Abstract
Color is a key indicator of quality and roast level in coffee beans. Surprisingly, little is known about the effect of different "roast profiles," i.e., the temperature versus time inside the roaster, on the dynamics of color during roasting, with most prior work focused on laboratory-scale roasters with little control over the roast profile. Here, we investigate seven roast profiles of the same total duration but varying dynamics inside a 5 kg commercial drum roaster, using coffee from three origins. We show that despite the dramatic differences in roast profiles and coffee origins, the bean color always maps onto a "universal roasted coffee color curve" when plotted in the L*a*b* color space. This universal color curve was modeled using polynomial mixed-effects regression and validated through a systematic review of existing literature following the PRISMA protocol to demonstrate its broad applicability. Although the dynamics of roast color development varied with roast profiles, the coffees always had approximately the same L*a*b* values at significant roast milestones, including color change, first crack, and second crack. We discuss how these results provide insight into color measurements and how they can quantitatively inform roast-level standards in the coffee industry for both real-time and post-roast applications.