If cellulosic refineries require large-scale switchgrass production, management recommendations to farmers will be aimed at optimizing yearly yields while minimizing input costs. Many empirical studies have focused on the management practices required to enhance yields. However, management is often regionally associated with climate, rainfall, and soil nutrients. Field trials are expensive and time consuming making it virtually impossible for data to be collected for all soil types and climate conditions. Instead, mechanistic models can be used to assess many different management practices in a timely manner to make local recommendations.

ALMANAC, DAYCENT, and EPIC models have been used to simulate the effects of irrigation, fertilization, and harvest frequency on switchgrass yields (Kiniry et al. 1996; Brown et al. 2000; Kiniry et al. 2008b; Thomson et al. 2009; Davis et al. 2011; Lee et al. 2011). The ALMANAC model has been used to determine the optimum management for several locations in the Northern Great Plains and Texas. For the Northern Great Plains, the response of yields to supplemental irrigation and N fertilizer was analyzed. Irrigation does not seem profitable in the Northern Great Plains in ND and SD because supplemental irrigation resulted in at best a small (< 1.5 Mg/ ha) increase (Table 4). Increasing the amount of N fertilizer substantially increased yields by 70% and 48% in NE and SD but had no impact in ND. In Texas, simulating a one — versus two-cut harvest revealed that two-cut does not significantly increase yields.

Other models have revealed similar results. Davis et al. (2011) used the DAYCENT model to estimate the yields of fertilized and unfertilized switchgrass. They determined that yields increased with N addition and consequently C emissions were reduced. Brown at al. (2000) and Lee et al. (2011) similarly found a positive relationship between yield and N fertilizer using the EPIC and DAYCENT models, respectively. Additionally, Lee et al.

(2011) reported higher yields with two or three cuts as opposed to one cut per year. Thomson et al. (2009) only reported increased yields with two-cut for upland ecotypes but not lowland ecotypes.