Simultaneous assessment of nitrogen and water status in winter wheat using hyperspectral and thermal sensors

Abstract

Listen

Translate article

Remote sensing is a valuable tool for reducing the environmental impact of agricultural practices by detecting crop nitrogen (N) and water status for site-specific N fertilization and irrigation. The interaction between N and water status may produce confounding effects in the acquired spectral reflectance, making it difficult to separate crop deficiencies. The objective of this study was to evaluate the potential of visible and infrared hyperspectral and thermal imaging sensors for N and water status assessment with reduced confounding effects. A winter wheat (Triticum aestivum L.) field experiment combining four N and two irrigation levels was conducted in Central Spain over 2 years. The Nitrogen Nutrition Index (NNI) was monitored (mid stem elongation, final stem elongation, flowering stage) and the crop water status was measured with a leaf porometer at flowering. Two hyperspectral sensors covering the visible and near infrared regions (400–850 nm) and part of the short-wave infrared (950–1750 nm) together with a thermal camera were installed on-board an aircraft to acquire images 300 m above the experiment. In addition, canopy reflectance (400−1000 nm) was measured with a handheld spectroradiometer at ground level. The relationship between the ground-based determination of N and water status with indicators based on remote sensors was analyzed. The planar domain Canopy Chlorophyll Content Index (CCCI) reduced soil background noise and correlated with the NNI in all cases (R2 > 0.44; P < 0.001). Reliable assessment of water status was achieved by using the Water Deficit Index (WDI), which is calculated using the Vegetation Index-Temperature trapezoid. The CCCI distinguished between N levels reducing the confounding effect of the water status, in contrast to the WDI which was mostly affected by the water status. Combining the CCCI and WDI to assess the crop NNI reduced the root mean square error to 0.109, suggesting that the combination of spectral and thermal information could improve the adjustment of N fertilization and irrigation to crop requirements. However, the approach must be validated in other cultivars and environments before making N fertilization and irrigation recommendations.