Establishment and application of prediction model of soil water in walnut orchard based on unmanned aerial vehicle thermal infrared imagery

Abstract: Studying the soil water status of the walnut orchard and conducting reasonable irrigation play an important role in relieving the pressure of irrigation water in arid or semi-arid areas. With the development of thermal imaging technology, thermal image is a viable alternative to point measurements, since it offers the possibility of rapidly measuring a large number of plants and integrating plant temperatures over entire fields and producing a map of the plant water status distribution in the field. In this study, the canopy temperature of walnut tree was continuously observed with thermal infrared instrument (A310 f) at 13:00 and 14:00 per day in the main growing season of 2016 and 2017, and meteorological factors (air temperature, air humidity, solar radiation, wind speed, and precipitation) and the soil water content in 0-80 cm depth were observed from May 2016 to September 2017. On August 11th, 2017, thermal image acquisition was carried out using unmanned aerial vehicle's thermal imaging system (TC640) in the continuous irrigation area and drought area. The results showed that 40-60 cm soil layers may be the main areas where walnut roots absorb water. In general, when canopy temperatures reach the highest in the afternoon (13:00 and 14:00), canopy temperature will be higher than air temperature and the range of variation is 0-5 ℃. In sunny weather conditions in 2016 and 2017, multiple regression analysis was performed based on canopy-air temperature difference, solar radiation, wind speed, vapor pressure deficit (VPD) and relative water content (RWC) in 40-60 cm depth. Coefficients of determination in 2 fitting equations were 0.57 and 0.69 respectively. The canopy-air temperature difference was negatively correlated with the soil water content, but positively correlated with the solar radiation, and the contribution of soil water content reached 75%, which was higher than the solar radiation through principal component analysis. The soil water prediction model was established using the data of canopy-air temperature difference and soil water content at 13:00 in 2017, with the coefficient of determination of 0.64. At the same time, the measured data at 14:00 were used to verify the model established, and the coefficient of determination was 0.61, indicating that the model had a certain accuracy degree of fitting. Finally, the soil water model was used to diagnose 2 different water conditions of walnut areas, which proved that it had a good practical application effect. The range of RWC change in the continuous irrigation area was 0.5-0.6, while the range of variation for RWC in the drought stress area was 0.41-0.5. For the 3 sample trees in the continuous irrigation area, RWC values from the simulation were 0.57, 0.53 and 0.55, and measured values were 0.62, 0.57 and 0.52, respectively. For the 3 sample trees in the continuous drought area, RWC values from the simulation were 0.49, 0.44 and 0.42, and measured values were 0.44, 0.42 and 0.39, respectively. This study combines the fixed thermal imaging equipment with the thermal imaging system of unmanned aerial vehicle to study the canopy temperature of walnut trees, and successfully achieves the conversion from the theoretical model to the practical application and the extension from the individual level to the regional scale. Finally, the constructed soil water model will provide a basis for scientific water resources allocation in walnut orchards of the northern China.