Multispectral Imaging Techniques for Nutritional Monitoring of Schima superba and Their Applications

Schima superba is a valuable timber species of great importance in southern China, with extensive plantation areas. Rapidly evaluating the growth status of S. superba forest stands using multispectral imaging technology (MSI) holds significant practical importance for contemporary forest management practices. In this study, we acquired multispectral images via UAV remote sensing from a 4-year-old S. superba forest in Badu, Longquan, in 2023. Based on original spectral bands and 39 vegetation indices, random forest (RF), support vector machine (SVM) and partial least squares regression (PLSR) models were constructed to predict leaf nitrogen (N), phosphorus (P), potassium (K), carbon (C), chlorophyll (Chl) and anthocyanin (Anth). The results showed that the RF model was superior for predicting the 6 physiological traits ( R t r a i n 2 = 0.88 ∼ 0.91, R t e s t 2 = 0.45 ∼ 0.83). The vegetation index RDVI showed strong predictive ability for all 6 physiological traits. Meanwhile, SR, NDVI, SAVI and WDRVI indices were effective in predicting N, P, Chl, and Anth. Based on the RF model, we found that high N, Chl, and Anth content along with low P and K in the canopy were associated with vigorous tree growth. Specifically, the canopy N, Chl, and Anth content peaked in November, indicating vigorous growth. Nevertheless, the growth of S. superba is significantly influenced by the environmental conditions of the forestland. Consequently, phosphorus-efficient genotypes were suggested and management practices were recommended for S. superba forest in southern China.