GM( 1,n) is established by the Spotted mackerel( Scomber australasicus) resources of the Pacific group which is supplied by Japan's central fisheries research institute from 1995 to 2012,combined with the sea surface temperature( SST) of spawning and feeding ground and tidal range. There are six models: GM( 1,1 ) without environmental factor; four GM( 1,2) models established by the SSTs of the feeding ground( 140°E - 160° E,35° N - 50° N,SST1 ), spawning ground1( 130° E - 132° E,30° N - 32° N,SST2 ), spawning ground2( 138°E - 141°E,34°N - 35°N,SST3) and tidal range of Kuroshio; GM( 1,5) established by all the factors. The average error of each model is 6.72%,3.73%,4.41%,4.78%,29.56% and 19.38% respectively. The results show that the GM models based on feeding and spawning grounds temperature have a high accuracy in the spotted mackerel resource forecast,and can be used into more researches. By analyzing the gray parameters a and b,it is concluded that the SST2 and SST3 are the most restrictive to the model,which means the fluctuation of those factors had the greatest effect on the amount, and the temperature of spawning ground has a high correlation with the resources. When compared with the optimum temperature,it is presumed that the mackerel resources in the appropriate temperature range can increase with the temperature of the spawning ground. The global warming and gradual increase of SST are likely to have a positive effect on the spotted mackerel stock.