Abstract: Based on the late rice flood disaster risk assessment model, this article evaluates and predicts the late rice flood disaster risk degree in its production and growth period, as well as its temporal and spatial differences.  The distribution of the late rice flood disaster risk degree decreases generally from southeast to northwest in South China, that is, the highest in the southeast coast region, moderate in the central part, and the lowest in the northern Guangxi Province. The risk degree in southeast coast of South China area is 1.4-2 times of that in north of Guangxi Province.  From the changes of the flood disaster risk at different growth periods, influenced by the increase in the intensity of tropical cyclone precipitation during last flood season in South China area, the effects of precipitationinduced disaster are increasing during the period from sowing to jointing stage; but the effects of precipitation on the period from jointing to mature stage are contrary to that on the period from sowing to jointing stage. In recent 30 years, the flood disaster risk degree of late rice has a gradually decreasing trend in South China overall. Meanwhile, abrupt changes of the risk degree appeared obviously at the end of the 1990s, and thereafter, the risk degree decreases most significantly at a rate of 0.277 10 a^-1.  The change rate of the late rice flood disaster risk degree is significantly regional and transitional. There is an increasing trend in southeast coast of South China, especially in the Pearl River Delta Plain and Chaoshan Plain, and the increasing rate is more than 0.10 10 a^-1, while there is a decreasing trend in the inland mountain of South China where the descending rate is more than 0.05 10 a^-1. The main reason for such changes is the topographic transitional belt. Therefore, the key periods of the late rice flood disaster risk in South China are from sowing to jointing stage, and the most attentions should be paid on the southeast coast region.

Key words: Pinus massoniana plantation., edge effect, recalcitrant component, forest gap