Abstract
In this research, the partial least square regression models for ground truths of first and second crops of green tea were estimated using a ground-based hyperspectral camera. The regression coefficients of the models were compared and analyzed. As the green tea plant grows, factors relating to yield increase, such as the number of new shoots, dry weight, dry weight of the 100 new shoots, and nitrogen content. However, factors governing quality, such as nitrogen concentration, decrease. Since factors controlling the yield are variable, the accuracy of the models for each crop vegetation stage was better than that for the combined data of first and second crops. The accuracy of the nitrogen concentration model, which was less variable at each crop vegetation stage, increased (r=0.759 and RMSEP=0.270) by using the combined data of first and second crops. The nitrogen concentration of both data of first and second crops had been more influence over nitrogen contents comparing that of each crop.
