Model of moisture content of paddy rice leaf based on canopy spectral reflectance

Based on the relationship between paddy rice canopy spectral reflectance and the leaf moisture content, a model of the leaf moisture content was built. Both the canopy spectral reflectance and the moisture content of leaf in booting stage were measured. According to the correlation coefficient of the paddy rice leaf moisture content and the spectral reflectance, the characteristic wave bands which had the higher correlation coefficient were selected. The genetic algorithm was used to optimize BP neural network’s initial weights. The prediction models were built using BP neural network, the GA-BP-Network and the traditional multiple linear regression method. The test results showed that the average error rate of the predicted moisture content value and the real value was 3.9% with GA-BP-Network model and the largest error rate was 6.1%. The prediction capability of the GA-BP-Network is better than that of BP neural network and the multiple linear regression, and the model can improve the accuracy of the moisture content prediction.