Use of neural networks to discriminate between control leaves of wheat or those deficient in nitrogen, phosphorus, potassium, and calcium using spectral data

Rapid identification of deficiencies in major elements using spectral characteristics would be a useful tool in precision farming and in other nutrient intensive agricultural production systems such as those proposed for long term space missions. A Multilayer Perceptron (MLP) neural network and backpropagation algorithm was used to discriminate between control leaves of wheat (Triticum aestivum L.) and those deficient in nitrogen (N), phosphorus (P), potassium (K), and calcium (Ca) using hyperspectral data. The network consisted of three layers (input, hidden, and output) with spectral reflectance of the leaves in wavelengths 401 nm to 770 nm as the input layer and the quantified nutrient concentrations of each element as the output layer. Based upon the values of actual nutrient concentrations (ppm), plants were classified as either deficient or normal. Wheat plants were grown for ∼100 d under both hydroponic conditions in the greenhouse and semi-hydroponic conditions in a growth chamber using Hoagland´s complete nutrient solution with selected elements removed to induce specific nutrient deficiencies. Control plants received complete nutrient solutions. The MLP model was trained and tested successfully within 1000 epochs as the MSE of the sample training curve approached zero. The back propagation algorithm performed well with the following accuracies for the classification model: N 90.9%, P 100%, K 90%, and Ca 100%. Using the regression model, the following accuracies were obtained: N 93.0%, P 87.2%, K 91.9%, and Ca 97.4%. This affirms the potential of using spectral data coupled with either a classification or regression neural network models to identify quickly leaves deficient in these four major elements so that remedial applications of those nutrients can be made before the crop is substantially impacted.