Electric vehicles (EVs) are vital to decarbonization, yet few EV adoption models operate at the fine spatial and demographic scale required for understanding household-level adoption. This model is one of the first attempts to employ a synthetic population to examine EV distribution at this level. We apply latent class clusters of first-time adopters, developed in earlier work, to a synthetic population at the Census-Tract-level which is enriched with household vehicle attributes. Then, we apply Bass diffusion models to each cluster. The result is an EV adoption model that can examine minority EV adopters at granular spatial scale. Spatial microsimulation is used to develop a representative population of California. This technique relies on two datasets, a constraints table of aggregate variable estimates and a sample table of representative households. The former is available at the Census-Tract-level, which allows our analysis to be local. The synthetic population includes variables like income, household size, number of vehicles, and home type from the ACS. Additional variables are imputed from the California Vehicle Survey including household fleet vehicle body type & age. Latent clusters developed in previous work are used to score the synthetic population and bass diffusion curves are fit to each cluster. Probabilities of belonging to one of eight clusters of first-time EV adopters is calculated for all households in the synthetic population. In earlier work, Ramadoss et. al. classify first-time EV adopters into eight latent clusters: four single-vehicle and four multi-vehicle household clusters. While high-income, older suburban homeowners comprise most EV buyers, many of these clusters are young, lower-income, renter, or urban EV buyers. Finally, Bass diffusion curves are fit to EV adoption in each cluster. The result is a forecast of EV adoption by year and region, at the local level considering adoption by groups that have been underrepresented to-date.