Diagnosis of nitrogen and magnesium deficiencies based on chlorophyll distribution features of cucumber leaf

Abstract: Nitrogen (N) and magnesium (Mg) elements play important role in the growth of cucumber plants, N and Mg deficiencies in cucumber plants drastically affects the quality and most importantly yield of agricultural products. In the published papers, chlorophyll content was used as an indicator for diagnosing N deficiency and Mg deficiency. However, leaf with low chlorophyll content appears both in N deficient and Mg deficient plans, which makes it is difficult to simultaneously detect N and Mg deficiencies using chlorophyll content. In this study, new indicators based on chlorophyll distribution features of the whole cucumber leaves were proposed for diagnostics of N and Mg deficiencies. N deficient, Mg deficient and control cucumber plants were cultured in a greenhouse with special nutrient supply. The content of N and Mg nutrient elements in N deficient, Mg deficient and control leaves were determined to test the nutrient status of cucumber plants in N deficient, Mg deficient and Control groups. 100 fresh cucumber leaves were collected and used as samples for detecting a chlorophyll distribution map. Firstly, hyperspectral images of cucumber leaves in the calibration set were collected and chlorophyll content of the cucumber leaves was determined using high performance liquid chromatography technology. Chlorophyll content calibration models were built using the hyperspectral images and chlorophyll content. Secondly, the hyperspectral images and chlorophyll content of cucumber samples in testing set were used to test the chlorophyll content calibration models, and the chlorophyll content calibration model with the best performance was selected as the optimal calibration model. The chlorophyll content distribution maps of N deficient, Mg deficient and control cucumber leaves were measured using the optimal chlorophyll content calibration model. After hyperspectral image collecting, hyperspectral image data of N deficient, Mg deficient and control leaves were obtained. Then, the spectral data of every pixel in the hyperspectral images was extracted and substituted in the optimal chlorophyll content calibration model to calculate the chlorophyll content at each pixel. The chlorophyll content of all pixels were displayed in two dimension spastically, then the chlorophyll content distribution maps of N deficient, Mg deficient and control leaves were obtained. The chlorophyll content distribution maps of 25 N deficient cucumber leaves, 25 Mg deficient cucumber leaves and 25 control cucumber leaves were determined. Compared with the distribution map of chlorophyll content in the control leaves, N deficiency led to the decrease of chlorophyll content in the whole leaf, and Mg deficiency led to the decrease of chlorophyll content in the area between the main veins. According to these results, two chlorophyll distribution features, the average and standard deviation of chlorophyll content at every pixels in a chlorophyll distribution map, were extracted for diagnosing N deficiency and Mg deficiency. Result showed that an average of chlorophyll content (11.5 mg/g) could be used as a threshold value to diagnose N deficiency, and the diagnostic rates for the calibration set and prediction set were 100% and 90%, respectively. A standard deviation of chlorophyll content (2.20 mg/g) could be used as a threshold value to diagnose Mg deficiency, and the diagnostic rates for the calibration set and prediction set were 93.3% and 90%, respectively. The result indicated that the extracted features could reflect the characteristic of N and Mg deficient cucumber leaves and could be employed to diagnose N and Mg deficiency nondestructively.