Phenomenological crystal plasticity models are widely used in materials engineering as they strike the right balance between complexity (of the captured physical phenomena) and ease of modeling: they capture the anisotropic nature of plastic slip while having to calibrate only a handful of parameters. Still, identifying a unique set of parameters for these models remains a challenge. This is because, in contrast to physics-based models where the material parameters hold physical meaning, the parameters in phenomenological models are related to ad hoc function used to describe the state of the material. Hence, they typically cannot be obtained from simulations or experiments at small length scales such as molecular dynamics or electron microscopy. However, the way in which phenomenological models address forest hardening through dislocation movement is analogous to a majority of physics-based models. In this study, we investigate how strengthening coefficients that are identified from discrete dislocation dynamics simulations for the use physics-based models can be incorporated into the latent hardening matrix that is used to describe forest hardening in phenomenological models. We compare this against other established parametrizations that have been widely used for decades. Through comparison against reference result obtained from a physics-based model, we conclude that adopting these physics-backed coefficients leads to significant gains in the prediction of hardening behavior. Most importantly, these coefficients have already been identified for most practically-relevant materials and therefore can be directly incorporated into phenomenological models at no additional cost.