Page Banner

United States Department of Agriculture

Agricultural Research Service

Vision, Mission, Strategic Plan
headline bar

Unit Vision

Provide leadership in systems research for developing sustainable and adaptive integrated agricultural systems.

Unit Mission

Enhance economic and environmental sustainability of agricultural production systems by:

  • Synthesizing and quantifying biological, chemical and physical processes at the whole-system level.
  • Conducting cooperative field research to fill knowledge gaps, especially for plant response to stress conditions & global change, and scaling up from plots to fields and larger areas.
  • Developing computer models of agricultural systems to support field research and analysis of major issues, emphasizing water quality and water conservation, production, precision farming, and climate change.
  • Providing farm-level computerized decision support technology and information system packages to farmers, ranchers, agricultural consultants and action agencies for evaluating sustainability of alternative farming/ranching options.
  • Creating an Object Modeling System with a library of science and related tools to assemble customized modular models.
  • Collaborating nationally and internationally to evaluate and improve knowledge and products.

Strategic Issues/Problem Areas

1. Whole System Integration and Modeling — Essential to Agricultural Science and Technology in the 21st Century
In the 20th Century, we made tremendous advances in discovering fundamental principles in different scientific disciplines that created major breakthroughs in management and technology for agricultural systems, mostly by empirical means. However, as we enter the 21st Century, agricultural research has more difficult and complex problems to solve. The environmental consciousness of the general public requires us to modify farm management to protect water, air, and soil quality, while staying economically profitable. At the same time, market-based global competition in agricultural products is challenging economic viability of the traditional agricultural systems, and requires the development of new and dynamic production systems. Frequency of extreme climatic events such as droughts has recently increased, possibly due to global climate change, requiring changes in management. Our customers, the agricultural communities, are asking for more timely transfer of research results in an integrated usable form to aid them in meeting these challenges through improved cropping and management. Fortunately, the new electronic technologies and remote sensing can provide us a vast amount of real-time information about crop conditions, near-term weather, and global markets, that can be utilized to develop a whole new level of management tools. However, we need the means to capture and make sense of this vast amount of site-specific data.

To address all the above needs requires us to better understand the whole system. Agricultural systems involve highly complex interactions among soil, plant, weather, and management components that are beyond a human brain to comprehend quantitatively. Modern computer technology can complement and assist the human brain in this process. The scientists need to improve on how to present their research results in the context of the whole agricultural system; this requires synthesis and quantification of experimental data based on fundamental principles and laws. The system models are indeed the product of this synthesis and quantification of current knowledge based on fundamental principles and laws. These models then form the basis of decision support systems for transferring research findings to producers under different soil and climatic conditions for making complex management decisions.

Integration of system models with field research has the potential to raise agricultural research to a higher plateau. It is also an essential first step to improve model usability and make a significant impact on the agricultural community. This integration will benefit field research and models in the following ways:

  • Promote a systems approach to field research.
  • Facilitate better understanding and quantification of research results.
  • Promote quick and accurate transfer of results to different soil and weather conditions, and different cropping and management systems outside of experimental plots.
  • Help research to focus on the identified fundamental knowledge gaps and make field research more efficient, i.e., get more out of research per dollar spent.
  • Provide the needed field tests of the models before delivery to other potential users—agricultural consultants, farmers/ranchers, state extension agencies, and federal action agencies (NRCS, EPA, and others).

The National Science and Technology Council, Committee on Environmental and Natural Resources, 1997 report entitled, “Integrating the Nations Environmental Monitoring and Research Network and Programs: A Proposed Framework”, recommended a national framework that links systematic observations and monitoring with predictive modeling and process research. The 1999 report of the National Academy of Sciences and Engineering on “New Strategies for America’s Watersheds” identified several critical support functions; one of these is the integration of theory, data, simulation models, and expert judgment to solve practical problems and provide a scientific basis for decision-making.

The field-tested model can be used as a decision aid or guide for best management practices, including site-specific management or precision agriculture (Ahuja and Ma, 2002), as tools for in-depth analysis of problems in management, environmental quality, global climate change effects, and other new emerging issues, and as guides for planning and policy making. The most desirable vision for agricultural research and technology transfer is to have a continual two-way interaction among the field research, process-based models of agricultural systems, and management decision support systems (Figure 1). The field research can certainly benefit from the process models as described above, but also a great deal from the feedback from the management decision support systems (DSSs), to be cutting-edge. On the other hand, field research forms the pivotal basis for models and DSS. The DSSs generally have models as their cores (simple or complex).

Figure 1. Our vision of the role of system models in agricultural research and technology transfer

 

System modeling has been a vital step in many scientific disciplines. We would not have gone to the moon successfully without the combined use of good data and models. In automobile designing, computer models of the system have replaced the scaled physical models of the past. Models have also been used extensively in designing and managing water resource reservoirs and distribution systems, and in analyzing waste disposal sites. Although agricultural system models have shown promise (Ahuja et al., 2002), a lot more work is needed to realize their great potential applications in field research and technology transfer. We have been and are attempting to do this by working cooperatively with ARS field scientists of several Units in evaluation and application of models, providing training in models use to many groups, and in giving guest lectures on models to a number of classes at the Colorado State University.

 

A need for a whole-system approach does not mean that we will always need a whole- system model as a guide to management.  After an initial analysis of the system has shown that only a certain component, such as soil erosion, or a certain insect or disease, is the main problem that needs to be corrected, a model or a technology guide that addresses only this component of the system may be adequate.  An important requirement, however, is that there is a consistency in the science used among components and whole-system models.

 

2. Models to Optimize the Use of water in Limited-Water Agricultural Systems

Much of agriculture in the arid western US is water-limited. Typical production systems consist of rain-fed crop/range-livestock systems or irrigated cropping/ pasture systems where water is available. Prolonged drought in the last few years has aggravated the availability of water for both dryland and irrigated production. Many climate models predict a more intense hydrologic cycle in the future with altered precipitation patterns, and greater frequency of severe droughts in mid-continental North America. As a result, there is a great concern of crop productivity, especially for dryland systems. At the same time, the increasing water demands for drinking, sanitation, urban irrigation, industry, and ecosystem service are outbidding and reducing the amount of irrigation water available for agriculture. To obtain optimum return from limited rainfall and irrigation water, while maintaining the natural resources, producers need whole-system and quantitative management tools to help them manage the use of available water and minimize associated inputs on site-specific or field-specific bases. These tools should help determine appropriate crop sequences, and optimal use of limited rainfall and irrigation water with respect to the amounts and timings of rainfall, critical crop growth stages, soil fertility, and weather conditions. Information is also needed to help determine potential alternative crops that would be suitable during periods of droughts and help producers determine how to optimally allocate limited water among crops. There is currently great excitement about growing bio-energy crops in the area, including the dryland oil seed crops and irrigated corn or other biomass crops. The tools being investigated should also be able to evaluate the long-term economics of bio-energy crops and see to it that enough crop residues remain on the soil surface to prevent erosion and maintain soil organic matter and productivity.

 

The above decision tools for optimization of the complex issues require the use of process level models of cropping systems as a whole to synthesize the available experimental data, extend the results beyond the experimental periods and soil conditions, and derive alternate management scenarios. Process level system models are based on synthesis and quantification of disciplinary knowledge and important interactions among the system components. These models also provide a systematic approach to design and implement a multi-disciplinary research program that will greatly enhance the efficiency of future field research for developing sustainable agriculture, enable a fast transfer of technologies to farmers, provide guides for planning and site-specific management, and inform policy makers and the general public on the major issues and tradeoffs, costs and benefits of alternatives.

 

These system models need to be calibrated and validated with good field experimental data at selected locations in the region, and then used to: 1) extend current research results to multiple years of historical climate beyond the limited experimental years and to other soils in the area; 2) derive the new optimum cropping and management strategies for future selective field testing; and 3) derive simpler management guides or tools for producers. Furthermore, in order to prepare for the projected climate-change effects on water, along with increased temperatures and carbon dioxide, there is an urgent need for validated models to project the effects of these changes on agricultural systems in the arid western US, devise and evaluate adaptation strategies to deal with these changes.

 

3. Models to Evaluate Effects of Management and USDA Conservation Practices on Water Quality and Crop production at Field, Farm, and Watershed Sales

Since the 1970s, water quality impairment of both surface and ground waters from agricultural sources has been recognized as major national issue in the U.S. As a part of the new research to ameliorate water quality, the ARS also recognized the need for computer models to evaluate water quality impacts of cropping systems and the associated management practices, and to guide a change in management to minimize these impacts. In the mid 1980s, two ARS teams were set up to develop two major models—NLEAP (Nitrate Leaching and Economic Assessment Package) and RZWQM (Root Zone Water Quality Model). Along with the Soil Plant Nutrients Unit, the GPSR Unit had a major part in developing NLEAP. It was designed as a screening model for quick estimation of nitrate leaching hot spots in cropping systems, and has been used extensively by NRCS and scientists all over the world. On the other hand, under leadership of the GPSR Unit, the RZWQM was developed as a comprehensive model of root zone processes that influence water quality, as well as soil water storage and efficient water use. It is an integrated physical, chemical, and biological process-based model that simulates the effect of management on plant growth and the movement of water, nutrients, and pesticides in runoff and through the root zone of a cropping system to shallow groundwater. The model allows simulation and evaluation of a wide spectrum of management practices and scenarios on water and water quality, such as no-tillage and residue cover vs. conventional tillage; rates, methods, and timing of application of water, fertilizers, manures, and various pesticide formulations; and different crop rotations up to 100 years. The model contains the special features of rapid transport of surface-applied chemicals to through soil macropores to deeper depths, groundwater and tile flow. The first version of the model was released in 1992, following which it underwent extensive evaluation, refinement, and validation, in cooperation with numerous ARS and university users. A new Windows98 user interface was developed to make it easier to use.

The model development was essentially completed in 1999. The model is now being used nationally and internationally, and the Unit is providing some user support. The model as a whole or its parts are also being used in other projects of the Unit described below. For the latter purpose, crop growth models developed by the DSSAT group and used all over the world, and improved routines for water and temperature simulations in residue-covered soil from SHAW model, were linked recently to RZWQM. At the request of U.S. Geological Survey and EPA, the model application has been extended from just the root zone to 30 meters depth.

 

In 2004, we started developing a prototype component-based farm to watershed scale model in the Object Modeling System to assess the effects of the USDA-NRCS recommended conservation practices on water, water quality, and production. The work is in progress.

 

4. Improving Modeling of Management Effects on Soil Properties and Processes and the Response of Crops under Stress Conditions

For models to guide precision agriculture and precision management of our water resources, it is crucial to be able to quantify the effects of agricultural management and cropping practices on soil properties and soil-plant-atmosphere processes, including soil water/chemical movement and retention, and water quality. The Unit initiated new pioneering research in this critical knowledge-gap area during the development of RZWQM, but which is continuing for the ongoing and future projects. New models and parameters are needed for how tillage and consolidation change the soil water retention curve and hydraulic conductivity via the change in effective porosity, effect of wheel tracks on these properties, how long-term no-tillage cropping and residue systems change surface aggregation and soil water properties, infiltration, soil water storage, evaporation,  soil temperature, and crop emergence, development, and yield; and how grazing intensity affects rangeland water storage and production in the long run. Further improvements are needed in modeling preferential transport to groundwater through no-till caused decayed root channels and earrs, the role of surface soil aggregates in holding back applied chemical and increasing macropore transport.

 

Modeling root growth is still a black box. How root growth and active root distribution with depth change with soil conditions and how they affect the root water uptake, nutrient uptake, and soil water, chemical, and heat movement? Is this movement still one-dimensional--perhaps for closely planted crops like wheat, but what about row crops? We either need a 2-dimensional model of root growth, water uptake, and water flow and chemical leaching in crop row vs. interrow zones and from bands of N in the cropped ridges, or ingenious 1-D simplifications. The processes of water and nutrient uptake, passive and active, also need improvements.

 

Most of the existing crop growth models do not adequately respond to water stress conditions. Effects of high temperatures and CO2 also need to be improved. Earlier, the Unit developed SHOOTGRO model that has the most detailed phenology and development processes for wheat crop. Based on this work, the Unit is developing new improved models of crop phenology and development under water stress. These new models will then be linked to the advanced knowledge of plant physiological processes of leaf growth, transpiration, photosynthesis, and carbon allocation to different plant parts under stress conditions at different growth stages. The new models should also include how increased carbon dioxide and temperature will interact with water stress and effect plant growth and yield in crop and rangeland systems. The new approaches based on the detailed energy balance will guide the improvements needed in current models.

 

5. Scaling and Modeling Space-Time Variability of Landscape Processes for Precision Management

Agricultural land management has to deal with a high degree of spatial variability of land and temporal variability of weather even at field scale.  This complexity increases a great deal further at farm and watershed scales.  In the last few years, there has been a nation-wide interest in so-called “precision agriculture”, which consists of site-specific, spatially variable, management to optimize production and minimize both on-site and off-site adverse effects on water quality and quantity ( Vanden Heuvel, R. M. 1996. The Promise of Precision Agriculture.  Journal of Soil and Water Conservation 51:38-40).  However, no methods and tools are available to guide this management of landscape and climate variability.  For this purpose and to aggregate expected results from plots to field, farm, and watershed scales, we need a framework for integrated spatial analysis, and distributed modeling, and up- and down-scaling for different land areas. This framework will then allow downscaling of watershed level observations, such as TMDLS, to farm and field levels for the purpose of site-specific management.  A scientifically robust scaling framework is the greatest need for making breakthroughs in transferring research knowledge across scales, and in understanding and managing large areas (National Research Council, 1991: Opportunities in Hydrologic sciences). 

Measurement of agricultural processes across multiple scales has become more feasible in recent years due to the improvement of automated data collection technology.  Tremendous amounts of site-specific data (e.g., soil properties and crop yield) are being collected; unfortunately, scientific analysis capability lags behind the technological advances driving data collection efforts. Models for predicting plant growth/development and flow/transport in agricultural landscapes are simultaneously improving, but nearly all of these models are designed only for homogeneous areas, or represent soil-plant-water processes in a very simplistic manner at large scales.  To manage variable, real-world landscape and climate variability across field, farm, and watershed scales for improving water quality and increasing efficiency of precision agriculture, research efforts must: 1) quantify spatial variability of landscape by relating to causal factors, such as the topography, and develop new physical and statistical methods of scaling; 2) improve spatial data analysis and model parameterization through application of these scaling methods; 3) provide improved spatial modeling tools, including surface and subsurface interactions among different spatial units, for enhancing site-specific management; and 3) improve simulation of management effects on soil and plant processes in a field and the way their effects on water, water quality, and crop growth are transmitted across to other fields, farm, and watershed. 

 

6. Simpler Planning and Management Decision Tools to help producers in planning and management

 

The US agricultural community is facing severe hardships at the farm and ranch levels from depressed commodity prices, weather fluctuations ranging from droughts to flooding, and foreign competition. Concerns from the environmental community about agricultural impacts on water quality and greenhouse gas emissions (e.g., global climate change) have further challenged farmers and ranchers. Decision aids based on the synthesis of current research knowledge for the production systems can help farmers deal with complexities. The GPSR Unit team recently developed a whole farm/ranch Decision Support System (DSS), Great Plains Framework for Agricultural Resource Manage

Agriculture has become a highly complex socio-economic-environmental enterprise. ment (GPFARM), for long-term strategic planning for the central Great Plains. GPFARM allows evaluation of different cropping systems, range-livestock systems, and integrated cropping-range-livestock systems, and associated management practices, for their economic, environmental, and resource sustainability on a long-term  (15-20 years+). Integrated farm/ranch enterprise management is a unique feature of GPFARM. Farm/Ranch cooperators of GPFARM have asked for an enhanced DSS to allow yearly or seasonal tactical planning and management of their operations based on climatic (especially rainfall) conditions and price fluctuations, and other risk factors. For example, the producers want a tactical tool to decide in Spring what crop to plant in May based on the known soil moisture, expected climate, and price conditions, and the risk associated with the decision. Ranchers would like to predict forage production in the coming season, especially during ongoing droughts, to decide the number of cattle to keep. The tool may also be used to guide within-season management, such as irrigation, fertilizer, and herbicide applications. There is also a national interest in developing a tool for evaluating carbon sequestration and green-house gas emissions in different agricultural systems. Further, the GPFARM is currently limited to six crops common in a four-state region of the central Great Plains, and to cow-calf animal operations on rangeland.  Actual and potential GPFARM users have requested a broader geographic scope extending to a national level, with enhanced capability for modeling  crop growth and addition of an environmental and economic risk assessment component for seasonal cropping and animal management.  This expanded DSS would provide nationwide assistance to farmers and ranchers, and would help Federal and State conservation and regulatory agencies to make better management decisions across the country.

 

8. Integration of Research Information into a Database for Decision Support for Resource Conservation Planning and Water Quality Protection by NRCS

 

Decision-making about the appropriate conservation practices to place on the landscape for protecting or enhancing water quality requires the integration of data, expert opinion, and knowledge. Evaluation of best management practices for water quality in the past has not accounted for the differences in efficacy of these practices due to soil, landscape, climate, and crop or agronomic characteristics.  The problems being considered were often single factor (e.g., erosion) that did not require the interaction of several management components to make decisions about the effect of various options on water quality criteria and production. These options may include changes in crop rotations, tillage, nutrients, pesticides, and conservation practices.  All of these considerations include some aspect of type, rate, or timing.  This type of multiple  decision-making is central to the improvement of water quality and it require the integration of several interacting factors that include weather, position in the watershed, management history, soils, and geology.  Producers and action agencies, e.g., Natural Resources Conservation Service (NRCS), do not have a system tool available in which comparisons among different system configurations can be easily made for a range of conditions and no formal method of comparing among systems for their water quality or economic impact.  Development and implementation of best management practices for impacting water quality and production at field and watershed scales will require procedures that can identify the best combination of practices before producers will alter their management systems.

Resource conservation planning tools are primitive compared to current science and information technology. Recent development in agricultural system models, decision support tools, and computer technologies could provide an improved tool for NRCS conservation planning that would lead to increased adoption of practices enhancing water quality and profitability. However, use of models directly by NRCS field personnel is still not feasible at this time. A new approach towards a DSS is to create an integrated research information database as a core of the DSS in place of a model. A system model, validated against available experimental data, is used to generate production and environmental impacts of different management practices for all major soil types, weather conditions, and cropping systems outside the experimental limits. This model-generated information is then combined with available experimental data and the long-term experience of farmers and field professionals to create the database. The database can be combined with an economic analysis package. It may also be connected to a so-called “Multi-objective Decision Support System” for determining trade-offs between conflicting objectives, such as economic return and environmental quality. It is also very flexible in generating site-specific recommendations.

 

9. An Advanced Modular Modeling Framework for Agricultural Systems and International Collaboration for Building Models of the Future

 

At present, the ARS and university scientists have more than a hundred small to large models or software packages to help evaluate or guide management practices; some of them are system-wide models whereas others are component models or software.  Examples of large system models are:  cotton and soybean models, Gossym and Glycim or Soygro; the crop models of CERES and CROPSGRO family;  erosion models, WEPP and WEPS; water quality models, GLEAMS, RZWQM, AGNPS, and SWAT; and general purpose models, EPIC and CROPSYST.  Examples of smaller, single-purpose, models are NLEAP and WEEDCOM.  Unfortunately, these tools now exist as an array of individual disparate models.  The large system models have been extremely expensive to develop ($15-30 million each).

 

At ARS customer workshops, the NRCS and other users of models or software have reported that different models or software tools do not give the same results for, say, soil erosion or crop yield, because these tools use different science in key areas like hydrology.  There may also be conflicting data, scales, and methodologies.

 

Moreover, it is becoming extremely difficult to maintain and support such a large number of packages.  In particular, yesterday’s monolithic models are very difficult to update, add to, or interface with other models, have diminishing technical support as the original developers retire, and lack the flexibility to meet today’s needs for more integrated analysis of changing resource issues.  It is difficult to transfer the new knowledge to the customers quickly through the use of these old tools.

 

All of the above reasons indicate a need for a new framework of model development that integrates all existing models and all future models into a common, collaborative, and flexible system.  Such a system will maintain modularity, reusability, and interoperabililty or compatibility of both science and axiliary components.  The system will also recognize the fact that different categories of applications may require different levels of scientific detail and comprehensiveness, as driven by problem objectives, scale of application, and data constraints.  In other words, there can be alternate components for the same process. These functionalities of the system will be obtained by establishing standard libraries of inter-operable science and axiliary components or modules that provide the building blocks for a number of similar applications.  Module libraries have been successfully used in several domains, such as the manufacturing, transport, and other systems (Top et al, 1997; Breunese et al, 1998; Praehofer, 1996).  One of the earliest modular model developments was done for SHE, the European Hydrologic System Model (Abbot et al, 1986; Ulgen et al, 1991).  Leavesley et al (1996 reported the conversion of the Precipitation Runoff Modeling System (PRMS) to a Unix-based Modular Modeling System (MMS) for hydrologic modeling.  Leavesley et al (2002) presented some successful applications of this concept.

 

In 1997, an ARS-NRCS-USGS Interagency Workshop reviewed the MMS and other similar approaches and unanimously endorsed the development of an advanced Object Modeling System (OMS) for Agricultural and Natural Resource Systems that will:

 

  • reduce duplication of effort in agricultural and natural resource modeling
  • improve the quality and currency of model code
  • make simulation models much easier to build, access, understand and use
  • facilitate long term maintainability of existing and new models
  • lead to greater consistency of modeling for particular problems and scales
  • enhance response and delivery times in scientific modeling projects
  • ensure creditability and security of  model implementations
  • function on any major computing platform

 

There is no software commercially available to build models from modules composed of existing source code.  Some technology is available which uses a common interface for running different pre-built models, such as groundwater and surface water models.  These systems cannot build models from modules.

 

Most current investments in model development programs by our university partners (e.g., The University of Florida DSSAT Modeling Group), other government agencies (Corps of Engineers, Environmental Protection Agency), and foreign countries (e.g., the Cooperative Research Centre (CRC) for Catchment Hydrology in Australia; the APSIM group in Australia; the European Commission Framework Programme 5: Harmon IT) are in the area of modular or object modeling, all of which use a basic concept of providing proven and sound scientific modules of the past made available for use in custom-designed integrated analyses.

 

In 2001, several federal agencies – Nuclear Regulatory Commission, NOAA, EPS, COE, USGS, and USDA – entered into a Memorandum of Understanding to collaborate in developing and using models.  These agencies enthusiastically endorsed the development and deployment of the Object Modeling System.

 

The ASR has developed the OMS in partnership with NRCS. The system is now being used to develop a prototype watershed model for NRCS Conservation Effects Assessment Project (CEAP).

 

 

 


Last Modified: 11/19/2008
Footer Content Back to Top of Page