods for large round bales Direct-cut alfalfa silage systems vs. wilted silage Dry hay vs silage systems Economic value of losses in forage harvest and storage Comparison of silo types, bagged silage and bale silage systems Unloading methods for bunker silos Alternative strategies for manure handling Conventional, conservation and no-till tillage systems 35

39 DAFOSYM Grazing vs. confinement feeding on dairy farms Alfalfa vs. corn silage systems on dairy farms Processing of corn silage on dairy farms Alternative sources of protein supplementation for dairy cattle Level of grain feeding on grazing dairy farms Potential economic benefits for pasture yield measurement Management strategies to reduce phosphorus loading on dairy farms In addition to its primary purpose as a research tool, DAFOSYM also provides an effective teaching aid. Students in Bio-Systems Engineering, Agronomy, and Dairy Science can use the model to learn more about the complexity of the many interactions that occur within a crop and livestock production system. Students may study the effects of relatively simple changes such as the size of a tractor or other machines. Such a change influences the timing of field operations, fuel and labor requirements, the quality of feeds produced, and milk production as well as the cost of production and farm profit. More complex problems may be studied such as maximizing the profit of a given size farm or optimizing the machinery set or structures used on a farm. The model can also be used in extension type workshops. Extension field staff, private consultants, and producers may use the model to study the impacts of various technological changes on farms in their area. With some experience, the model can be used to assist with strategic planning such as providing useful information on the selection of equipment and structures for optimal farm performance or expansion. Various cropping systems and feeding strategies can also be compared along with numerous other options in farm management to determine more economical and environmentally friendly production systems. References for more information: A Windows version of the DAFOSYM model is available from the home page of the Pasture Systems and Watershed Management Research Unit ( The program operates on computers that use Microsoft Windows 95 or higher operating systems. Instructions for downloading and setting up the program are provided on the home page. The software contains an integrated help system that includes a reference manual with a more detailed description of the internal functions and algorithms of the model. Numerous journal articles have been published that document components and applications of the model. The following is a list of publications describing the major components of the model. A more extensive list can be found in the reference manual. Borton, L.R., C.A. Rotz, H.L. Person, T.M. Harrigan, and W.G. Bickert Simulation to evaluate dairy manure systems. Appl. Eng. Agric. 11(2): Buckmaster, D.R., C.A. Rotz, and R.E. Muck. 1989a. A comprehensive model of forage changes in the silo. Trans ASAE. 32(4): Buckmaster, D.R., C.A. Rotz, and D.R. Mertens. 1989b. A model of alfalfa hay storage. Trans ASAE. 32(1): Fick, G.W The mechanisms of alfalfa regrowth: A computer simulation approach. Search Agriculture. 7(3):

40 DAFOSYM Fick, G.W. and D.W. Onstad Statistical models for predicting alfalfa herbage quality from morphological or weather data. J. Prod. Agric. 1(2): Harrigan, T.M., W.G. Bickert, and C.A. Rotz Simulation of dairy manure management and cropping systems. Appl. Eng. Agric. 12(5): Harrigan, T.M, C.A. Rotz, and J.R. Black A comparison of large round bale storage and feeding systems on dairy farms. Appl. Eng. Agric. 10(4): Jones, C. A. and J. R. Kiniry (Eds.) CERES-Maize: a simulation model of maize growth and development. Texas A&M Univ. Press. College Station, Texas. Jones, J.W., K.J. Boote, S.S. Jagtap, and J.W. Mishoe Soybean development. In Modeling Plant and Soil Systems, eds. J. Hanks and J.T. Ritchie. Agronomy Monograph 31, ASA- CSSA-SSSA, Madison, WI. Mohtar, R.H., D.R. Buckmaster and S.L. Fales A grazing simulation model: GRASIM, A: Model Development. Trans. ASAE. 40(5): Ritchie, J.T. and S. Otter Description and performance of CERES-Wheat: A user-oriented wheat yield model. Pp In ARS Wheat Yield Project. ARS-38. Natl. Tech. Info. Serv., Springfield, VA. Rotz, C. A., D.R. Mertens, D.R. Buckmaster, M.S. Allen, and J.H. Harrison A dairy herd model for use in whole farm simulations. J. Dairy Sci. 82: Rotz, C.A Loss models for forage harvest. Trans. ASAE. 38(6): Rotz, C.A., D.R. Buckmaster, D.R. Mertens, and J.R. Black DAFOSYM: a dairy forage system model for evaluating alternatives in forage conservation. J. Dairy Science. 72: Rotz, C.A. and Y. Chen Alfalfa drying model for the field environment. Trans. ASAE. 28(5): Savoie, P., L.D. Parsch, C.A. Rotz, R.C. Brook, and J.R. Black Simulation of forage harvest and conservation on dairy farms. Agric. Systems. 17: Sinclair, T.T Water and nitrogen limitations in soybean grain production I. Model development. Field Crops Research. 15: Soder, K.J. and C.A. Rotz Economic and environmental impact of four levels of concentrate supplementation in grazing dairy herds. J. Dairy Sci. 84: Tsuji, G.Y., J.W. Jones, G. Uehara, and S. Balas (ed.) DSSAT version 3. University of Hawaii, Honolulu, HI. 37

41 DAFOSYM Inputs and outputs: Input information is supplied to the program through three data files: farm, machinery, and weather parameter files. The farm parameter file contains data that describe the farm. This includes crop areas, soil type, equipment and structures used, number of animals at various ages, harvest, tillage, and manure handling strategies, and prices for various farm inputs and outputs. The machinery file includes parameters for each machine available for use on a simulated farm. These parameters include machine size, initial cost, operating parameters, and repair factors. Most farm and machinery parameters are quickly and conveniently modified through the menus in the user interface of the program. Any number of files can be created to store parameters for different farms and machinery sets for later use in other simulations. The weather data file contains daily weather for many years at a particular location. Weather files for about twenty locations are available with the model, and new files may be created for other locations. All files are in a text format so they can be created or edited with most spreadsheet and text editors. When creating a new weather file, the exact format of the file must be followed. This format is similar to the standard format for weather data established by the International Benchmark Sites Network for Agrotechnology Transfer (IBSNAT) project. The first line contains a site code, the latitude and longitude for the location, the atmospheric carbon dioxide level, and an unused parameter set to zero. The remainder of the file contains one line of data for each day. The daily data includes the year and day of that year, solar radiation (MJ/m²), maximum temperature ( C), minimum temperature ( C), and total precipitation (mm). Only 365 days are allowed each year, so one day of data must be removed on leap years. DAFOSYM creates output in four separate files. Following a simulation, the files requested appear in overlaying windows within the primary DAFOSYM window where they can be selected and viewed. The four output files are the summary tables, report tables, optional tables, and parameter tables. The summary output provides several tables that contain the average performance, costs, and returns over the number of years simulated. These values include crop yields, feeds produced, feeds bought and sold, manure produced, costs of manure handling and feed production, other farm costs, income from products sold, and the net return or profitability of the farm. Values are provided for the mean and standard deviation of each over all simulated years. The report tables provide more extensive output information, which includes all the data given in the summary tables and more. In the report tables, values are given for each simulated year as well as the mean and variance over all simulated years. Optional output tables are available for a closer inspection of how the components of the full simulation are functioning. These tables include daily values of crop growth and development, a summary of the suitable days for fieldwork each month, daily summaries of forage harvest operations, annual summaries of machine, fuel, and labor use, and a breakdown of how animals are fed. Optional output is best used to verify or observe some of the more intricate details of a simulation. This output can become very lengthy and as such is only available when requested. Parameter tables can also be requested. These tables summarize the input parameters specified for a given simulation. Any number of tables can be requested where tables are grouped for major sections of model input. These sections include: crop, soil, tillage and planting parameters, grazing parameters, machine parameters, harvest parameters, storage and preservation parameters, herd, feeding, and manure parameters, and economic parameters. These tables provide a convenient method for documenting the parameter settings for specific simulations. 38

42 DAFOSYM Measures of farm performance, production costs, and the net return over those costs are determined for each simulated year of weather. All input parameters, including prices, are held constant throughout the simulation so that the only source of variation is the effects of weather. Distribution of the annual values obtained can then be used to assess the risk involved in alternative technologies or strategies as weather conditions vary. Using statistical terminology, each system alternative can be considered a treatment, and each simulated year is a replicate of farm performance for the specific weather conditions of the year. Thus a multiple year simulation provides an estimate of the frequency or probability of attaining a certain level of system performance. A wide distribution in annual values implies a greater degree of risk for a particular alternative. The selection among alternatives can be made based upon the average annual measure of performance or the probability of attaining a desired level. Several aspects of the model output can be plotted. These include the pre-harvest and postharvest crop yields, total feed and manure costs, net return for the farm, and the whole farm balance of the three major crop nutrients. Annual values of these output numbers are ranked from smallest to largest and plotted as a cumulative probability distribution. These plots can be viewed on the monitor and printed on a compatible printer. Example reports: A typical summary output for a simulation follows: Table 3.1. Average crop yields and nutritive contents over a 25 year farm analysis Preharvest Postharvest Yield Crude Yield Crude (ton DM/ac) Protein NDF (ton DM/ac) Protein NDF ALFALFA, 100 acres Cutting Cutting Cutting Cutting Total CORN, 100 acres Silage HM grain OATS, 20 acres HM grain

43 DAFOSYM Table 3.2. Feed Production and utilization for a 25 year analysis of a farm with 100 cows and 85 young stock on 220 acres of land. Unit Mean SD High-quality hay production ton DM Low-quality hay production ton DM High-quality silage production ton DM Grain crop silage production ton DM High-moisture grain production ton DM Forage sold ton DM 8 63 Grain purchased ton DM Soybean meal, 44% purchased ton DM 7 3 User defined feed purchased ton DM 40 5 Mineral and vitamin mix purchased ton DM 6 0 Average milk production lbs/cow Table 3.3. Nutrients available, used, and lost to the environment for a 25 year analysis of a farm with 100 cows and 85 young stock on 220 acres of land. Unit Mean SD Nitrogen imported to farm lb/ac Nitrogen exported from farm lb/ac Nitrogen available on farm lb/ac Nitrogen lost by volatilization lb/ac Nitrogen lost by leaching lb/ac Nitrogen lost by denitrification lb/ac Average nitrogen concentration in leachate ppm Crop removal over that available on farm % 51 4 Phosphorous imported to farm lb/ac Phosphorous exported from farm lb/ac Phosphorous available on farm lb/ac Phosphorous loss through runoff lb/ac Soil phosphorous build up lb/ac Crop removal over that available on farm % Potassium imported to farm lb/ac Potassium exported from farm lb/ac Potassium available on farm lb/ac Potassium loss through runoff lb/ac Soil potassium build up lb/ac Crop removal over that available on farm %

44 DAFOSYM Table 3.4. Annual manure production, nutrient availability and handling cost for a 25 year analysis of a farm with 100 cows and 85 young stock on 220 acres of land. Unit Mean SD Manure handled ton Manure applied to alfalfa land ton Manure applied to corn land ton Manure applied to oats land ton Manure nitrogen over crop requirement % 74 9 Manure phosphorous over crop requirement % Manure potassium over crop requirement % Machinery cost $ Fuel and electric cost $ Custom hauling cost $ Storage cost $ Labor cost $ Bedding cost $ Total manure handling cost $ Total cost per mature animal $/cow Table 3.5. Crop production, feeding and manure handling costs and the net return over those costs for a 25 year analysis of a farm with 100 cows and 85 young stock on 220 acres of land. Unit Mean SD Machinery cost $ Fuel and electric cost $ Feed, manure and machinery storage cost $ Labor cost $ Seed, fertilizer and chemical cost $ Grain drying and roasting cost $ 0 0 Land rental $ 0 0 Purchased feeds and bedding cost $ Income from feed and bedding sales $ Net feed and manure cost $ Net cost per unit of milk $/cwt Net cost as portion of milk income % Income from milk sales $ Net return over feed and manure costs $ Net return per mature animal $/cow

45 DAFOSYM Table 3.6. Total production costs and net return to management for a 25 year analysis of a farm with 100 cows and 85 young stock on 220 acres of land. Unit Mean SD Total feed cost $ Total manure cost $ Animal facilities cost $ Milking and milk handling equipment cost $ Milking and animal handling labor cost $ Animal purchase and livestock expense $ Milk hauling and marketing fees $ Property tax $ Income from milk sales $ Income from feed and bedding sales $ Income from animal sales $ Return to management and unpaid factors $

46 Soil Moisture Model Name of model : Model Purpose : The Soil Moisture Model (SMR) Pierre Gérard-Marchant 1 and Tammo Steenhuis 1 The Soil Moisture Routing model (SMR) To simulate the hydrological behavior of small rural watersheds with shallow soils and moderate slopes, and in particular to identify the surface runoff generating variable areas. Model Developers : ver. 2.0 : Pierre Gérard-Marchant ver. 1.0 : V.K. Mehta, M.S. Johnson previous versions : J. Boll, E. Brooks, J. Rossing, W.-L. Kuo, T.S. Steenhuis, M.T. Walter, J. Zollweg. Contact Persons : P. Gérard-Marchant & T.S. Steenhuis, Soil & Water Lab, Dept. of Biological \& Environmental Engineering, 222 Riley-Robb Hall, pg56@cornell.edu / tss1@cornell.edu Description of model : The Soil Moisture Routing (SMR) Model is a continuous, physically based, spatially distributed model, fully integrated into the GRASS (Geographic Resources Analysis Support System) Geographic Information System (GIS). For the sake of simplicity, the term SMR will refer in this document to both the set of conceptual and mathematical relationships among variables, and the actual computer implementation. SMR is intended as a management tool for planners or groups interested in watershed management. Therefore, it is designed to use readily available data, which can usually be obtained in electronic form: topography (Digital Elevation Map or DEM), land use (Land Use and vegetative cover Map or LUM) and soil hydrodynamic characteristics, as a combination of a Soil Type Map (STM), its corresponding Soil Characteristics Table and some statistical relationships between textural information and hydrodynamic properties. One useful and unique feature of SMR is that no significant calibration is needed. The modeled watershed is divided into small square elements, or cells. On each cell, geotopographical and soil hydrodynamic properties are assumed homogeneous, or at least uniform with respect to the hydrologic response. The cell dimension depends on the resolution of the maps available, and is usually between 10 and 30m. The cells are then decomposed in several horizontal layers, as described in fig First, two functional layers are defined: an evapotranspiration zone (corresponding to the root zone) and an underlying transmission zone. In a second complentary approach, N structural layers are defined according to their hydrodynamic properties and superimposed on the two functional layers. The number of structural layers, N, depends on the available geological data. In both cases, we assume that the shallow subsoil is bounded by a restrictive layer such as bedrock or fragipan. 1 CALS, Dept. of Biological and Environmental Engineering, Cornell University 43

47 Soil Moisture Model Figure 1: Vertical discretization of the subsoil. (a) Two Functional layers (b) N Structural layers z 1 z 2 z N Simulations are typically performed on a daily time step. The choice of this time interval is a good compromise among computation speed, result accuracy, and data availability. It significantly simplifies the resolution of the water mass balance. The user must supply an input file containing the simulation data : daily average temperature, daily precipitation and daily potential evaporation rate. The SMR code is written as a series of unix and GRASS commands grouped in scripts. An overview of the organization is presented figure 4.2. The flexibility brought by this organization allows the user to easily edit and adapt the code for his or her own needs. A first script "setup" transforms the three input maps (DEM, LUM, STM) into a set of maps of hydrodynamic characteristics. The different variables needed for the simulation are then initialized through the "init" script. The "balance" script is the backbone of SMR. It recursively processes the daily simulation data and calculates the mass water balance on each cell, by calling different subscripts : "meteo" calculates the daily volume of rainfall (RF) and snowmelt (SM), "roots" determines the root distribution in each structural layer, "redist" calculates the distribution of water in the subsoil and "output" processes and displays the resulting maps. The "postprocess" script calculates the daily hydrograph. 44

48 Soil Moisture Model Figure 4.2 Structure of SMR. The distribution of water in the subsoil is calculated at each time step for each structural layer by solving the equation : (1) where z k,i the thickness of the k th of cell i [m]; θ the mean water content [m 3 /m 3 ]; D k the volume of water flowing downward from layer k, expressed in units of cell area [m 3 /m 2 ]; W the cell dimension [m]; Q in and Q out the volume of water exchanged laterally with neighboring layers [m 3 /m 2 ]; ET the volume of water evapotranspirated [m 3 /m 2 ]; and R the saturation excess [m 3 /m 2 ]. By convention, we define for the top layer D 0 as the sum of rainfall (RF) and snowmelt (SM) volumes, and for the bottom layer D N as the volume percolating through the restrictive layer. Even though the components of the mass balance are tightly coupled one to each other, they are estimated successively for each cell for computational simplicity. An overview of the balance algorithm is presented in fig

49 Soil Moisture Model Figure 4.3 Schematic description of the water mass balance algorithm First, the average water content of layer k is updated by adding the various inputs. Rainfall and snowmelt volumes are derived from daily precipitation data and daily average temperature. The volume of water drained to layer k+1 is then calculated by solving a simplified Richards' equation: (2) where K(θ) is the hydraulic conductivity [m/s], and where capillary effects are neglected. The volume evapotranspirated is then estimated by solving the equation: (3) where U is a factor between 0 and 1 accounting for the concentration of roots in the current layer, and E the evapotranspiration rate [m/s]. We assume that E varies linearly with water content from 0 (for water contents lower than permanent wilting point) to the potential evapotranspiration rate PET, when moisture exceeds a given limit usually set to field capacity. Vertical drainage and evapotranspiration are calculated successively for each layer, starting from the top one. The volume D N is then added to a lumped `subsurface reservoir' in order to represent flow to the groundwater table. Any water in excess of saturation in layer N is then redistributed to layer N-1. Then, the water volume drained by lateral flow is calculated with a simplified Darcy's law: (4) where K eff,j,i is the transmissivity of layer j [m/s], h the hydraulic gradient, assumed equal to the actual land slope, and t the time step. This volume is distributed among the layer's downslope lateral neighbors according to elevation differences. The volume Q in is then defined as the sum of the contributions received from the upslope neighbors. Saturation excess redistribution and lateral drainage are calculated successively for each layer, up to layer 1, where the saturation excess defines the `saturation excess runoff' (SE). 46

50 Soil Moisture Model SMR outputs consist in a series of maps and a streamflow hydrograph. The choice of the maps to be output is at the user's discretion. A typical choice is a degree of saturation map, as presented in fig 4.4a. Zones with saturation excess are represented in red. By counting the days of a given month for which each cell is saturated, a map as the one presented in fig 4.4b can be obtained, which indicates areas very prone to saturation and areas where saturation is rare. Figure 4.4 Example of local output maps (a) Daily saturation degree map (b) Monthly saturation areas map By selecting contiguous cells, a soil moisture profile can be estimated along a transect. Figure 4.5 is an example of a comparison between simulated and observed water content along a transect in upstate NY. 47

51 Soil Moisture Model Figure 4.5 Comparison between observed and simulated soil moisture profiles along a transect. Hydrographs are determined at the end of the simulation. A given fraction, φ, of the volume stored in the lumped subsurface reservoir (SR) is assumed to flow to the streams, constituting the baseflow; runoff is then added to obtain the total streamflow SF, as summarized in eq. 5 : A comparison between simulated and observed hydrographs is given in fig Model Applications SMR has been successfully applied to a number of watersheds in the Catskills Mountains (Townbrook, Biscuit Brook, Irondequoit Creek, Trout Creek...). Comparison with other more complex models such as the Hydrological Simulation Program - Fortran (HSPF) showed that the accuracy of SMR was similar to these in terms of streamflow hydrograph. Only SMR predicted the spatial distribution of moisture content, saturated and runoff generating areas. Another difference between SMR and other models is that for SMR moisture distribution in the landscape and runoff generation depends mainly on the geotopography of the watersheds, when most models consider only variations with land use. Current developments of SMR include routing of surface water to streams and the transport of pollutants and pathogens. A simple overland flow routing algorithm has been developped and should soon be implemented in the code. The coding of an algorithm describing the transport of phosphorus through surface runoff is in progress. It relies on a "extraction coefficient'' approach, that links the soil test P to concentrations released in baseflow and surface runoff. (5) 48

52 49 Soil Moisture Model

53 SWAT Cannonsville Watershed References for More Information The redaction of a user manual for the new version of SMR is currently in progress, and should be finished early in Meanwhile, the reader may consult information on previous versions of SMR in references 1-8 below. References Rossing Frankenberger, J. Identification of critical runoff generating areas using a variable source area model. Ph.D. dissertation, Cornell University, Ithaca, NY, USA, 1996 Rossing Frankenberger, J., E.S. Brooks, M.T. Walter, M.F. Walter, and T.S. Steenhuis. A GISbased variable source area hydrology model. Hydrological Processes, 13(6): , Johnson, M.S. Comparative analysis of two watershed hydrologic models for a central New- York state watershed: Hydrological Simulation Program - Fortran (HSPF) and the Soil Moisture Routing model (SMR). M.Sc. dissertation, Cornell University, Ithaca, NY, USA, Kuo, W.L. Spatial and temporal analyisis of soil water and nitrogen distribution in undulating landsacpes using a GIS-based model. Ph. D. dissertation, Cornell university, Ithaca, NY, USA, 1998 Kuo, W.L., T.S. Steenhuis, C.E. McCulloch, C.L. Mohler, D.A. Weinstein, S.D.DeGloria and D.P.Swaney. Effect of grid size on runoff and soil moisture for a variable source area hydrology model. Water Resources Research, 35(11): , Mehta, V.K. A multi-layered soil moisture routing (SMR) model applied to distributed hydrological modeling in the Catskaills. M.Sc. dissertation, Cornell University, Ithaca, NY, USA, Zollweg, J.A. Effective use of Geographic Information Systems for rainfall-runoff modeling. Ph.D. dissertation, Cornell University, Ithaca, NY, USA, Zollweg, J.A., W.J. Gburek and T.S. Steenhuis. SMorMod - a GIS-integrated rainfall-runoff model. Transactions of the ASAE, 39(4): ,

54 SWAT Cannonsville Watershed Modeling the Cannonsville Watershed for Water Quality Assessment and Management Christine A. Shoemaker and Jennifer Benaman 1 Name of model: Model purpose: SWAT To use spatially distributed information to estimate effects of changes in weather and management practices on phosphorous loading to the Cannonsville Reservoir and to quantify the uncertainty in these estimates. Model developers: Arnold et al, 1998, developed SWAT 2000 and Prof. Shoemaker s research group and staff at the Cornell Water Resources Institute are developing the application of SWAT 2000 to the Cannonsville Reservoir. (See Additional Information for more details.) Contact person: Model type: Prof. Christine Shoemaker, CAS12@cornell.edu This is a semi-empirical model, developed by USDA and TEXAS A&M University. It has been applied throughout the US, including in Texas, Illinois and Pennsylvania. The model allows for crop growth, harvesting, rotation and different fertilizer application processes. SWAT is a distributed model with subregions, hydraulic response units (HRUs), and routing of water and constituents through streams and rivers. Description of model: The Cannonsville Watershed has been under a phosphorous-loading restriction due to the high phosphorus concentrations within the water body. These high concentrations are assumed to be related to the non-point source loads from the agricultural land in the basin. The county and regulatory agencies need a way to evaluate the effectiveness of the BMPs for phosphorous reduction. In addition, it would be useful to have a tool to project future conditions of the reservoir resulting from changing land use practices and BMPs as well as variability in weather. Our approach of the evaluation of the BMPs and land use practices is to adapt a basinwide watershed model for the Cannonsville Basin. The model simulates hydrology, sediment, and phosphorous transport throughout the watershed, which ultimately becomes an input to the reservoir. The model being applied is the Soil and Water Assessment Tool (SWAT), which is combined with Geographical Information Systems (GIS). SWAT is a spatially distributed model, requiring that the entire basin be divided into smaller sub-basins of similar land use, soil properties, and topography. The model uses this information, along with meteorological data, 1 School of Civil and Environmental Engineering, Cornell University, Ithaca, NY 14853; Copyrighted 2002 by Christine Shoemaker and Jennifer Benaman, Cornell University 51

55 SWAT Cannonsville Watershed river geometry information, and constituent parameters to develop the fate and transport of phosphorus and sediment over the land surface. The calibration of the model for hydrology and suspended sediment reported in Benaman et al. (2002) has an R 2 of.8 for flow and was within 1% of the average sediment loadings estimated from point data. In addition, on-going research on phosphorous calibration of the model and examination of the effects of alternative methods for determining precipitation inputs will be discussed. Introduction The Cannonsville Reservoir in Delaware County, New York, part of an 1178 km 2 watershed, is dominated by dairy pastures and forested land. This system has been designated phosphorus restricted by the New York City Watershed Memorandum Agreement (MOA). The County, the Department of Environmental Conservation (DEC), Cornell University, and other organizations are co-operating on research and projects that could lead to reducing phosphorus loadings to the reservoir (Water Resources Institute and Delaware County 1999). This ongoing co-operation for water quality improvement presents a unique opportunity for a modeling study. The focus on the watershed has resulted in a significant amount of data, which can be used for model development and calibration. Cornell s Water Resources Institute (WRI), the DEC, the United State Geological Survey (USGS), and the New York Department of Environmental Protection (NYDEP) have provided data for this study, including land use, soils, flow, and water quality measurements. In addition, because a modeling effort may aid in phosphorus management of the watershed, the county has a concerted interest in potential model developments. Figure 5.1 Location of Cannonsville Reservoir and related watershed (from Benaman et al., 2002) 52

56 SWAT Cannonsville Watershed Model Selection and Calibration Approach Distributed and Lumped Models NYDEP developed a lumped parameter watershed model of the entire New York City water supply system (New York Department of Environmental Protection 1996). This effort used a customized version of the Generalized Watershed Loading Functions (GWLF) model created by Haith et al. GWLF was used to estimate total phosphorus loading to the reservoir, and for this purpose it was an appropriate model. GWLF is a lumped parameter model, meaning all parameters are spatially averaged and the watershed is modeled as one large system. However, we wish to assess the different impacts of weather and management practices applied in various spatial locations in the watershed. In a lumped model like GWLF, no routing through the watershed is performed in the model, making it difficult to assess the impact of spatially specific management practices. However, the fact that proposed management practices would vary spatially throughout the watershed supported the development of a distributed watershed model by our group. For this purpose we choose the Soil and Water Assessment Tool SWAT. Distributed models differ from lumped parameter models in that parameters are allowed to vary spatially in the watershed instead of being incorporated into one average number. In addition, depending on how the watershed is subdivided, results from a distributed model, such as phosphorus loading, can be analyzed on a number of different spatial scales. Small scale (e.g.. farm level) or short time period (e.g. storm event modeling) cannot give us the information required to assess management decisions in terms of their long-term effect on phosphorous loading to the reservoir. SWAT is designed to establish long-term, average annual or monthly loading from the watershed. SWAT is the latest watershed model developed by the Agricultural Research Service of United States Department of Agriculture (USDA). The principle purpose of SWAT is computation of runoff and loadings from rural, especially agriculturally dominated, watersheds (Williams and Arnold 1993). Recent improvements to SWAT have made it applicable to large watersheds (up to several thousand square miles), which can be further subdivided into smaller sub-basins. Researchers have also upgraded many of its processes, including the incorporation of auto-fertilization and auto-irrigation management options, the addition of a CO 2 component to crop growth, and the integration of instream nutrient kinetics to the channel-routing equations (Benaman et al. 2000). The use of the model will probably increase in the coming years, as the EPA recently integrated the latest version of SWAT into their TMDL tool, Better Assessment for Point and Non-Point Sources (BASINS). BASINS is a GIS-based interface currently promoted by the agency as viable tool for TMDL development. Calibration of SWAT for Cannonsville Reservoir Models like SWAT have many parameters, which are adjusted within allowable limits to fit observed data in a process called calibration. Logically, a watershed model is calibrated in the following sequence: hydrology (i.e. streamflow and runoff), sediment loading, and constituent loading. This approach was applied over a 9 year, 7 month period from January 1990 to July First, the streamflow was calibrated using traditional techniques. This calibration occurred on differing spatial and temporal scales. The longest-running flow gage at Walton for the watershed drains approximately 80% of the watershed and was the primary calibration location. In addition, gages located throughout the watershed that drain smaller subbasins and have shorter periods of record (~2 years) were used during the calibration procedure. The use of these smaller subbasins 53

57 SWAT Cannonsville Watershed ensured that the model is accurately simulating the watershed on different spatial scales. The flows were compared on a daily, monthly, seasonal, and annual basis to determine if there are trends in model output or error. An R 2 value of.8 was obtained for the flow at the primary gauge station at Walton. Suspended sediment in streamflow is important to phosphorus loadings because some phosphorus is attached, or adsorbed, to sediment particles in the water. Calibration of sediment loading took a slightly different approach than flow because sediment (TSS-total suspended solids) is measured infrequently in comparison to flow. The USGS maintains flow measurements on fifteenminute intervals and reports the data as mean daily flow. Total suspended solids (TSS), which is the measurement used for sediment calibration, is not typically recorded on such short intervals. As a result, this research calibrated suspended sediment using the following approach: Data at a semicontinuous station near the primary flow gage was used, along with data interpretations performed by the DEC. These interpretations combined approximately bi-weekly TSS sampling with high-intensity sampling during high flow events to estimate monthly sediment loading. These monthly loading estimates were compared to model output using graphs and percent difference calculations, bearing in mind that the data estimates are most likely, underestimates of sediment loading (e.g. some high flow events may be missed by the sampling effort). Preliminary Results and Observations Available Data and Watershed Delineation The Cannonsville Reservoir and its surrounding watershed are under careful study due to its current phosphorus load restriction imposed by New York City. As a result, a significant amount of data exists to aid in the development and calibration of a watershed model. In addition, because the SWAT watershed model is a spatially distributed model, it requires spatial information to accurately simulate the system. Table 5.1 outlines the data used in model development and calibration. Figure 5.2 shows the primary water quality and flow gauge locations within the watershed. Because SWAT is a distributed model, it is possible to analyze the response of the watershed at varying spatial scales. In order to accomplish this, the full watershed must be subdivided into a number of smaller subwatersheds. The model output can then be analyzed at each of the subwatershed outlets to determine the impact of that area on the entire catchment. The subwatersheds established for the Cannonsville watershed followed those designated by the NYDEP that were based on major tributaries entering the West Branch Delaware River, the main river within the basin, and Cannonsville Reservoir. These 31 basins (Figure 5.2) were delineated with the aid of GIS using a digital elevation model and stream network (Neitsch and DiLuzio 1999). Each subbasin is further divided into Hydrologic Response Units (HRUs), which are determined by unique intersections of the land use and soils within the basins. These HRUs are the spatial level that the model establishes management practices such as crop growth, fertilizer application, and livestock management. For the Cannonsville, analysis determined 301 HRUs for the entire basin, resulting in an average of 10 HRUs per subbasin. 54

58 SWAT Cannonsville Watershed Table 5.3 Summary of data used in model development and calibration (from Benaman, Shoemaker, and Haith, 2002). Data Locations Per. of Rec. Water Quality Beerston 1991-pres Monitoring Town Brook 1998-pres. Phosphorus Numerous low flow Total Suspended stations Solids Streamflow W. Br. Walton Little Delaware Trout Creek Town Brook 1950-pres pres pres pres. Source NYDEP and DEC USGS Primary Use Model-to-data calibration and validation Model-to-data calibration and validation Land Use Basin-wide 1992 NYDEP Model input (HRU generation and land management) Stream Network Basin-wide Unknown NYDEP Model input and subbasin delineation Soils State Soils Geographic Database (STATSGO) Digital Elevation Model (DEM) Point Source Dischargers Basin-wide 1994 USDA/ NRCS Model input (HRU generation and land management) Basin-wide Unknown EPA Model input and subbasin delineation Basin-wide 1990-pres. WRI Model input 55

59 SWAT Cannonsville Watershed Figure 5.2 Cannonsville basin showing subwatersheds and monitoring gauge locations (from Benaman et al, 2002). Additional insight is gained by analyzing the spatial distribution of runoff and baseflow from a preliminary calibration run (Figure 5.3). This plot shows that the upland regions contribute less to runoff, on a depth basis, than many of the lowland regions. This fact may be due to a number of possibilities, including the presence of more forested and crop land in the upland region. With more forests, evapotranspiration is higher, resulting in less surface runoff to the streams. Figure 5.3 demonstrates strength of distributed modeling and will become important for sediment and phosphorus loadings. Knowing which subbasins may be contributing more of the load will help in pinpointing priority basins for implementing management practices. Figure 5.3 Spatial distribution of baseflow and surface runoff over the Cannonsville subbasins (from Benaman and Shoemaker, 2002). 56

60 SWAT Cannonsville Watershed Initial Sediment Calibration The calibrated results for sediment have average loadings that are within 1% of the DEC estimated sediment loading based on a time series measurements of sediment loading. As mentioned earlier in most watersheds, sediment measurements are taken much less frequently than flows so total sediment loading must be estimated from the point estimates. In this case the loading estimates were made for NY State DEC by Longabucco and Rafferty (1998). A flood event in January 1996 resulted in a load of almost 60,000 metric tons (MT). The model is unable to capture such a high flow event due to water movement into the flood plain and the subsequent erosion of the flood plain, which is not simulated in SWAT. Sediment-loading results can be displayed on a spatial scale (Figure 5.4). This information provides guidance to decision makers as to priority watershed which may be contributing a high percentage of total sediment load to the reservoir. It is important to note that this spatial plot is of sediment loading measured at the outlet of each subbasin. These loads are further routed through the watershed (primarily down the West Branch Delaware River) before reaching the reservoir. The causes for this spatial distribution are being currently investigated. The spatial results show a wide variance throughout the watershed, but high erosion seems to be closely tied to land use, soil type and rock cover. Figure 5.4 Total sediment erosion by subbasin for the Cannonsville Watershed. Validation Results for Flow and Calibration Benaman et al. (2002) also report validation results for flow and sediment. Validation is quite different from calibration results. In calibration, the model parameters are adjusted (within allowable limits) so that the model output is as close as possible to the observed data in the calibration data set. Validation is a step following calibration. (Often only calibration without 57

61 SWAT Cannonsville Watershed validation is done in watershed modeling.) In validation, the model is run using the parameter values that were determined during the calibration run for a new set of observations that is independent of the of the calibration data set. For the Cannonsville, the set of observations used for validations was the years 1990 to The result of the flow validation was an R 2 of 0.8. Phosphorous Modeling Incorporation of phosphorous into the SWAT model is currently underway. Subbasin manure production rates were estimated based on a cattle population survey. Manure spreading practices have been incorporated. Point source phosphorous loading has been specified. Subbasin groundwater TDP concentrations have been estimated from DEP low flow data. Our procedure for phosphorous will be similar to the process for flow and sediment, in that we will calibrate the model first for phosphorous and then later do validation runs. Future Scenario Selection After calibration and validation, the next use for the model will be to choose proposed management practices and simulate the response of the watershed to these changes. This task will allow Delaware County and DEC to analyze different options in reducing phosphorus loading to the reservoir. Ultimately, the goal would be to find a solution which is beneficial, while still being economically viable and implementable on the watershed level. Some ideas of potential management practices which could be tested include: reduction of waste water treatment plant loadings; more efficient and environmentally friendly manure storage practices; modified fertilization application rates and times; revised management of livestock grazing; change in feeding practices of livestock; or erosion abatement tillage for crops. This research will use the results from previous studies and ongoing projects (WRI and Delaware County 1999) to determine the changes in model parameters that must occur to accurately simulate proposed management practices. The actual number of management practices and forecasts play an active role in determining what scenarios will be simulated and evaluated. Uncertainty both in weather input and in model parameters will be considered. Acknowledgements This paper was written by Shoemaker and Benaman as a summary report on the completed and on-going Cannonsville modeling effort by Professor Shoemaker s research group done in co-operation with staff in the Cornell Water Resources Institute. In addition to the authors, Prof. Shoemaker s students who contributed through their research to this report include Bryan Tolson, Ching Pei Yang, and Zying Shen. The authors and Shoemaker s students have w orked closely with Steve Pacenka and Keith Porter of the Water Resources Institute, Center for the Environment, Cornell University, PH (607) , ksp2@cornell.edu. Prof. Douglas Haith is a member of Benaman s Ph.D. committee, and he contributed substantially to the model calibration reported in Benaman et al (2002). Financial support for Shoemaker, Tolson, Shen, Pacenka, and Porter s time on this project was provided in part by funding from provided by NYS Department of Environmental Conservation under the Federal Safe Drinking 58

62 SWAT Cannonsville Watershed Wa ter Act and by Cornell University. Benaman s tuition and stipend were funded under the EPA s Science to Achieve Results (STAR) fellowship, Grant U Delaware County, the DEC, Cornell University s WRI, the USGS, and the NYDEP all provided data and insight which was vital to this project s completion. Patricia Longabucco Bishop (DEC) and Steve Pacenka (Cornell WRI) were instrumental in the attainment of the sediment water quality data. In add ition, particular thanks are given to Keith Porter (Cornell WRI), Carol Stepcheck (NYDEP), and Dean Fraiser (Delaware County) for their dedication, support, and advice throughout the duration of the project. The advice of the researchers at the Blackland Research Center (Texas A&M), especially Drs. Jeff Arnold, Susan Neitch, and Nancy Sammons is greatly appreciated. References American Society of Civil Engineers (ASCE). Task Committee on Definition of Criteria for Evaluation of Watershed Models of the Watershed Management Committee, I. a. D. D., (ASCE) (1993). "Criteria for Evaluation of Watershed Models." Journal of Irrigation and Drainage Engineering 119(3): Arnold, J. G., R. Srinivason, R. R. Muttiah and J. R. Williams (1998). "Large Area Hydrologic Modeling and Assessment Part I : Model Development." Journal of the American Water Resources Association 34(1): Benaman, J., G. Ward, D. R. Maidment and W. K. Saunders (2000). "A Critical Assessment of Water Quality and Watershed Models for the Texas TMDL Process". Watershed 2000, Vancouver, B.C. Benaman, J., C. Shoemaker, D. Haith, (2002)"Calibration and Validation of a Distributed Watershed Model for Basin-Wide Management", submitted journal paper. Haith, D. A., R. Mandel and R. S. Wu (1992). Generalized Watershed Loading Functions: Version 2.0 User's Manual, Department of Agricultural and Biological Engineering Cornell University. Longabucco, P and MR Rafferty Analysis of Material Loading to Cannonsville Reservoir: Advantages of Event-Based Sampling. Journal of Lakes and Reservoir Management 14(2-3): Neitsch, S. L. and M. DiLuzio (1999). ArcView Interface for SWAT 99.2, USDA Agricultural Research Center and Texas A&M University Texas Agricultural Experiment Station. New York Department of Environmental Protection, NYDEP. (1996). GWLF Modeling of the New York City water Supply System. Kingston, NY. Peterson, J. R. and J. M. Hamlett (1998). "Hydrologic Calibration of the SWAT Model in a Watershed Containing Fragipan Soils." Journal of the American Water Resources Association 34(3):

63 SWAT Cannonsville Watershed Water Resources Institute, WRI. and N. Y. Delaware County (1999). Delaware County Comprehensive Strategy for Phosphorus Reductions. Cornell University, Center for the Environment, Prepared for the Delaware County Board of Supervisors. Williams, J. and J. Arnold (1993). A System of Hydrologic Models. Water-Resources Investigations Report , U.S. Geological Survey, Proceedings of the Federal Interagency Workshop on Hydrologic Modeling Demands for the 90's. Model Inputs and Outputs Inputs: Precipitation and other climate data as well as the values of hundreds of parameters. Spatial inputs include land use, soil type, and topographic information (elevation and slope). Outputs: Full water balance information including soil moisture, evapotranspiration, baseflow, lateral flow, and surface water runoff. Flow, sediment, and pollutant mass leaving each subbasin each day, plus summary statistics including monthly totals, R 2 values based on differences between measured and modeled variables. 60

64 Modeling P Movement from Agriculture to Surface Waters Modeling Phosphorus Movement from Agriculture to Surface Waters 1 Introduction Andrew Sharpley, Peter Kleinman, Margaret Gitau, Bil Gburek, and Ray Bryant 2 Phosphorus (P), an essential nutrient for crop and animal production, can accelerate freshwater eutrophication (Carpenter et al., 1998; Sharpley, 2000). Recently, the U.S. Environmental Protection Agency (1996) identified eutrophication as the most ubiquitous water quality impairment in the U.S., with agriculture a major contributor of P (U.S. Geological Survey, 1999). Given general environmental concerns from harmful algal blooms (Burkholder and Glasgow, 1997) and regulatory pressure to reduce P loadings to surface waters via implementation of Total Maximum Daily Loads (TMDLs) (U.S. Environmental Protection Agency, 2000), much research is now focused on better understanding factors controlling P loss from agricultural watersheds. Because of the time and expense involved in field assessment of management impacts on P loss, models often represent a more efficient and feasible means of evaluating management alternatives. Indeed, many process-based models have been developed to simulate the fate of P in soil and its transport to surface waters. In their most comprehensive form such models integrate information over a large scale, helping to define watershed scale processes relevant to P transport, highlighting appropriate best management practices (BMPs) and identifying critical source areas where BMPs are most likely to affect watershed-scale P losses. The AGNPS (Agricultural Nonpoint Pollution Source; Young et al., 1989, 1995) model was originally developed to provide estimates of runoff water quality from watersheds of up to 50,000 acres (20,000 hectares) and to quantify the effects of BMPs targeted to specific areas. To make model output more meaningful to decision makers, such as conservationists and farmers, AGNPS, which ran on a storm or flow event basis, was recently superceded by an annualized version, AnnAGNPS (Bingner et al, 2001; Croshley and Theurer, 1998). The model operates on a cell basis that makes it possible to analyze spatially discrete management units (fields) within a watershed, thereby enabling identification of individual fields that may serve as critical source areas of nutrient export. The SWAT model (Soil and Water Assessment Tool) was developed to assess the impact of land management on water quality in watersheds and large river basins (Arnold et al., 1998). The model runs on a continuous time step and is currently being utilized in a variety of largescale studies to estimate the off-site impacts of climate and management on water use and nonpoint source loadings. 1 This article is a summary of an invited paper in the Journal of Soil and Water Conservation, 2002 and presented at the Annual ASA Meeting s Nutrient Management Symposium held November 2001 in Charlotte, NC. 2 USDA-ARS, Pasture Systems and Watershed Management Research Unit, Curtin Road, University Park, PA Andrew Sharpley s ans3@psu.edu 61

65 Modeling P Movement from Agriculture to Surface Waters Other process-based nutrient transport models include, but are not limited to, ANSWERS-2000 (Areal Nonpoint Source Watershed Environment Response Simulation 2000, Beasley et al., 1985; Bouraoui and Dillaha, 1996), GAMES (Guelph Model for Evaluating the Effects of Agricultural Management Systems on Erosion and Sedimentation, Cook et al., 1985), HSPF (Hydrologic Simulation Program - Fortran, Johanson et al., 1984), ARM (Agricultural Runoff Model, Donigian et al., 1977), EPIC (Erosion-Productivity Impact Calculator, Sharpley and Williams, 1990), and the lumped parameter model GWLF (Generalized Watersheds Loading Functions, Haith and Shoemaker, 1987). For more detailed information on these models and the approaches used, reviews by Hook (1997), Leavesley et al. (1990), National Research Council (2000), and Rose et al. (1990) are available. Export coefficient models have also been widely used to predict P loading of receiving water bodies (Beaulac and Reckhow, 1982; Hanrahan et al., 2001; Johnes et al., 1996). Export coefficients define P loss from a particular source or land use in watershed, and are usually derived from actual field measured losses of P (Johnes, 1996; Johnes and Heathwaite, 1997). The models calculate watershed export of P as the sum of individual loads from each source in the watershed. This approach accounts for the complexity of land-use systems, spatial distribution of data from various sources (point and nonpoint), and permits scaling up from plot to watersheds. As export coefficients are empirical, these types of models are as accurate as input data (as are also process-based models) (Hanrahan et al., 2001). Coefficients derived from short-term monitoring of small drainage areas, however, can contribute to predictive variability (Lathrop et al., 1998). A common limitation to model application is the lack of detailed parameterization data on soil physical, chemical, and biological properties as well as on crop and tillage details. Thus, existing databases are increasingly being linked to nonpoint source models, often via Geographical Information Systems (GIS). Generally, key input data for nutrient transport models involve land use, soil texture, topography, and management practices. Once these data are in digital form, GIS techniques can be used to combine them with experimental or model results to extrapolate other properties needed for model application. In this summary we will discuss how models represent soil P release and transport, effects of mineral fertilizer and manure management on P loss, spatial resolution, and channel processes that translate edge-of-field losses to water body inputs. We then highlight future modeling efforts that will address these issues. Soil phosphorus release and transport Dissolved P Most nonpoint source models simulate dissolved P transport in overland flow as a function of the extractability of P in the surface 5 cm of soil (e.g., CREAMS, AGNPS). This can be represented by: Dissolved P = Extraction Coefficient * Available soil P * Overland flow volume [1] where dissolved P is orthophosphate loss in overland flow (kg ha -1 ), available soil P is the amount of P in a unit depth of surface soil (usually 5 cm; Sharpley, 1985b) as estimated by recommended soil test P methods (STP; kg ha -1 5 cm -1 ), and extraction coefficient is the fraction of STP that can be released to overland flow for a given flow event volume (cm). Extraction 62

66 Modeling P Movement from Agriculture to Surface Waters coefficients can be determined as the slope of the linear regression of STP and overland flow dissolved P (Fig 6.1a). Figure 6.1 Relationship between the concentration of dissolved P in overland (a) and subsurface flow (b) from 30 cm deep lysimeters and the Mehlich-3 extractable soil P concentration of surface soil (0-5 cm) from a central PA watershed (adapted from McDowell and Sharpley, 20001a and Sharpley et al., 2000) a. Overland flow 1000 Di ssolved P (µg L -1 ) b. Subsurface flow from lysimeters y = 1.98x + 79 R2 = y = 0.93x + 60 R2 = Mehlich-3 extractable soil P (mg kg -1 ) 63

67 Modeling P Movement from Agriculture to Surface Waters A similar relationship holds for subsurface flow P and surface STP, although the slope of the relationship (0.93) is almost half that for overland flow (slope of 1.98) (Fig. 6.1b). The dependence of dissolved P transport in subsurface flow as well as overland flow, suggests the importance of preferential flow pathways, such as earthworm burrows and old root channels, in the downward movement of P through the soil profile (Kleinman et al., 2002; McDowell and Sharpley, 2001a; Sims et al., 1998). Most models (e.g., CREAMS, AnnAGNPS etc) use a constant extraction coefficient value, assuming that STP extractability is similar among soils. A reanalysis of data published by McDowell and Sharpley (2001a), Pote et al. (1999), and Sharpley and Smith (1994) relating STP and overland flow dissolved P, revealed a range of extraction coefficient values (Fig. 6.2). Figure 6.2 Extraction coefficient (slope of the relationship between soil test P and dissolved P in overland flow) as a function of erosion to represent soil vegatative cover for sites in Arkansas, Oklahoma, New York, and Pennsylvania (data adapted from Pote et al., 1999; McDowell and Sharpley, 2001a and Sharpley and Smith, 1994). 20 Extraction coefficient Native grass / pasture No till Reduced till Conventional till y = 1.25x 0.30 R 2 = ,000 10,000 Erosion (kg ha -1 yr -1 ) Decreasing soil cover 64

68 Modeling P Movement from Agriculture to Surface Waters Extraction coefficients were much greater for cropped (8 to 17) than grassed watersheds (1 to 4). Using erosion as a surrogate for land cover, extraction coefficients increased with greater erosion or decreased soil cover (Fig. 6.2). A larger soil P extraction coefficient represents a greater release of P as overland flow dissolved P per unit STP increase. This can be attributed to a lower degree of interaction between surface soil and overland flow with a protective grass cover than for a cropped situation, where the soil is more exposed to overland flow. Other factors which influence P release among soils, include the dominant forms of P in soil, texture, aggregate diffusion, degree of interaction between soil and water, organic matter content, vegetative soil cover, and P sorption capacities. Particulate P As the sources of particulate P in overland flow and stream include eroding surface soil, stream banks, and channel beds, processes determining erosion also control particulate P transport. In general, eroded particulate material is enriched with P compared to source surface soil, due to the preferential transport of finer (i.e., clay-sized), more sorptive soil and organic particles or greater P content than coarser inorganic particles (i.e., sand-sized). Sharpley (1985a) found that the plant available P content of sediment in overland flow was on average three times greater (or more enriched) than that of source soil and 1.5 times greater for total, inorganic, and organic P. The degree of P enrichment is expressed as a P enrichment ratio (PER); that is the P concentration of sediment discharged divided by that of source soil. In assembling enrichment ratio information for the CREAMS model, Menzel (1980) concluded that for particulate P, a logarithmic relationship as in Equation [2] seemed to hold for a wide range of soil vegetative conditions. Ln ER = Ln Sediment discharge [2] where sediment discharge is in kg ha -1. Most nonpoint source models adopted this approach to predicting particulate P transport in overland flow. This relationship is based on the welldocumented increase in particulate P loss with increasing erosion (Fig. 6.3). Based on the total P concentrations of source soils for each of the watersheds represented in Figure 6.3, PER decreases with an increase in erosion. As erosion increases, there is less particle-size sorting by overland flow, relatively less clay-sized particles are transported, and P enrichment, thus, decreases (Fig. 6.3). Once an appropriate PER is obtained form sediment discharge, particulate P loss can be calculated as: Particulate P = Total soil P * Sediment concentration * PER * overland flow volume [3] where particulate P loss in overland flow (kg ha -1 ), total soil P is the amount in a unit depth of surface soil (usually kg ha -1 5 cm -1 ), sediment concentration is g sediment L -1 overland flow, PER is calculated from Equation [2], for a given flow event volume (cm). 65

69 Modeling P Movement from Agriculture to Surface Waters Figure 6.3 Particulate P loss and enrichment ratio of eroded sediment as a function of erosion in overhead flow from watersheds at El Reno, OK (adapted from Sharpley et al., 1991 and Smith et al., 1991). ha -1 ) kg Particulate P ( Particulate P P enrichment ratio P enrichm e nt ratio ,000 10, ,000 Erosion (kg ha -1 ) Fertilizer and manure management Fertilizer and manure management, as it affects P availability to overland flow over the near term, can profoundly affect prediction of P transport in overland flow. While soil P represents a source of P enrichment in overland flow, the application of fertilizer and manure to soil, including type, method, timing and rate of P application, can temporarily overwhelm relationships derived between STP and P in overland flow (Sharpley and Tunney, 2000). As such, accounting for fertilizer and manure management in P models is essential to their accuracy under certain conditions. However, most models do not directly address the effect of applied P (fertilizer or manure) on P transport in overland flow. Rather, added P is incorporated into the soil P pool and the extraction coefficient adjusted. Thus, P transport in overland flow as affected by the amount, type, method, and time after applying P is, in general, poorly represented and predicted. Mineral fertilizer and manure represent concentrated sources of soluble P that can greatly increase dissolved P losses in overland flow. Consequently, the concentration of soluble P in these sources may provide effective predictions depending upon the solubility of the P source, method of application, rate of application and timing of application relative to the overland flow event (Fig. 4; Kleinman et al., 2002). Surface application of manure and mineral fertilizer concentrates P at the extreme soil surface where it is vulnerable to removal by overland flow (Eghball and Gilley, 1999; McDowell and Sharpley. 2001b; Sharpley et al., 1984). Although 66

70 Modeling P Movement from Agriculture to Surface Waters injection, knifing and immediate incorporation of manure and fertilizer may decrease P losses, cultivation may increase site vulnerability to particulate P loss due to greater erosion potential (Romkens et al., 1973; Andraski et al., 1985). Modifying the effect of P source and application method on P concentrations in overland flow is the timing of application relative to when overland flow event occurs (Sharpley, 1997; Westerman and Overcash, 1980). Immediately following application of a P source, the potential for P loss peaks and then declines over time, as applied P increasingly interacts with the soil and is converted from soluble to increasingly recalcitrant forms (Edwards and Daniel, 1993). As a result, fertilizer and manure management effects on overland flow P are predictable. Figure 6.4. Relationship between water extractable manure P and the dissolved P in overland flow one week after manure or mineral fertilizer was broadcast (90 kg total P ha - 1 ) on a Hagerstown silt loam soil (7 cm hr -1 rainfall for 30 minutes). P in overland flow (µg L -1 ) Dissolved 6000 Poultry manure Swine slurry 4000 Dairy compost Poultry litter 2000 Dairy manure Poultry compost Water extractable manure P (g kg -1 ) 67

71 Modeling P Movement from Agriculture to Surface Waters Spatial data requirements for modeling Models that assess non-point sources of P loss from agricultural lands rely on spatial data as input. Land use, soil properties, and topographic data that include stream locations and watershed boundaries are commonly required inputs. However, with an expansion in the geographical scale at which watershed processes are to be modeled, there is a great increase in the size of associated spatial databases. Data and parameter requirements also increase rapidly as models become more mechanistic to better represent physical and chemical processes and spatial interactions involved in P loss. The complexity of managing these large databases in support of a watershed model, can limit the degree of spatial resolution of existing models. Spatial parameters are frequently lumped so that units having similar soil, land use and topographic characteristics, respond the same to driving variables, such as those used to simulate runoff generation. However, spatially lumped parameters can pose a problem when responses from lumped units cannot distinguish between relative spatial locations of individual units, which can be critical in determining P export from a watershed to a water body. To overcome the spatial data limitations thus far identified, a nested modeling approach is recommended. Field and farm scale models that incorporate the knowledge of P source and transport processes involved in P loss, can be supported with highly detailed spatial databases that are already available in some areas or could be easily developed in others. Results and generalizations from these models could be aggregated to represent sub-basins in a simpler, less mechanistic model that requires lower spatial resolution. Similarly, results from sub-basin models could be further aggregated to represent whole watersheds of several hundreds of square kilometers in size. Beyond that scale and with enough knowledge of processes operating in individual sub-watersheds, the principles of mapping could be invoked to derive generalizations about large watersheds that span multiple physiographic regions, such as the Chesapeake Bay Watershed and Mississippi River Basin. Map units of the Major Land Resource Areas (MLRA) of the U.S. are defined on the basis of topography, soils and land use, and, therefore, are ideally suited for extrapolating detailed studies of whole watersheds to the broader area of the MLRA map unit. Defining future best management practices The implementation of P control measures has often been carried out with insufficient knowledge as to the suitability of these practices for P control. A large number of BMPs exist; their suitability likely varying depending on the particular situation. Given that BMP impacts are largely site specific (Deer and Company, 1995; U.S. Environmental Protection Agency, 1993; Baker and Johnson, 1983), defining future BMPs for P control depends a great deal on being able to establish the effectiveness of these BMPs under the variety of field conditions that are constantly encountered. 68

72 Modeling P Movement from Agriculture to Surface Waters There are several factors that complicating BMP assessment in a field situation, which include site variability, lack of controlled replication, and length of study needed. In turning to models, we try to eliminate some of these complications. While it is recognized that models greatly simplify the natural system, they do provide a means of carrying out complex BMP evaluations. Nonetheless, the large amounts of data that has accumulated over the years can be extremely useful in working on a modeling approach to BMP evaluation (Gitau et al., 2001). An initial step in modeling BMPs would be the characterization of the BMPs of concern with regard to their mechanisms of operation - such as source (soil P and type, rate, and form of P applied) and transport (runoff and erosion) factors controlling P loss. This characterization would enable identification of source and transport mechanisms impacted by particular BMPs, and thus the determination of model changes that would be necessary to fully represent the BMPs (Gitau et al., 2001). Transport and impact of P in surface waters In-channel processes modify the potential for agriculture to impact a downstream freshwater body. As surface water impacts drive activities such as TMDL development, understanding in-channel P transport processes and the impact of transported P on downstream water bodies is necessary to link upstream changes in agricultural management with downstream water quality impacts. For example, McDowell et al. (2002) examined the processes controlling sediment P release to the Winooski River, VT, the largest tributary to Lake Champlain. Iron-oxide strip P (algal-available P) of the river sediments adjacent to agricultural land (3.6 mg kg -1 ) was significantly greater (P<0.05) than that of sediments adjacent to forested land (2.4 mg kg -1 ). Notably, impoundment (731 mg kg -1 ) and reservoir sediments (803 mg kg -1 ) had greater total P concentrations than did river sediments (462 mg kg -1 ). This was attributed to more fines (< 63 µm) in impoundments and reservoirs (64%) than in river sediments (33%). Consequently, impoundment and reservoir sediments had lower abilities to release P to solution in the shortterm, thereby acting as P sinks. The results of this research clearly demonstrate that there is a strong influence of fluvial hydraulics on the properties of sediment within river systems. The input and delivery of fine sediment enriched with P was influenced by adjacent land use. The fluvial sediment, particularly at the outflow of the river into Lake Champlain, represents a store of P, which has a long-term potential to release a large amount of P to overlying waters. In the short-term, however, river flow and physical properties of the sediments will influence the amount of sediment P leaving the watershed in the Winooski River, VT. Thus, modeling of channel processes must account for variability in flow, local sources of P, and sediment properties, particularly near the point of impact. Because of these complexities, channel processes and changes in P forms and loads are not currently simulated in most watershed models (Hanrahan et al., 2001). Intuitively, biological responses are different among water bodies, with variations in geographic location, climate, water residence times, and surface area and depth of water body. For example, the Cannonsville Reservoir (part of the New York City water supply system) 69

73 Modeling P Movement from Agriculture to Surface Waters flushes in a matter of months, while Cayuga Lake (the longest Fingerlake in New York State) has a mean water residence time of about 12 years. Also, the Chesapeake Bay has a completely different set of critical biological indicators in comparison with the Gulf of Mexico (National Research Council, 2000). In fact the ratio of watershed drainage area and bay water volume (2410 km 2 km -3 ) is nearly an order of magnitude greater than any other lake or bay in the world (next is Gulf of Finland at 380 km 2 km -3 ). As a result, simulation as well as management of the biological response within the Chesapeake Bay presents unique challenges, because of the relatively large area for nutrient source inputs that must be considered. Although we can simulate P loss in overland flow, the related effects of agricultural management, and how nutrients cycle within a water body, it is still difficult to relate P loss as a function of watershed management to the biological response of a receiving water body. Because of the scales involved, connectivity, and dominant processes in terrestrial and aquatic systems, watershed and water body response models have tended to develop independently. Summer et al. (1990) attempted to link watershed (AGNPS) and lake process (FARMPND) models. However, a lack of adequate water monitoring data (chemistry and flow rate) limits rigorous testing of their ability to simulate a lake s response to changes in agricultural management and climate. Summary This discussion has presented background information on processes controlling P transport in overland and subsurface flow from agricultural landscapes and how nonpoint source models have attempted to simulate P losses. New information on soil and site dependency of extraction coefficients relating STP and overland flow dissolved P and the use of enrichment ratios to estimate particulate P transport, should be incorporated into these models. Also, incorporation of new formulations describing the release and transport of inorganic and organic P from manure in overland and subsurface flow, will improve model predictions of P loss following land application of manures. However, it is clear that there is a lot of information already available on the fate and transport of P in agricultural landscapes and the effectiveness of various BMPs to minimize this loss through source or transport controls. Mechanisms are being developed to apply this information through innovative data base management and integration with existing models, to better use existing data, rather than recreating the wheel. Many complex models are available and are gaining greater acceptance with managers and planners, as computers become more powerful, cheaper and we are generally more comfortable using them. However, because models yield clear numerical results with which to gauge progress, they have a strong appeal to policymakers and managers, an appeal that can sometimes bring false confidence and misconception (Boesch et al., 2001). It has been said that while all models are wrong, some are useful. It is of critical importance that modelers clearly define what his or her model is useful for and what it is not designed to do. Likewise, users must decide what they want to accomplish with a model. For example, one must consider the scale (field, watershed, or basin), time (flow event, annual, or multi-year), and level of accuracy (0.1 or 10 lbs ac -1 year -1 ) that needs to be simulated, as well as the amount of parameterization data available. Thus, a key to useful simulation of P loss is selection of the appropriate model and data to run it. If, for instance, one needs to identify areas in a watershed at greater risk for P loss 70

74 Modeling P Movement from Agriculture to Surface Waters to target remedial BMPs, then site vulnerability tools such as the P Index are available (Lemunyon and Gilbert, 1993; Gburek et al., 2000). On the other hand, P indices are not designed to quantify P loss as are many nonpoint source models described earlier in this discussion. Even so, it is clear that there can be a great deal of uncertainty in model computations. Uncertainty arises in connection with an imperfect representation of the physics, chemistry and biology of the real world, caused by numerical approximations, inaccurate parameter estimates and data input, and errors in measurements of the state variables being computed. Whenever possible, this uncertainty should be represented in the model output (e.g., as a mean plus standard deviation) or as confidence limits on the output of a time series of concentrations or flows. The tendency described earlier for decision makers to believe models, because of their presumed deterministic nature and exact form of output, must be tempered by responsible use of the models by engineers and scientists, such that model computations or predictions are not over-sold or given more weight than they deserve. Above all, model users should determine that the model computations are reasonable in the sense of providing output that is physically realistic and based on input parameters that are within accepted ranges. The role of modeling will be more and more important over the next decade in making management and policy decisions related to conservation programs and water quality enhancement and enforcement. Also, the availability of water monitoring data is increasing in response to water quality concerns in the U.S. and other parts of the world and providing new opportunities to develop, calibrate, and test watershed models. As we move forward, however, an interdisciplinary approach is needed that involves hydrologists, soil scientists, engineers, economists, animal scientists, and possibly rural sociologists. With the knowledge that many and varied working models exist, our efforts should be directed to the improvement or adaptation of exiting models, rather than reinventing or developing new models, except where major limitations have been clearly defined. Finally and most importantly, it is essential that the most appropriate model be carefully selected to meet a user s needs, in terms of level of predictive accuracy needed, input data available, and scale of simulation being considered (both time and space). More detail on information given in this report can be obtained from the full article that was published in the Journal of Soil and Water Conservation, September - October 2002 issue. References Andraski, B.J., D.H. Mueller and T.C. Daniel Phosphorus losses in runoff as affected by tillage. Soil Sci. Soc. Am. J. 49: Arnold, J.G., R. Srinivasa, R.S. Muttiah, and J.R. Williams Large area hydrologic modeling and assessment Part 1: Model development. J. Am. Water Resour. Assoc. 34:

75 Modeling P Movement from Agriculture to Surface Waters Baker, J.L. and H.P. Johnson Evaluating the effectiveness of BMPs from field studies. p In Agricultural Management and Water Quality. F.W. Schaller and G.W. Bailey (eds.), Ames, IA: Iowa State University Press, IA. Beasley, D.B., E.J. Monke, E.R. Miller, and L.F. Huggins Using simulation to assess the impacts of conservation tillage on movement of sediment and phosphorus into Lake Erie. J. Soil Water Conserv. 40: Beaulac, M.N., and K.H. Reckhow An examination of land use nutrient export relationships. Water Res. Bull. 18: Bingner, R.L., F.D. Theurer, R.G. Cronshey and R.W. Darden AGNPS 2001 Web Site. Boesch, D.F., R.B. Brinsfield, and R.E. Magnien Chesapeake Bay eutrophication: scientific understanding, ecosystem restoration, and challenges for agriculture. J. Environ. Qual. 30: Bourao ui, F. and T.A. Dillaha ANSWERS-2000: Runoff and sediment transport model. J. Environmental Engineering 122(6): Burkholder, J.A., and H.B. Glasgow, Jr Pfiesteria piscicidia and other Pfiesteriadinoflagellates behaviors, impacts, and environmental controls. Limnol. Oceanogr. 42: Carpenter, S.R., N.F. Caraco, D.L. Correll, R.W. Howarth, A.N. Sharpley, and V.H. Smith Nonpoint pollution of surface waters with phosphorus and nitrogen. Ecol. Applic. 8: Cook, D.J., W.T. Dickinson, and R.P. Rudra GAMES - The Guelph Model for Evaluating the Effects of Agricultural Management Systems in Erosion and Sedimentation. User's Manual. Tech. Rep School of Engineering, Univ. of Guelph, Guelph, Ont. Croshley, R..G and F.G. Theurer AnnAGNPS Non Point Pollutant Loading Model. In Proceedings First Federal Interagency Hydrologic Modeling Conference April Las Vegas, NV. Deere and Company Managing Nonpoint Source Pollution in Agriculture. Technical Report No Moline, IL: Deere and Company Technical Center. Donigian, A.S., Jr., D.C. Beyerlein, H.H. Davis, Jr., and N.H. Crawford Agricultural Runoff Management (ARM) Model Verison II: Refinement and testing. 294 pp. U.S. Environ. Prot. Agency. EPA 600/ Environ. Res. Lab., Athens, GA. 72

76 Modeling P Movement from Agriculture to Surface Waters Edwards, D.R., and T.C. Daniel Drying interval effects on runoff from fescue plots receiving swine manure. Trans. Am. Soc. Agric. Eng. 36: Eghball, B. and J.E. Gilley Phosphorus and nitrogen in runoff following beef cattle manure or compost application. J. Environ. Qual. 28: Gburek, W.J., A.N. Sharpley, A.L. Heathwaite, and G.J. Folmar Phosphorus management at the watershed scale: A modification of the phosphorus index. J. Environ. Qual. 29: Gitau, M.W., E. Schneiderman, W.J. Gburek, and A.R. Jarrett An Evaluation of Best Management Practices Installed in the Cannonsville Reservoir Watershed, New York. Proceedings 9th National Non Point Source Monitoring Workshop, August. Indianapolis, IN. Haith, D.A. and L.L. Shoemaker Generalized watershed loading functions for stream flow nutrients. Water Resources Bulletin 23(3): Hanrahan, G., M. Gledhill, W.A. House, and P.J. Worsfold Phosphorus loading in the Frome catchment, UK: Seasonal refinement on the coefficient modeling approach. J. Environ. Qual. 30: Hook, R.A Predicting Farm Production and Catchment Processes. A Directory of Australian Modelling Groups and Models. CSIRO Publishing, Melbourne, Australia. Johnes, P.J Evaluation and management of the impact of land use changes on the nitrogen and phosphorus load delivered to surface waters: The export coefficient modeling approach. J. Hydrol. 183: Johnes, P.J., and A.L. Heathwaite Modelling the impact of land use change on water quality in agricultural catchments. Hydrol. Proc. 11: Johnes, P.J., B. Moss, and G. Phillips The determination of total nitrogen and total phosphorus concentrations in freshwaters from land use, stock headage and population data: Testing of a model for use in conservation and water quality management. Freshwater Biology 36: Kleinman, P.J.A., A.N. Sharpley, B.G. Moyer and G.F. Elwinger Effect of mineral and manure phosphorus sources on runoff phosphorus. J. Environ. Qual.: In press. Lathrop, R.C., S.R. Carpenter, C.A. Stow, P.A. Soranno, and J.C. Panuska Phosphorus loading reductions needed to control blue-green algal blooms in Lake Mendota. Can. J. Fish. Aquat. Sci. 55:

77 Modeling P Movement from Agriculture to Surface Waters Leavesley, G.H., D.B. Beasely, H.B. Pionke, and R.A. Leonard Modeling of agricultural non-point source surface runoff and sediment yield - a review from the modeler=s perspective. p In: D.G. Decoursey (ed.), Proc. Int. Symp. Water Quality Modeling of Agricultural Non-point Sources, Part l. USDA-ARS 81. U.S. Government Printing Office, Washington, DC. Lemunyon, J.L., and R.G. Gilbert The concept and need for a phosphorus assessment tool. J. Prod. Agric. 6: McDowell, R.W., and A.N. Sharpley. 2001a. Approximating phosphorus release from soils to surface runoff and subsurface drainage. J. Environ. Qual.30: McDowell, R.W., and A.N. Sharpley. 2001b. Phosphorus losses in subsurface flow before and after manure application to intensively farmed land. Sci. Tot. Environ. 278: McDowell, R.W., A. Sharpley, and A.T. Chalmers Chemical characterisation of fluvial sediment: The Winooski River, Vermont. Ecological Engineering. In press. Menzel, R.G Enrichment ratios for water quality modeling. p In. W. Knisel (ed.), CREAMS - A Field Scale Model for Chemicals, Runoff and Erosion from Agricultural Management Systems. Vol. III. Supporting Documentation, USDA, Cons. Res. Rep. 26. U.S. Govt. Printing Office, Washington, DC. National Research Council Clean coastal waters: Understanding and reducing the effects of nutrient pollution. National Academy Press, Washington, DC. Pote, D.H., T.C. Daniel, D.J. Nichols, A.N. Sharpley, P.A. Moore, Jr., D.M. Miller, and D.R. Edwards Relationship between phosphorus levels in three Ultisols and phosphorus concentrations in runoff. J. Environ. Qual. 28: Romkens, M.J.M, D.W. Nelson and J.V. Mannering Nitrogen and phosphorus composition of surface runoff as affected by tillage method. J. Environ. Qual. 2: Rose, C.W., W.T. Dickenson, H. Ghadiri, and S.E. Jorgensen Agricultural nonpointsource runoff and sediment yield water quality (NPSWQ) models: modeler s perspective. p In: D.G. DeCoursey (ed.), Proc. Int. Symp. Water Quality Modeling of Agricultural Non-point Sources, Part 1. USDA-ARS 91. U.S. Government Printing Office, Washington, DC. Sharpley, A.N. 1985a. The selective erosion of plant nutrients in runoff. Soil Sci. Soc. Am. J. 49: Sharpley, A.N. 1985b. Depth of surface soil-runoff interaction as affected by rainfall, soil slope, and management. Soil Sci. Soc. Am. J. 49:

78 Modeling P Movement from Agriculture to Surface Waters Sharpley, A.N Rainfall frequency and nitrogen and phosphorus in runoff from soil amended with poultry litter. J. Environ. Qual. 26: Sharpley, A.N Editor Agriculture and Phosphorus Management: The Chesapeake Bay. CRC Press, Boca Raton, FL. Sharpley, A.N. and S.J. Smith Wheat tillage and water quality in the Southern Plains. Soil Tillage Res. 30: Sharpley, A.N. and H. Tunney Phosphorus research strategies to meet agricultural and environmental challenges of the 21st century. J. Environ Qual. 29: Sharpley, A.N., and J.R. Williams (eds.) EPIC-Erosion/Productivity Impact Calculator. I. Model documentation. U.S. Department of Agriculture Tech. Bull Sharpley, A.N., S.J. Smith, B.A. Stewart and A.C. Mathers Forms of phosphorus in soil receiving cattle feedlot waste. J. Environ. Qual. 13: Sharpley, A. N., S.J. Smith, J.R. Williams, O.R. Jones, and G.A. Coleman Water quality impacts associated with sorghum culture in the Southern Plains. J. Environ. Qual. 20: Sharpley, A.N., P.J.A. Kleinman, R.J. Wright, T.C. Daniel, B. Joern R. Parry, and T. Sobecki The National Phosphorus Project: Interfacing agricultural and environmental phosphorus management in the U.S. p In Steenvooreden, J. (ed.) International Conference on Agricultural Effects in Ground and Surface Waters. International Association of Hydrologic Sciences, Wageningen, The Netherlands. Sims J. T., R.R. Simard, and B.C. Joern Phosphorus losses on agricultural drainage: Historical perspectives and current research. J. Environ. Qual. 27: Smith, S. J., A.N. Sharpley, J.W. Naney, W.A. Berg, and O.R. Jones Water quality impacts associated with wheat culture in the Southern Plains. J. Environ. Qual. 20: Summer, R.M., C.V. Alonso, and R.A. Young Modeling linked watershed and lake processes for water quality management decisions. J. Environ. Qual. 19: U.S. Environmental Protection Agency Guidance Specifying Management Measures for Sources of Non-point Pollution in Coastal Waters. EPA-840-B93-100c. U.S. Govt. Printing Office, Washington, DC. U.S. Environmental Protection Agency Environmental indicators of water quality in the United States. EPA 841-R U.S. EPA, Office of Water (4503F), U.S. Govt. Printing Office, Washington, DC. 75

79 Modeling P Movement from Agriculture to Surface Waters U.S. Environmental Protection Agency The Total Maximum Daily Load (TMDL) program. EPA 841-F U.S. EPA, Office of Water (4503F), U.S. Govt. Printing Office, Washington, DC. U.S. Geological Survey The quality of our nation=s waters: Nutrients and pesticides. U.S. Geological Survey Circular 1225, 82 p. USGS Information Services, Denver, CO. Westerman, P.W., and M.R. Overcash Short-term attenuation of runoff pollution potential for land-applied swine and poultry manure. p In: Livestock waste - A renewable resource. Proc. 4th Int. Symp. on Livestock Wastes, Amarillo, TX. April Am. Soc. Agric. Eng., St. Joseph, MI. Young, R.A., C.A. Onstad, D.D. Bosch, and W.P. Anderson AGNPS: A nonpointsource pollution model for evaluating agricultural watersheds. J. Soil Water Conserv. 44: Young, R.A., C.A. Onstad, and D.D. Bosch AGNPS: An agricultural nonpoint source model. p In: V.P. Singh (ed.), Computer models of watershed hydrology. Water Resources Publication, Highlands Ranch, CO. 76

80 Modeling P Movement from Agriculture to Surface Waters Introduction New Nutrient Management Approaches Using Precision and Information Technologies Harold M. van Es 1 Precision Technologies and Agricultural-Environmental Management: Sustainability of livestock farms can be improved through the effective use of precision and information technologies in creating comprehensive resource management plans for each farm, and basing management decisions on space-time referenced information sources. The components that must be linked include farm profitability, crop and animal productivity and nutrition, and the impact of farm practices on the environment. Decision tools need to be developed and implemented that capitalize on these new technologies, including computer-based record keeping, global positioning systems, automated variable-rate applicators and yield monitors, geographical information systems, dynamic simulation models, site-specific weather data, and remote sensing for the purpose of improved management of farm inputs and wastes while enhancing farm system profitability. The complexity of livestock systems, the dominant agricultural industry in New York, and the elevated environmental concerns associated with it provide considerable potential for the successful adoption of precision and information technologies, and consequently for advances in improving farm management and reducing environmental impact. Stakeholders: The 50-member New York Precision Agriculture Alliance was organized in 1998 to bring together researchers, educators, farmers and industry representatives and provide direction to the adoption of new technologies. In a December 2000 meeting Alliance members identified strategic needs and developed a plan ( to better capitalize on the potential use of precision agriculture technology in environmental management, including better linkage of watershed-scale and farm scale use of GIS and GPS technologies. Watershed programs (e.g., Upper Susquehanna Coalition, New York City water supply system, Skaneateles Lake, and Lake Champlain) and professional organizations like the Northeast Dairy Producers Association and the New York State Agricultural Business Association similarly are looking for better integration of new technologies. 1 Department of Crop and Soil Sciences, Cornell University. 77

81 Modeling P Movement from Agriculture to Surface Waters Management Issues Field-Scale Nutrient Variability: The recent interest in precise management of crop inputs has resulted in a re-evaluation of the concept of the management unit. Soil sampling for nutrient analysis is now done at the sub-field scale by using grid sampling. Natural soil and geologic variability as well as variations in management history often resulted in considerable variability in nutrient and ph status of soils. We recently characterized site-specific soil test results for five farms and determined that field-scale variability was much higher on dairy farms than cash grain operations. P levels on one field receiving regular manure applications ranged from 13 to 70 mg kg -1, and had a welldefined spatial distribution (Fig. 7.1). Such variability implies that site-specific nutrient applications are imperative, and that environmental assessments (e.g., P Index) require consideration of the site-specific nature of both source and transport factors. Fig. 7.1: Variability of soil test P levels on a 10 ha field receiving dairy manure. Nutrient Indexing: The Phosphorus Index (PI) is being adopted as a field scale assessment tool that can relate site characteristics and management practices to the relative movement of phosphorus from that site. It was designed to be a management tool that uses readily available site information and separately considers source and transport factors. Due to high variability in source and transport factors on many manured fields (Fig. 7.1) a precision (site-specific) approach to field management appears an obvious next step in improving P management. Similarly, N management can be more precisely conducted using a dynamic indexing approach, which is currently being considered at the national level. Variations in Fertilizer Response: Most current methods for determining fertilizer and manure rates, including those promulgated by Cornell University, are based on an expected (average) yield response based on information including yield potential, soil type, manure applications, cropping history, etc. This approach is scientifically sound, but does not account for the spatial and temporal variations in yield response and environmental losses of nutrients. Notably, supplemental N fertilizer needs vary greatly from year to year depending on the weather patterns, especially during the early 78