Abstract: Tomato 'Kaisa' was used as test material for pot trials in the artificial climate chamber of Nanjing University of Information Science and Technology from June to August 2022. Using whiteflies as media, through stinging and sucking for virus transmission to tomato plants. The maximum/ minimum air temperature was set at 28℃/18℃ and the relative air humidity at 70% (±5 percentage points) in the artificial climate chamber. The spectral reflectance and relative chlorophyll content (SPAD value) of tomato seedling leaves at different levels of the infection were measured by the SOC710 spectrometer and the SPAD-502 chlorophyll meter. The original spectra was first-order derivative transformed to analyze the relationship between each of the hyperspectral parameters and SPAD value of tomato leaves at different infection levels. In turn, the sensitive bands and characteristic parameters of tomato were screened in order to construct the best model for the estimation of the SPAD value. The results show that different infection levels of tomato leaves had different positions of the red edges. The leftward shift in the position of the spectral curve peak occurred at 710nm, with the maximum peak wavelengths occurring at 711nm, 700 nm, 695 nm, and 690 nm for healthy, slight, moderate, and severe infested tomato leaves, respectively. The elements R₅₃₁, R₈₁₆ (positive), R'₅₀₀, and R'₇₂₁ (positive) with the maximum highly significant correlation (P<0.01) were selected as spectral feature parameters by correlation analysis of spectral parameters with SPAD value. Spectral indexes were constructed using spectral feature parameters and the elements that were highly significantly correlated with SPAD values were Dy, λb, λr, Rg, RVI, DVI, PRI, GNDVI, NDVI, mNDI and SAVI. The hyperspectral estimation model of SPAD value of tomato leaves under virus disease infection was constructed by screening mNDI, GNDVI, R₈₁₆, Rg, and R'₅₀₀ by stepwise regression method, and the R² of the validated model was 0.835. This study showed that the use of spectral technology was feasible for the measurement of SPAD value in tomato leaves affected by virus diseases, which provided a reference for early non-destructive diagnosis of plant diseases.

Key words: Tomato, Virus disease, SPAD value, Hyperspectral, Estimation model