Advancing dairy farm simulations: A 2-step approach for tailored lactation curve estimation and its systemic impacts.

Lactation curve models are a foundational component of dairy farm simulation models because they support prediction of individual animal milk production over time. For farm simulation models to be applicable as decision-support tools, the predicted baseline milk production should match farm reported production as accurately as possible. However, individual animal lactation curve parameters are not easily accessible farm data. This study introduces a straightforward and effective calibration method for determining the parameters of Wood's lactation curve, leveraging a previously published database of parameters and 3 additional data inputs readily available on farms: annual herd milk production (AHMP), number of milking cows, and herd parity composition. Our method involves (1) adjusting curve parameters based on previously reported national estimates and farm contextual metadata (i.e., temporal, geographic, and management factors) and (2) further optimizing the scale parameter for each parity to match the 305-d milk yield derived from observed AHMP, number of milking cows, and herd parity composition. The Ruminant Farm Systems (RuFaS) model, a comprehensive dairy farm simulation platform, was employed to evaluate this method on 10 commercial Holstein dairy farms in New York (n = 3), Texas (n = 3), and Wisconsin (n = 4). By using lactation curve parameters estimated by this calibration method, we achieved greater accuracy in estimating AHMP with RuFaS for these case study farms, reducing the root mean square percentage error from 40.6% to 2.22%. To evaluate the downstream effects of lactation curve estimation methods within a farm systems context, we used the RuFaS Animal Module to simulate key performance and environmental metrics, including dry matter intake, enteric methane production, and manure excretion. We calculated gross feed efficiency and associated feed and manure emissions using emission factors derived from established literature. The results underscore the critical role of lactation curve modeling in dairy farm system simulation models and its substantial effects on environmental footprint predictions. In conclusion, this study demonstrates the effectiveness of our lactation curve parameter calibration method. With this method, farm system simulation models such as RuFaS can greatly increase their reliability in generating farm-specific predictions without requiring extensive farm data collection.