In addition to continuing to collect empirical data, models of switchgrass biomass production are now being used to address many of these questions. Models of bioenergy crops can be split into two distinct types. Correlational or statistical approaches rely on relationships between environmental variables and empirical biomass estimates. In contrast, mechanistic or process-based approaches simulate the actual processes that govern crop growth. Correlational or statistical models have been used to estimate yields across large spatial extents (Barney and DiTomaso 2010; Evans et al. 2010; Jager et al. 2010; Wullschleger et al. 2010). However, little information is included on soil type, nutrient availability, and management practices, which are known to have large impact of biomass production (Muir et al. 2001; Fike et al. 2006). Instead, processes-based simulations of plant growth include detailed information on climate, soil dynamics, and management (Williams et al. 1989; Kiniry et al. 1992; Kiniry et al. 1996; Arnold et al. 1998; Del Grosso et al. 2005; Di Vittorio et al. 2010; Miguez et al. 2011). These models have been used to not only estimate yields but also to analyze water use efficiency, management practices, long-term effects on soil properties, and the impact of climate change (Kiniry et al. 1996; Kiniry et al. 2008; Brown et al. 2000; Sarker 2009). Due to their wider range of applications, this chapter will focus on process-based simulation models and will highlight recent applications to switchgrass.

Mechanistic models of plant growth that have been used to simulate switchgrass production include Agro-BGC, ALMANAC, BIOCRO, DAYCENT, EPIC, and SWAT (Williams et al. 1989; Kiniry et al. 1992; Kiniry et al. 1996; Arnold et al. 1998; Del Grosso et al. 2005; Di Vittorio et al. 2010; Miguez et al. 2011). These models were created for different purposes (i. e., tracking greenhouse gas (GHG) emissions, water erosion, nutrient cycling, plant growth, etc.). Correspondingly, these models vary in their functions and amount of detail incorporated to simulate growth. Despite these differences, each model shares the following basic functionality. First, they simulate biomass production by specifying light interception, conversion of sunlight to biomass, and partitioning of biomass into structural components (such as below ground roots and above ground shoots). Second, they simulate soil water dynamics, which depends on precipitation, run off, and evapotranspiration. Third, they simulate soil C and N dynamics. Lastly, each model simulates the effect of water stress on plant growth. Models with more complex functions for the effects of environmental stress on plant growth, such as ALMANAC and EPIC, incorporate more stress effects; temperature stress, N and P nutrient stress, salinity, low pH, aluminum toxicity, and soil aeration.

There are several different data types required to parameterize and run each model (Williams et al. 1989; Kiniry et al. 1992; Kiniry et al. 1996; Arnold et al. 1998; Del Grosso et al. 2005; Di Vittorio et al. 2010; Miguez et al. 2011). First, each model requires plant parameters that characterize the developmental stage and rate of biomass accumulation over time. Next, each model requires daily weather values that include daily maximum temperature, minimum temperature, precipitation, and solar radiation. Third, each model requires basic soil properties such as: soil type, nitrogen, texture, moisture availability, or water holding capacity. Lastly, most models incorporate basic crop management practices such as fertilizer application, planting date, harvesting dates, and removal rates.