Abstract: Based on the 1963-2008 meteorological data from 68 weather stations in Fujian Province, and by using mathematic statistics and GIS technique, the short-term subtle forecast and early warming methods for cold (freezing) damage of fruit trees in the province were approached. The forecast period for cold (freezing) damage of fruit trees was from early December to next early February, and the key period for warning low temperature was from mid December to next mid January. The short-term forecast equations for the minimum daytime temperature in Fuzhou, Xiamen, and Shaowu were established by stepwise regression analysis, which could be used for the forecast of short-term minimum temperature by all of the meteorological observation stations in Fujian Province by means of differential algorithms. Based on the data of longitude, latitude, and altitude, the geographical relational prediction model of the minimum daytime temperature was established, and the distribution map of forecasted low temperature was drawn with GIS technology, which could subtly forecast the minimum daytime temperature of whole Fujian Province. In combining with the cold (freezing) damage indices of fruit trees such as litchi, longan, and banana in southern subtropical monsoon climate zone, the forecast information of warning cold (freezing) damage of fruit trees were popularized, and the occurrence, development, and range of cold (freezing) damage for fruit trees were forecasted in short-term. By means of differential algorithms, the short-term minimum temperature was forecasted by all weather stations in the Province in 2009, and the forecast accuracy of ≤1 ℃, ≤1.5 ℃, and ≤2 ℃was 58.3%, 83.3%, and 91.7%, respectively. It could be seen that our short-term forecast model possessed a certain forecast capability, and could be used for the quantificational forecasting of low temperature in winter.

Key words: Landscape, Spatial dynamic model, Remote sensing, GIS, Model test