Based on the monthly mean air temperature data at 26 stations in Northeast China, the 74 circulation characteristic indexes from the National Climate Center, the NCEP/NCAR reanalysis data from 1961 to 2015, and the rice yield per unit data from 1986 to 2015 in Liaoning Province, the temporal variation of rice harvest was analyzed and the annual harvest forecast model of rice in Liaoning was established by using a multiple linear regression method. The results are as follows:(1) The actual yield, trend yield and meteorological yield of rice in Liaoning Province showed an increasing trend, and the time for their climatic mutations was 1992, 1997, and 1994, respectively. The relationship between actual output and meteorological yield was closer; (2) Through the F-test, the rice harvest forecast model reached the statistical significance. The basic accuracy rate of annual forecast model for rice was 81.9%. And the predicted yields of rice in 2014 and 2015 by using this model were close to the actual values.

%U http://www.ghqx.org.cn/EN/10.11755/j.issn.1006-7639（2018）-03-0501