Based on the K-Means spatial clustering algorithm, this paper introduces the features of food emergency reserves and influencing factors according to the food emergency reserve goals, and uses the distance of the gravity center of the elements as the test standard to achieve the zoning of provincial emergency food reserves and formulate corresponding food emergency reserve strategy. The results show that, by fully taking disasters, population, economy and other factors into consideration, the provincial grain emergency reserve zoning model, based on the K-Means spatial clustering algorithm, can achieve deliver better results. For the purpose of fine division, a certain degree of cost savings and efficiency are achieved. This paper provides a theoretical reference for the study of the regional emergency reserve of grain in China's provincial regions, and guides practices.