Forests cover nearly 40% of low-elevation mountains worldwide and are often considered a nature-based solution against granular geophysical flows [1]. Simulating the complex interactions between flow and forest in a high-fidelity way is a challenging task. This is because numerical simulators struggle to model the contact behaviours between flows and trees and the effects of tree uprooting. Due to these difficulties, existing numerical simulators are difficult to conduct high-fidelity simulations of flow-forest interactions. To address these issues, this study presents a novel numerical strategy for simulating complex flow-forest interactions, including the uprooting of trees. The simulator is based on coupled material point method and level-set discrete element method (MPM-LSDEM), which incorporates the GJK-EPA scheme [2] for contact detection and a sampling point-level-set strategy to build a realistic contact manifold. Experimental results from both self-developed model forests and literature were used to validate and calibrate the MPM-LSDEM simulator. Our findings demonstrate that the MPM-LSDEM simulator can produce numerical results that closely approximate experimental results. Using the developed simulator, we show that interactions between overlapping bow shocks at the tree scale significantly affect the mobility of dry granular flows moving through a forest. Bow shocks can either reduce flow momentum or concentrate momentum, increasing the flow runout distance by up to 1.5 times compared to bare terrain without trees. Additionally, tree uprooting dissipates flow kinetic energy and reduces flow momentum. These results highlight the potential of the developed simulator as a new guiding tool for assessing the design of cultivated forests in impeding dry granular flows.