In order to study the ability of spectrum to estimate the water content of winter wheat plant components and analyze the separation rule of spectral information by wavelet technology, the canopy spectral information of winter wheat and the corresponding measured values of water content of leaves, stems and ears of winter wheat were used as data sources. Then, the partial least squares (PLS) algorithm was used to construct the estimation model of winter wheat plant component water content, which was verified and evaluated. The results showed that after wavelet technology decomposition, the absorption characteristics of winter wheat canopy spectrum were separated into high frequency information by decomposition level, and the absorption characteristics represented by each decomposition level were distributed in the H1~H10 decomposition level of high-frequency information. The accuracy and stability of the estimation model of winter wheat ear water content was strong, that of the stem was the second, and the leaf stability was the worst. This showed that the current situation of water supply of winter wheat after poplar flowering stage was no longer suitable to use only leaf water content for evaluation, and the detection index should be added or replaced.