Management data tend to be sparsely available and representing continuity of plant populations is challenging. Advancing our ability to understand how grasslands are managed – to understand, for example, what species are planted, what inputs are provided, what grazing management is applied – is centrally important for improving our ability to model pasture and range land systems. Planted pastures and native grazing lands both contain a variety of species, some of which are more palatable, nutritious, grazing-resistant, or fire-resilient than others. A more open, data-rich environment could facilitate evaluation of a variety of approaches for representing long-term dynamics, which could address several important grassland management and assessment issues. Managing grass swards to maintain desirable plants is a primary goal of grassland management, but one for which modeling tools have offered limited assistance. Models that represent vegetation dynamics are also desirable for understanding longer-term changes in species that can impact productive capacity, sensitivity to degradation, and carbon dynamics . Year-to-year variability is a key component for understanding potential utility and risk of relying on grassland forage resources. Next generation models that enhance our ability to forecast this risk would mark a substantial and meaningful advance. There is a need for better links between the agricultural modeling communities and ecological researchers studying long-term vegetation dynamics. The primary use of forage resources is for grazing animals, yet most grassland models are only loosely coupled with grazers . Better integration through grazing effects on grasslands, grazer distributions across landscapes,plastic pots plants forage demand/consumption, livestock/wildlife movement, etc., would enhance the ability of models to contribute to important emerging issues.

For example, holistic grazing management, in which several aspects of management vary in response to a variety of different cues from the land and expectations about future conditions, can be impossible to evaluate with current modeling frameworks. A system that integrated user demand into the model development process could lead to implementation of new data-management feedback loops within models. Such interactions between users and producers of information could direct data collection to facilitate model use. Models that better represent grazer-grassland interaction are also crucial for understanding how efficiently livestock use forage resources, what is necessary to sustain wildlife populations, and how much grassland output might be available for other uses .The Intergovernmental Panel on Climate Change has reviewed the existing evidence for how climate change may affect weeds, pests, and diseases . One issue with this evidence base is that there is a clear publication bias towards reports of increased threats – people often do not bother to write up no-effect results. There is a general recognition that we need good models to help tease out different effects that changing weather will have simultaneously on both crops and the organisms that compete with or attack them. There has already been some work applying crop physiology-type models to weeds, and developing more mechanistic models of the effect of temperature on insect pests. There is an opportunity and need for more integrated models that include interactions between organisms, for example between weeds and crops, and between pests and the predators and parasites that attack them. A variety of different approaches are possible, and there is a need for an AGMIP-type approach to help the community decide how best to move forward.Highly contagious diseases of livestock present a major threat to agriculture, both in the developed and developing worlds.

Diseases may be chronic in livestock populations, emerge from wildlife reservoirs, or possibly be introduced deliberately by man as an act of bio-terrorism. Models are required to help understand how a disease will spread, and to help policymakers design optimal interventions. These models must encompass not only the epidemiology of the disease but also how it is affected by agricultural practices and in particular the movement of livestock by farmers. There have been significant recent advances in this area, often building on work on human diseases. For example, it is now possible to take livestock movement data and use it to parameterize an epidemiological model . There are the beginnings of a model comparison movement in human epidemiology; livestock disease epidemiology would also benefit from this approach.There is intense current research activity into novel genetic methods of insect control. Most of this work is currently directed at the insect vectors of human diseases such as malaria, though the same methodology can be applied to insect pests of crops and of course the vectors of livestock diseases. The greatest advantage of these approaches is that they involve self-sustaining interventions that spread naturally through a pest population, although because they are nearly all classified as genetically modified, the regulatory issues surrounding them are complex. Cutting-edge modeling work in this field involves joint population and genetic dynamic models, many of which are explicitly spatial. This topic is likely to be one of the most important and exciting areas of modeling as applied to agriculture over the next few decades.Integrated agricultural technologies, defined as the integration of improved genetics, agronomic input, information technology, sensors, and intelligent machinery, will play a pivotal role in agriculture in the years to come. These innovations will be driven by economic forces, by the need to produce more food with limited land and water for the increasing population, and at the same time by the push to save resources to reduce the environmental impact associated with food production. While these changes are occurring now in the commercial scale industrialized agricultures of the world, many of these technologies have the capability to be adapted to conditions in other parts of the world.

The cell phone now allows farmers in rural areas almost everywhere in the world to have low-cost information about prices, for example. Similarly, it is likely that unmanned aerial vehicles will rapidly be adapted to conditions around the world and used to carry out activities such as monitoring crop growth and pest occurrence, and improve management decisions. In large-scale, capital-intensive agricultural systems, these technologies are rapidly leading to the automation of many production activities, particularly machinery operation and decisions about input application rates. The automation of agriculture began in the mid-nineties, resulting in large amounts of data available to farmers and agribusiness companies. Farm machinery are now often equipped with high precision global positioning system controllers, which allow all activity on the farm to be recorded, geo-referenced, and stored on remote computers: “in the cloud.” All modern tractors collect data on a continuous basis and are equipped with wireless connectivity for data transmission. Harvesters record the yield at a particular location, planters can vary the plant spacing or type of seed by location, and sprayers can adjust quantity and type of fertilizer, fungicide or pesticide by location; all to a granularity of just a few square meters. Yield monitoring can now be linked to unmanned aerial vehicle imagery to produce a prescription map for the farmer to implement. These private data could also provide tremendous benefit to the researcher community, should access be increased. Producers in some regions of the world now have historical crop yield data for their fields at very high resolution. Combined with advanced satellite-based imagery, high-resolution spectral and thermal data obtained from UAVs,plastic nursery pots and weather forecasts, growers have most of the critical inputs required to convert this “big data” into an actionable management plan with equipment that can vary fertilizer and other inputs spatially within a field. Despite these rapid advances in the sophistication and automation of farm equipment, a vital piece of the equation is still lacking: the analysis of the vast amount of newly available data in order to provide the farmer with a map of what action to take where and when. Most variable rate application is currently managed by farmers, using rule-of-thumb and empirical approaches, and not by using a systems approach that accounts for the interaction of soil, crop, management, and weather. Thus much of the power of automation remains unexploited. In order to realize the full potential of more sophisticated equipment, new modeling systems for precision agriculture are needed. These systems could be based on comprehensive predictive crop yield models that combine publicly available data, such as soil type, weather, and prices, along with location-specific data from farmers’ yield maps of their fields, to provide a prescriptive crop management plan at high spatial resolution, as in Fig. 2. This type of system could deliver automated crop simulations, crop management strategy recommendations, process-based variable rate prescriptions, risk assessments, continual in season simulations, integration of in-season crop scouting UAVs flight information, pest management prescriptions and accurate harvest recommendations via simple-to-use apps, websites, or smart phones. In addition to the farm-to-landscape scale analysis represented in Fig. 2, there will be a growing demand for agricultural systems models to simulate and integrate the different components of the agricultural value chain, to meet both policy requirements and corporate sustainability goals . Genetics, agronomic management , weather, soil, information technology and machinery will need to be linked in a system approach to address these informational needs. This is a new frontier for agricultural system modeling that would extend to the broader food system and raise additional data and analytical challenges.

As illustrated in Fig. 2 and discussed in the previous section, various management and production data are becoming available through mobile technologies . An example of this analytical capability is the AgBizLogic™ software developed by several university extension programs, which allows managers to calculate short-term profitability and rates of return on long-term investments . Similar proprietary software tools are being developed and used. These analytical tools could be linked with modules that track or predict environmental outcomes such as soil erosion and net greenhouse gas emissions . Low-bandwidth versions of these tools need to be developed for use in areas where mobile phone technology is a limiting factor. Analytical tools need to be adapted to fit small-holder systems as indicated by the NextGen Use Cases. The flood of data on physical land-use, water availability and use, and yields coming from mobile devices and remote sensing systems suggest that both the biophysical and behavioral aspects of farm production at specific locations can be estimated by sequential learning processes. The use of advances in computational methods such as machine learning and remote sensing data is illustrated by analysis of the impact of the 2009 and 2014 droughts on California agriculture, which demonstrated the advantages of better data .To facilitate the use of models for various locations and systems, and to link to crop and livestock system simulation models, economic models need to be incorporated into modules with standardized inputs and outputs. Various types of economic models are available in the literature, including farm-level optimization models, regional positive quadratic programming models, econometric land-use models, and regional impact assessment models . User needs should dictate which types of models should be used depending on informational needs. Methods and protocols are required to link regional economic models with market equilibrium models . Some progress has been made on this front but much more development is needed to address various aggregation and dis-aggregation issues . Generalization of behavioral assumptions and investigation of their effects on investment and policy analysis is also needed.There is a rich literature on risk modeling which could be incorporated. Recent advances in the expectations formation literature and the behavioral economics literature could be investigated for use in agricultural systems models.The application of different farm improvement methods has explicit winners but also unintended ‘casualties’ and perverse incentives. From a development standpoint, it is essential to understand these dynamics to ensure that appropriate policies are developed to maintain equal opportunities for all sectors of society. For example, in many cases, rich farmers are the ones who adopt technologies early. This factor could potentially disrupt power relationships in markets, thus affecting poorer farmers. In this case it is essential to design alternative options and safety nets for poorer farmers to prevent widening the gap and making them more vulnerable. New models should improve our understanding of these processes, as we move from single farm models to multi-farm and regional models. Methods utilizing population-based data are providing improved capability to represent distributional impacts and vulnerability .Current agricultural system models typically operate at the point/ field scales with an emphasis on vertical fluxes of energy, water, C, N and nutrients between the atmosphere, plant and soil root zone continuum.