The sequestration of carbon dioxide in subsurface aquifers is an interesting process that poses fundamental questions concerning hydrodynamic stability, optimization methods, and perturbation methods. Following the injection of CO2 into an aquifer, the CO2 slowly dissolves into underlying groundwater. Because the density of the groundwater increases with the dissolution of CO2, this produces an unstable density gradient in which layers of heavy CO2-rich water overly layers of lighter water with low CO2 concentration. This leads to overturning and mixing of the groundwater that play a dominant role in the long term transport and dissolution of CO2. Unfortunately, though studied extensively, there is wide disagreement concerning the onset of this mixing. This presentation will demonstrate that previous disagreement stems from a sensitivity to how flow instabilities are measured. Furthermore, due to unique physical constraints, traditional hydrodynamic stability methods do not predict realistic results. To address these issues, we develop novel optimization and weakly nonlinear methods to predict the time required for onset of mixing as well as the flow structures associated with this mixing. All results are further validated by comparison to direct numerical simulations using high-order spectral methods.