Data fusion is one of the key techniques in Wireless sensor network (WSN). Data fusion can reduce quantity of data transmission in the network, extend network lifetime by reducing energy consumption, and improve efficiency of bandwidth utilization. At present, researches on data fusion in WSN mainly fall into the following aspects: aggregation tree construction and data processing. This paper proposes an aggregation tree construction method based on grid named ATCBG which makes some improvements over GROUP. The results of simulation experiment show that the average energy consumption of ATCBG is evidently lower than GROUP, and the lifetime of the network is much longer than GROUP before the emergence of node death.