Plant, Soil and Environment

J. Yang ¹, S. Shi ¹, W. Gong ¹, L. Du ¹, Y. Y. Ma ¹, B. Zhu ¹, S. L. Song ²

Affiliations
1 State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan, CN
2 Wuhan Institute of Physics and Mathematics, Chinese Academy of Sciences, Wuhan, CN

Abstract

Paddy rice is important for Chinese agriculture and crop production, which largely depends on the leaf nitrogen (N) levels. The purpose of this study is to discuss the relationship between the fluorescence parameters and leaf N content of paddy rice and to test their performance in inversing N content of crops through back-propagation (B-P) neural network. In the correlative analysis of the fluorescence parameters and the N content, we found that the correlation between fluorescence ratios (F740/F685 and F685/F525 (F740, F685, F525 - intensity of fluorescence at 740, 685 and 525 nm, respectively)) and the N content (R² are 0.735 and 0.4342, respectively) is weaker than that between the intensity of fluorescence peaks (F685 and F740) and N content (R² are 0.9743 and 0.9686, respectively). Our studies show that the accuracy and precision of N content inversion which is acquired from the intensity of fluorescence peaks through the B-P neural network model are significantly improved (root mean square error (MSRE) = 0.1702, the residual changes between -0.1-0.1 mg/g) compared with the fluorescence ratio (MSRE = 0.3655, the residual changes from -0.3-0.3 mg/g). Results demonstrate that the intensity of fluorescence peaks can be as a characteristic parameter to estimate N content of crops leaf. The B-P neural network model will be serviceable approach in inversing N content of paddy leaf.

Keywords

Remote Sensing, N Fertilization Strategies, Nutrient Stress, Chlorophyll.

Full Text