Impact of Apex Model Parameterization Strategy on Estimated Benefit of Conservation Practices.

Three parameterized Agriculture Policy Environmental eXtender (APEX) models for corn-soybean rotation on clay pan soils were developed with the objectives, 1. Evaluate model performance of three parameterization strategies on a validation watershed; and 2. Compare predictions of water quality benefits (WQB) from best management practices (BMPs; terraces, grass waterways and winter cover crops) for the validation watershed by the three parameterizations. Two of the three parameterizations were from previous studies on a 4.44-ha, no-till corn-soybean watershed in Novelty, Missouri located 90-km northwest of the validation watershed on similar soils.  The first parameterization (pre-buffer) used event-data from 1993-1997 (n=43); the second parameterization (post-buffer) used event-data from 1998-2008 (n=42) after the establishment of 4-m agro-forestry buffers. The third parameterization (local) was developed using event-data from 2012-2013 (n=10) from a 2.69-ha field-cultivated corn-soybean watershed in Chariton, MO. The validation data were from an adjoining 31.7-ha field-cultivated watershed with terraces, a grass waterway, and winter cover-crops. The validation event-data were from 2011-2013 (n=15).  Event-runoff was well simulated with all three parameterizations with coefficient of variation (r²): 0.7-0.8, Nash-Sutcliffe Coefficient (NSC): 0.3-0.6, and percent bias (Pbias): -20-6%.  Model performance of the pre-buffer and local parameterizations was acceptable for event-sediment (r²: ~0. 5, NSC: 0.4-0.5 and Pbias: 23-53%) and event-total phosphorus (TP) losses (r²: 0.8-0.9, NSC: 0.3-0.5 and Pbias: 66-70%).  Thirty-year average annual predictions by the three models for individual and combinations of BMPs varied by 1-15% for runoff, 2- 51% for sediment, and 1-24% for TP, the largest differences being obtained when the terraces were present. In spite of these differences, the BMPs effectiveness (reduction in runoff, sediment and TP) estimated by the three parameterizations remained within 12% of each other. All parameterizations predicted terraces were the most effective for reducing sediment (43-51%) and TP (52-54%). Cover crops were the most effective for runoff reductions (18-22%).  All three parametrizations showed reasonable robustness in quantifying the comparative WQB of BMPs.