Abstract:
Abstract: The measurement and estimation of evapotranspiration plays an important role in agriculture. In this study, we designed a plant phenotype and evapotranspiration monitoring system based on weighing lysimeter and multi-images. A total of 24 small weighing lysimeters, integrated RGB imaging sensors, multispectral imaging sensors and a 2D laser scanner were integrated with a gantry to control the movement in order to build weighing lysimeter plant phenotypic monitoring system, realizing automatic monitoring of RGB, red (668 nm), green (560 nm), blue (475 nm), the red edge (717 nm), near infrared (840 nm) image information and plant height information during plant growth period. Each lysimeter had the length of 1 m, a the width of 0.75 m and the depth of 2 m. The effective planting area was 0.75 m2. The total area was 18 m2. The intact soil was filled into the lysimeter. The lysimeter was equipped with data collecting system. The pressure signal was transformed into electrical signals. Wheather stations were installed to measure air temperature, air humidity, radiation, wind speed, precipitation, and the others. The phenotypic monitoring module was composed of RGB high speed color camera, 5-channel multi-spectral camera and laser scanner. The motion control module was of programmable logic controller in motor control cabinet in charge of moving ganty. The outside of programmable logic controller had the man-machine interaction interface. If the automatic control system failed, it could be manually controlled through the man-machine interaction interface. In this paper, phenotypic data, meteorological station data and lysimeter data were combined to not only estimate crop evapotranspiration in a large area, but also obtain various crop index and plant height information. The system was then tested at the designed normal speed and sampling frequency meeting the practical requirements. The results showed that the single journey time of the system was 142 s when the RGB and multi-spectral imaging sensor images were taken every 5 s, laser scanning once every 1 s. After a journey, the system could automatically collect 28 RGB and multi-spectral images, from which plant growth information could be derived. The obtained image data were stored in time format in a folder. During the motion control test, the single journey, time for single journey, pulse for single journey were recorded at designed motor rotation speed of 0.111 and 0.167 m/s. The relative error between the measured and the designed values was 1.8%-6.0%, indicating that the motion control performance was well. The system was used for estimation of evapotranspiration after seedling estimation of winter wheat. The RGB images were collected every 10 days. The average daily coverage and crop coefficient were calculated to calculate evapotranspiration. Finally, the estimated evapotranspiration had the relative error of 16.62% averagely, indicating the reliability of evapotranspiration estimation by the system. In addition, the acquired multi-spectral images and laser scanner data of summer maize were revealed, suggesting that the system could reliably obtain crop index and plant height information such as normalized difference vegetation index, difference vegetation index, ratio vegetation index, normalized difference green index, soil adjusted vegetation index and so on. In sum, this system integrated the lysimeter and multispectral images so as to provide an valuable technology and equipment support for real-time monitoring, accurate diagnosis of crop evapotranspiration and researches on crop growth status. In future, It is necessary to carry out researches on the acquisition of other phenotypic information and image fusion so as to obtain more crop information.