Concern is emerging regarding the challenges posed by spatial complexity for modelling and managing the area-wide elimination of parasitic infections. While this has led to calls for applying heterogeneity-based approaches for addressing this complexity, questions related to spatial scale, the discovery of locally-relevant models, and its interaction with options for interrupting parasite transmission remain to be resolved. We used a data-driven modelling framework applied to infection data gathered from different monitoring sites to investigate these questions in the context of understanding the transmission dynamics and efforts to eliminate Simulium neavei- transmitted onchocerciasis, a macroparasitic disease that causes river blindness in Western Uganda and other regions of Africa. We demonstrate that our Bayesian-based data-model assimilation technique is able to discover onchocerciasis models that reflect local transmission conditions reliably. Key management variables such as infection breakpoints and required durations of drug interventions for achieving elimination varied spatially due to site-specific parameter constraining; however, this spatial effect was found to operate at the larger focus level, although intriguingly including vector control overcame this variability. These results show that data-driven modelling based on spatial datasets and model-data fusing methodologies will be critical to identifying both the scale-dependent models and heterogeneity-based options required for supporting the successful elimination of S. neavei-borne onchocerciasis.