Abstract: Daily precipitation data from 348 stations in northern China are selected to construct the time series of the periods of the longest consecutive dry days in the warm half year (April-September) that is divided into four sub-regions. The empirical mode decomposition (EMD) method is used to obtain intrinsic mode functions (IMFs) by which variations are analyzed at various time scales. Based on wavelet analysis and other methods, sudden jump points are found. Results show that the variation of the longest consecutive dry period is mainly composed of two higher frequency intrinsic mode functions, and their contribution ratio is 80% to the total variance. This shows that 3-4 year and 7-8 year oscillations play important roles. Lower frequency components reflect the inter-annual variation of about 10 years. There is an increasing trend of the longest consecutive dry period in the warm half year in northeast China, north China, and eastern part of northwest China, and the droughts in most of northern China are increasing, but decreasing in the western part of northwest China. For the last 10 years, it get drier in northeast China, while it gets wetter in the eastern part of northwest China. The significant increase of the longest consecutive dry periods observed in north China in 1992 and in northwest China in 1960 was associated with the droughts. The results obtained may be different from each other owing to the differences in temporal and spatial scales.

Key words: empirical mode decomposition, northern China, temporal and spatial variations, the longest consecutive dry period, warm half year