Monitoring Rice Plant Potassium Content Using the Combination of Remote Sensing and Meteorological Data

Abstract:

Potassium (K) plays a significant role in the formation of crop quality and yield and accurate estimation of plant potassium content (PKC) using remote sensing (RS) techniques is of importance for the precise management of crop K fertilizer. Meteorological data mainly change environmental condition to influence crop K uptake. However, no studies currently focus on combining RS and meteorological data to estimate crop K nutrition status. The aim of this study is to understand whether the inclusion of meteorological into RS model can improve the K estimation accuracy. Correlation analysis was conducted between rice (Oryza sativa L.) PKC and transformed spectra [reflectance spectra (R), first derivative spectra (FD) and reciprocal logarithm-transformed spectra (log (1/R))] and meteorological data (temperature, humidity, precipitation, etc.). Then, the least absolute shrinkage and selection operator (LASSO) was used to select important bands (IBs) and important meteorological factors (IFs) that contributed the most to rice PKC. Finally, the spectral index and machine learning methods (partial least-squares regression (PLSR) and random forest (RF)) were used to construct rice PKC estimation models based on transformed spectra, transformed spectra + IFs and IBs, IBs + IFs, respectively. Results showed that normalized difference spectral index [NDSI (R₁₂₁₀, R₁₁₀₅)] had a moderate estimation accuracy for rice PKC and PLSR (FD-IBs) and RF (FD-IBs) models based on FD could improve the prediction accuracy. The R² of three models were 0.51, 0.69 and 0.71, respectively, and the RMSEs were 0.49%, 0.37% and 0.40%, respectively. Besides, among the meteorological factors, daily average temperature contributed the most to rice PKC, followed by daily average humidity. The estimation accuracy of the optimal rice PKC models could be improved by adding meteorological factors into the three RS models. The R² of the models were increased to 0.65, 0.74 and 0.76, respectively, and the RMSEs were decreased to 0.42%, 0.35% and 0.37%, respectively. Meanwhile, the optimal models showed better stability and temporal transferability in different datasets. This study demonstrates that the combination of spectral and meteorological data could effectively predict the K nutrition status of rice.