Development And Application Of Ground And Unmanned Aerial Vehicle Monitoring System For Carbon Exchange In Maize Field

The carbon cycle of agroecosystem is important for global CO₂ cycle balance.The CO₂ exchange between agroecosystem and atmosphere is affected by many factors,and its carbon pool is active.Timely and effective monitoring of farmland ecosystem carbon flux has become an inevitable choice for the development of precision agriculture.Carbon exchange between farmland ecosystem and atmosphere can be characterized in two ways:（1）by measuring carbon flux of farmland ecosystem;（2）by measuring CO₂ concentration in atmosphere.At present,the main equipment used in the two methods are chamber system and CO₂concentration sensor fixed-point monitoring equipment.At present,most of the chamber method systems need manual operation and measurement.It is difficult to collect carbon exchange data of high stalk crops such as maize,therefore,it is not convenient to collect high-resolution CO₂ flux data;the fixed-point monitoring of CO₂ concentration sensor can only carry out single point concentration monitoring,which is not convenient to collect high spatial resolution CO₂ concentration data.The shortcomings of these monitoring devices lead to the lack of research on farmland carbon exchange at the canopy scale.For example,at the canopy scale,it is still unclear whether promote or inhibit carbon exchange in the daytime and at night was same,the influence order of driving factors on carbon exchange is still unclear,and the research on the establishment of the carbon flux and CO₂ concentrationm model is insufficient.Based on the existing common chamber method,this study improved the current common static chamber method,which is convenient to improve the temporal resolution of carbon flux collection data;developed a set of UAV gas collection system,which is convenient to improve the spatial resolution of CO₂ concentration measurement;based on the equipment,analyzed the factors affecting the CO₂ exchange of maize ecosystem and established the model of CO₂ flux and CO₂ concentration.The main research contents and conclusions are as follows:（1）In order to solve the problem that the current static chamber method system is mainly manual and not suitable for collecting CO₂ flux data at night,a full-automatic static chamber method system is developed in this paper.The system can automatically complete six groups of CO₂ flux data acquisition each time,and can realize the measurement of CO₂ flux data change at night.The gas collection and automatic opening and closing device of the chamber can achieve accurate cooperation.The temperature and humidity controol device is designed,which can control the difference between the temperature and humidity inside the chamber and the temperature and humidity outside the chamber in a certain range.（2）In order to ensure the accuracy of the collected data,after the development of the automatic static chamber system,the performance of the system is tested and analyzed.It mainly includes the experiment of chamber sealing and gas collection time,the experiment of chamber environment changes and the experiment of chamber effect on crop physiology.Through data comparison and analysis,the performance of the static chamber system in this study is up to or even better than the commonly used static chamber system.The results indicted that the R²of the two methods was 0.986 and the average difference of CO₂ flux was0.079 u mol m^-2s^-1.（3）At present,CO₂ concentration monitoring can only carry out single point concentration monitoring,which is not convenient for high spatial resolution CO₂concentration data acquisition.In order to obtain high spatial resolution CO₂ concentration data above the canopy,the relationship model between ground measured CO₂ flux and CO₂concentration above the canopy is explored,and a gas acquisition system based on UAV is developed.The UAV gas acquisition system can automatically complete the gas acquisition by setting the route,and can also remotely complete the gas acquisition.The number of gas samples can be collected in one flight is 5,and the gas collecting time and location information of each sample collection point will be automatically recorded.（4）The influence of propeller disturbance is the main problem to be solved in the measurement of gas concentration based on UAV.In order to avoid the change of gas concentration caused by propeller disturbance,the airflow caused by propeller was simulated and analyzed.After simulation and experimental verification,the position of the air inlet of the sampler is determined.Finally,the concentration of gas samples collected by UAV is compared with that of ground monitoring samples.The difference of the CO₂concentration average value between the gas samples collected by UAV and ground station is 1.19 ppm.（5）The diurnal variation of maize carbon flux at canopy scale was obtained using the static chamber system of this study.Then,the influencing factors were analyzed,and the carbon flux model was established.The results showed that the increase of soil temperature and air humidity could promote ecosystem carbon emission,increase maize crop’s consumption of organic matter.The increase of other factors can promote the carbon absorption of maize ecosystem and increase the ability of maize to synthesize organic matter in the daytime,but this factors will play the opposite role in the evening.The analysis of random forest and structural equation model（SEM）indicated that PAR,leaf area and soil water content at 10 cm had great influence on CO₂ exchange in daytime,while leaf area was the most important influence factor of CO₂ exchange in nighttime.Through the analysis of SEM,it indicated that the important factors affecting CO₂ exchange have both indirect and direct effects,while the factors that have little impact on CO₂ exchange generally have only one of the direct or indirect effects.Machine learning method were used to establish NEE and NPP models,in which R² of NEE model established by XGB is 0.9706,and R² of NPP model established by RF is 0.9605.（6）The change of CO₂ concentration can directly reflect the CO₂ exchange in the regional ecosystem,but most of the researches still stay in the regular stage of CO₂ concentration monitoring and change,and lack of correlation analysis and model establishment research.The CO₂ concentration of 10 meters above the canopy of maize was measured by using the UAV gas sampling system in this study.The correlation between CO₂ concentration in the period of 6:00-7:00,and CO₂ emissions at night and main biological factors was analyzed.The multi regression model of CO₂ concentration in the period of 6:00-7:00 was established using CO₂ emissions and maize biological factors,and R² was 0.89.The SVR,RF and XGB models of CO₂ concentration in the period of 6:00-7:00 were established using the same factors.The results of XGB model R² were the largest,0.84.SVR,RF and XGB models of CO₂ concentration were established using the CO₂ concentration at 6:00-7:00 above the canopy,biological parameters and environmental parameters of maize.In the period of 11:30-12:30 and 19:00-20:00.XGB model has good accuracy for the CO₂ concentration estimating.