Using Nelder-Mead Optimization Routines with the USDA Soil Physics Model CHAIN_2D to Determine BMPs for Chemigation Practices.

Investigations of volatilization and subsequent air dispersion of agrochemical products has generally been limited to extremely volatile compounds such as soil fumigants. An effective way to minimize off-site fumigant transport is to reduce the amount of fumigant applied, albeit at the possible expense of efficacy against the pest being targeted. This work couples criteria using the USDA soil physics model CHAIN_2D, a Nelder-Mead numerical optimization procedure, and various pest organism mortality dose-response observations (based upon published values) to address environmental and efficacy issues for the fumigant 1-3,dichloropropene (1,3-D). Computer experiments are performed for a variety of common agricultural soils (sand, loamy sand, sandy loam, silty loam, loam) for chemigation performed using surface drip lines under a polymer membrane (tarp), or drip lines at a depth of 7.5 and 15.2 cm in soil. The objective function requiring minimization was a coupling between maximizing the fumigant efficacy in soil while minimizing volatility losses. Chemigation parameters include pulsed irrigation amounts and durations, along with the timing between successive irrigation pulses. Best management practices are functions of the soil type, organism dose-response function, and the endpoint being considered such as pest efficacy levels and the area in soil where efficacy is achieved. Endpoints include the cumulative mass loss volatilized from soil, the area within soil root zone where efficacious 1,3-D residues reside (as measured by a concentration × time index), and a combination of the two. Sensitivity analysis suggests that as the initial soil moisture increases, efficacy increases and volatility decreases. ANOVA analysis confirms the primary sensitivity of soil texture and initial soil moisture constant. Other sensitive parameters are associated with the chemigation pulse duration, the number of pulses, and the time between each consecutive chemigation pulse. R2 values for model predictions are generally > 0.85 indicating the empirical multivariable equation can serve as a surrogate to quantify the impact of Best Management Practices (BMPs) associated with chemigation.