The encounter desorption (ED) approach significantly deviates from the more rigorous microscopic Monte Carlo method when the surface H2 coverage is high. We propose an improved method that removes this limitation under high H2 coverage. This method is applied to explain the coverage-dependent H2 desorption energy observed in experiments, EexptD2(θ). Hydrogen molecules adsorbed on bare surfaces and those adsorbed on H2 substrates are distinguished and treated as different species, each characterized by distinct diffusion barriers and desorption energies. We find that the effective desorption energy calculated using the improved method shows much better agreement with EexptD2(θ) than that obtained with the conventional ED approach. Moreover, the improved method clearly outperforms the original approach under conditions of high surface H2 coverage.