Field-scale DNAPL dissolution is controlled by the topology of DNAPL distributions with respect to the velocity field. A high resolution percolation model was developed and employed to simulate the distribution of DNAPL within source zones. Statistically anisotropic permeability values and capillary parameters were generated for 10${\times}$10${\times}$10 m domains at a resolution of 0.05 to 0.1 m for various statistical properties. TCE leakage was simulated at various rates and the distribution of residual DNAPL in 'fingers' and 'lenses' was computed. Variations in finger and lens geometries, frequencies, average DNAPL saturations, and overall source topology were predicted to be strongly influenced by statistical properties of the medium as well as by injection rate and fluid properties. Model results were found to be consistent with observations from controlled DNAPL release experiments reported in the literature. The computed distributions of aquifer properties and DNAPL were utilized to perform high-resolution numerical simulations of groundwater flow and dissolved transport. Simulations were performed to assess the effect of grout or foam injection in bore holes within the source zone and of shallow point-releases of fluids with various properties on dissolution in DNAPL dissolution rate, even for widely spaced injection points. The results indicate that measures that induced partial flow reductions through DNAPL source zones can significantly decrease dissolution rates from residual DNAPL. The benefit from induced partial flow reductions is two-fold: 1) local flow reduction in DNAPL contaminated zones reduces mass transfer rates, and 2) contaminant flux reductions occur due to the decrease in groundwater velocity