The statistical analysis of meteorological data processing is dataintensive and always conducts on a single machine. The speed is too slow when the data set is large, which restrains researches of meteorological data. Hadoop and MapReduce are used to speed up the dataintensive processing of meteorological data. In allusion to the low efficiency of processing enormous and small files by using Hadoop, a preprocess is conducted to integrate the enormous and small files to a large one. The experiment proved that this method can solve the low efficiency problem when using Hadoop to process enormous and small files. Comparing with Oracle, it is more useful to use Hadoop to process dataintensive Meteorological computing.