Our understanding of human systems has been synthesized and advanced by computationally representing human decision-making in agent-based models. Whether representing individuals, households, firms, or larger organization, agent-based modelling approaches are often used to model processes (e.g., urban growth, agricultural land management) that directly effect and are affected by natural systems. Contemporary efforts coupling models of human and natural systems have demonstrated that results significantly differ from isolated representations of either system. However, coupling models of human and natural systems is conceptually and computationally challenging. In addition to discussing these challenges and approaches to overcoming them, this talk will also suggest that research quantifying natural processes at the decision-making scale of the land user is needed. Using structure-from-motion and unmanned aerial vehicle (UAV) imagery, we can accurately quantify natural processes like soil erosion to a high level of accuracy and that frequently modelled processes (e.g., flow accumulation) typically differ from reality. Novel data from the field or parcel scale are needed to calibrate and validate our representation of natural processes if we are to advance our representation of feedbacks between natural processes and human decision-making. By improving our representation of both natural processes and human decision-making at the scale of the decision-maker, we add confidence in our ability to scale out to larger spatial extents that are reflective of natural processes (e.g., watershed) or policy driving human decisions from municipal, state, or national governments.