Stem water content is an important parameter for evaluating plant physiological water conditions. Poplar trees were selected as research objects. The estimation model of stem water content was proposed by analyzing the variation relationship between stem water content and micro-environment parameter set. Considering the multi-collinearity between micro-environment parameter set, the maximum principal component PC1 of micro-environment parameter set was chosen as a feature variable via principal component analysis. The feature variable retained 98.89% of the original data information. In addition, the complexity of model was simplified by reducing the data dimension. PC1 as the input variable and stem water content as the output variable, the oblique ellipse model between the two was established. In six sunny days with similar micro-environment, the mean error of the model was less than 0.05%, root mean square error of the model was less than 0.06%, and decision coefficient of the model was greater than 0.94. The oblique ellipse model can precisely estimate real-time stem water content, but because of the differences in morphological indexes of poplar trees, the estimated parameters of different oblique ellipse models between stem water content and PC1 were different. Moreover, the estimation model of stem water content should be respectively established according to different seasons and weather environment.