Abstract: Nicotine and total nitrogen contents directly influence the accumulation of aroma precursors in tobacco leaves through nitrogen metabolism, and their dynamic monitoring during the field growing period plays a critical role in estimating tobacco yield and quality as well as enabling precision field management. However, current UAV-based remote sensing methods for estimating nicotine and total nitrogen contents are primarily focused on the canopy scale, making it difficult to capture the vertical heterogeneity characteristics of leaves at different stalk positions. This study proposes a method for simultaneously estimating nicotine and total nitrogen contents in upper and middle leaves based on UAV hyperspectral imaging. Through field experiments with different fertilization gradients, a three-step collaborative modeling framework was constructed that couples canopy spectral signals with the spatial distribution characteristics of nicotine and total nitrogen. First, machine learning algorithms were used to develop quantitative estimation models relating canopy spectral reflectance—dominated by the contribution of upper leaves—to nicotine and total nitrogen contents in upper leaves. Second, statistical regression analysis was employed to elucidate the relationships between the corresponding indicators in upper and middle leaves. Finally, using the estimated values of upper leaves as inputs, the relational models were coupled to indirectly invert nicotine and total nitrogen contents in middle leaves. The results demonstrate that the proposed method effectively captures the vertical distribution variation characteristics of nicotine and total nitrogen. The estimation accuracy for upper-leaf nicotine was R²=0.744, RMSE=0.34%, and for total nitrogen was R²=0.778, RMSE=0.43%. The indirect estimation accuracies for middle-leaf nicotine and total nitrogen were R²=0.715, RMSE=0.23% and R²=0.754, RMSE=0.30%, respectively, both outperforming direct spectral estimation of middle leaves. This study provides a non-destructive and scalable remote sensing monitoring solution for precision tobacco field management, with potential for extension to other multi-layered economic crops.