The landscape pattern and its change are important factors affecting surface runoff and spatial distribution and are also an approach of managing urban rainfall flood and governing ecological environment. Taking one million mu of afforestation project in the plain area, Beijing as the background, the remote sensing data during the first round of afforestation time (2011, 2013, 2015, and 2017) were selected in this paper. Defining the Third Ring Road, the Fifth Ring Road and the Sixth Ring Road in Beijing as the boundary, the plain area was divided into core area, central area, suburban area, and outer suburban area for analyzing the changes of landscape pattern of the whole forest land and the partitioned forest land, respectively. SWAT model was used to simulate the impacts of different patterns on storm runoff in different regions, and explore the dominant forestland landscape index affecting runoff change. The results show that 1) the forestland area in plain area presents an increasing trend, with an increase of 12.96%. Apart from some parts in the core area and the central area, the forestland patches are more concentrated and more complex in shape. 2) One million mu of afforestation project plays a positive role in reducing the risk of rainfall flood in the plain area. However, due to the influence of urban construction and human activities, the effecting intensity of afforestation and construction expansion on the runoff mitigation in different areas are slightly different. The effecting intensity of afforestation on runoff mitigation in the core area and central area is greater than the side effect of construction expansion on runoff, which is manifested as the case that the total runoff decreased, with the proportions of 1.98% and 4.29%, respectively. However, the situation is just on the contrary in the suburban area and outer suburban area:the total runoff increased by 0.09% and 6.82%. The intensity of construction expansion in the outer suburbs is the main factor affecting the runoff risk in plain areas. 3) The key forest landscape indices for different regions are different. The dominant forestland factors of runoff in core area are the number factors of patches and the shape factor; that for the central area is area factor; those for the suburban area are aggregation factor and shape factor; and those for the outer suburban area are area factor as well as aggregation factor. The control of dominant forestland factors will reduce the risk of rainfall flood more efficiently. In this study, through SWAT simulation analysis, the spatial changes of forest pattern and runoff characteristics in the plain area and the key forestland factors for runoff mitigation in different areas during the first round of afforestation were obtained which provides reference for improving the effect of mitigating rainfall flood during the second round of afforestation.