Prediction of chlorophyll content using spectral response characteristics of greenhouse tomato

In order to estimate chlorophyll content of greenhouse tomato leaves fast and accurately and improve the precision management of tomato crop, this research studied the correlation of the chlorophyll content and spectral response at different growth stages of greenhouse tomato. Leaf spectral measurements from each treatment (4 N levels: 0%, 33.3%, 66.6%, 100%) were taken in the greenhouse using ASD FieldSpec HH. Chlorophyll content of tomato leaves were measured by alcoholic-acetone extraction in chemistry lab. It was found that Chlorophyll content of tomato leaf was increasing continuously to the maximum 50 days after the transplantation, while red edge moved to direction of NIR(long wave), green peak position moved to direction of blue light(short wave) and green peak amplitude decreased. The chlorophyll content would decrease after fruiting stage, while red edge, green peak position and its amplitude moved to the opposite direction. For quantitative analysis the relationship between chlorophyll content and spectral response, red edge parameters (Sred(area of red edge), Dred (amplitude of red edge) and Pred (position of red edge ) ) in the first derivative reflectance curve were obtained at bands of 680 nm to 760 nm. Similarly, blue edge, green peak and red valley parameters were defined to reflect spectral character. Vegetation indices have been used extensively to estimate the vegetation growth status. So the following wavelength were used for developing RVI,NDVI and ARVI index: l440 nm, l500 nm, l550 nm, l680 nm, l770 nm, Pblue (position of blue edge), Pyellow (position of yellow edge), Pred (position of red edge), Pgreenpeak (position of green peak), Predvalley (position of red valley). Seven optimal spectral characteristics parameters were chosen with the method of Karhunen-Loeve from the above-mentioned 68 self-defined feature parameters. At last, stepwise multiple regression (SMLR), principal component regression (PCR), ridge regression (RR) and partial least squares regression (PLSR) were used to develop the prediction models of the chlorophyll content of tomato leaf. The best model was obtained by RR. Root MSE was 0.406 and R-Square was 0.839.