Climate change is a major threat to species, unless their populations are able to invade and colonise new landscapes of more suitable environment. In this paper, we propose a new model of the invasion process using a tool of landscape network sparsification to efficiently estimate a duration of the process. More specifically, we aim to simplify the structure of large landscapes using the concept of sparsification in order to substantially decrease the time required to compute a good estimate of the invasion time in these landscapes. For this purpose, two different simulation methods have been compared: full and R-local simulations, which are based on the concept of dense and sparse networks, respectively. These two methods are applied to real heterogeneous landscapes in the United Kingdom to compute the total estimated time to invade landscapes. We examine how the duration of the invasion process is affected by different factors, such as dispersal coefficient, landscape quality and landscape size. Extensive evaluations have been carried out, showing that the R-local method approximates the duration of the invasion process to high accuracy using a substantially reduced computation time.