FT-NIR Spectroscopy technique based analysis and Prediction on soil nutrient content of Lychee orchard—A case study in Zhongluotan of Guangzhou, South China

摘要

Precision farming with Geoinformatics is a new paradigm for agricultural production, especially in a developing country as China. Precision agriculture is an integrated application referring to the advanced 3S (remote sensing, GIS and GPS) technology and spectroscopy, and the key of this technology is to obtain the spatial and temporal information of the environmental factors such as soil nutrients, which has an influence on the growth of the Lychee crop and agricultural environment protection, and then the appropriate measurement in standard prescription can be adopted to realize "prescription farming" with efficiency for agricultural resource saving and environment protecting. Considering the actual problems and deficiencies of the present technology in the quick information extraction of Lychee soil nutrient, we systematically analyzed the relationships between the intelligent soil Fourier NIR spectrum information and Lychee orchard soil nutrient components such as nitrate (N), total phosphorus (P), potassium (K), organic matter (OM), calcium (Ca), and magnesium (Mg). The content of these components in the 15 in-situ soil samples of Zhongluotan, Guangzhou South China was analyzed against near infrared spectroscopy. The NIR spectral data of soil was treated by partial least square (PLS) analysis, and then the best dimension of analysis was obtained with multivariate linear regression. Upon this, the calibration model was established. The content of some components in Lychee orchard soil was predicted using the calibration model. It is shown that the result of Fourier NIR spectroscopy method was highly related with those of chemical analysis method. The regression coefficients between measured and predicted values of N, OM and pH were 0.93, 0.92, and 0.93, respectively. The correlation coefficient between measured and predicted P, K, Ca, and Mg were satisfactorily acceptable at 0.65, 0.74, 0.61 and 0.69, respectively. It showed that the Fourier NIR spectroscopy method is a good tool for soil nutrient prediction of Lychee garden, especially for soil N, OM and pH. The accuracy of the FT-NIR based PLS models for predict soil nutrient concentration was adequate to be used as a quick evaluation of nutrient composition of Lychee orchard soil for the precision agriculture application.