Plant stress prediction model on nutrient film technique hydroponic in a natural greenhouse

Abstract

Abstract This study aimed to identify plant stress on the NFT hydroponic module in the greenhouse using the mathematical model of the 4th-order polynomial regression equation. The artificial Neural Network (ANN) approach was also conducted to analyze factors affecting plant growth and yield. Tatsoi plants (Brassica rapa subsp. Narinosa) were grown in the NFT hydroponic module with three treatments of irrigation time, i.e., 6 hours (A6), 12 hours (A12), and 24 hours (A24). Measured parameters include microclimate (temperature, relative humidity, and light intensity) and plant morphology (leaf canopy area, root length, leaf number, and plant weight). The value of the resulting model’s coefficient of determination (R2) for A6, A12, and A24 were 0.9986, 0.994, and 0.9988, respectively. The results of the validation test also show that the prediction data built from the model is very close to the actual data of tatsoi plant’s leaf canopy area. Based on the ANN approach, the MAPE value (0.240) and error (0.00116) A6 treatment were the smallest, whereas the largest R2 value (0.979) was also obtained in this treatment. Furthermore, analysis of the plant weight showed that changes in the leaf canopy area due to water stress ultimately affect plant yield.