Abstract: Based on air temperature data from 90 metrological stations in winter from 1957 to 2010 in the northeast China, the study area was divided into the south and north subregions in terms of winter air temperature using methods of a rotated empirical orthogonal function (REOF) and a cluster analysis. The variation trends of winter air temperature and warm and cold winter were analyzed, and their relationships with main circulation indexes were discussed. The optimum forecast factor was selected by a back-method, and regression models were built in the whole study area and two subregions. The results indicate that winter air temperature increases obviously in the northeast China, and the ratio reaches 0. 45 ℃/ 10 a, especially in the north area. The correlation between winter air temperature and the simultaneous Euro-Asian zonal circulation indexes is significant. The subtropical high area index in August in the east Pacific Ocean, polar vortex area index in October in the Asia area and polar vortex area index in August in the north hemisphere are in the significantly positive correlations with winter air temperature, and their multiple correlation coefficients all reach 0.70. The above three factors are the key forecast factors in regression equations. If the subtropical high and polar area indexes are used as the important influencing factors when it forecast cold winter and warm winter, the forecast effect is good.

Key words: Northeast China, Winter air temperature, Rotated empirical orthogonal function (REOF), Circulation index, Correlation, Forecast