Abstract: With the adjustment of agricultural structure, the planting area and distribution of crops changed greatly in terms of space and time, which introduced much influence on environment, ecology and decision-making by government. When land use of Beijing and nearby region was investigated and updated through remote sensing, traditional method of extracting winter wheat produced inaccurate results because of mixing with clover whose planting area increased much just in recent years. The focus was put on monitoring crops in spring-summer and extracting their areas in this research. The spectra of winter wheat and clover were measured for four times from April to June, and it was found that there was little difference between winter wheat and clover during early period, but the difference increased gradually during late period. The great difference of spectra was located in near-infrared(NIR) region. It is the best time to classify wheat and clover during late May and early June. However clover in different lands was cut by farmers in different time from May to September. It was impossible to classify wheat and clover in one scene remote sensing image in May and June. The Landsat NIR bands of three days were used to compose a color map and it is easy to not only classify wheat and clover but also infer the cutting time of clover. Tested by ground data, classification results of this method were accurate. This method just used a NIR bands of Landsat, which improved the speed of data processing and reduced the cost, more importantly this method was less influenced by haze.