Monitoring Index Study Of Soil Water Content In Summer Maize Fields Based On UAV Remote Sensing Data

Soil water content is a key factor affecting crop growth and yield.Therefore,monitoring soil water deficit in arid areas is of great significance for the development of water-saving irrigation agriculture and the guidance of farmland irrigation decision-making.UAV remote sensing has the advantages of high resolution,high flexibility and low cost,which can quickly obtain high-resolution image data of planting areas,providing guarantee for timely acquisition of dynamic changes in soil moisture.Studies were conducted in maize fields with different irrigation treatments at seedling stage,jointing stage,tasseling stage,silking stage,blistering stage in 2019 and 2020.The thermal and RGB images of UAV were used to obtain the temperature data and spectral data at 9:00,11:00,13:00,15:00 and 17:00in the test area.Then maize physiological parameters including vegetation coverage and leaf area index,meteorological parameters including net solar radiation,air temperature,and atmospheric humidity,and monitoring parameters including temperature histogram features,image texture,and EXG index were obtained or calculated High spatio-temporal soil water content maps were obtained by the combination of all the above data,providing certain technical support for agricultural irrigation system with higher accuracy.The main research results are as follows:(1)The land surface temperature is affected by meteorological changes and crop growth factors,which have problems of poor accuracy of soil water monitoring.The variation of land surface temperature,solar altitude,vegetation coverage and vapor pressure deficit at different times during the whole growth period of maize in 2020 were studied for the establishment of the multifactor soil water monitoring model.The results showed that the diurnal variation of land surface temperature presented a negative skew distribution and fluctuated strongly in the maize growing season.The diurnal variation curve of the solar altitude was symmetrically distributed around 13:00,and the solar altitude angle decreased slowly at 13:00 during the experiment.Maize growth led to a gradual increase in vegetation coverage,which tended to be stable after maize stepped into silking stage,while the vapor pressure deficit showed a fluctuating trend.During the maize growing season,land surface temperature,solar altitude,vapor pressure deficit and soil water content showed negative correlation,vegetation coverage and soil water content showed positive correlation.Compared with the simple linear regression model based on land surface temperature,the multiple linear regression model considering environmental factors and crop growth changes could effectively improve the monitoring accuracy of soil water content(R~2=0.380,NRMSE=16.9%,n=300,P<0.001).(2)The establishment and analysis of a monitoring model of soil water content based on the absence of meteorological data.The characteristics of texture and temperature histogram of thermal infrared images at 13:00 of corn growing season in 2019 and 2020 were studied,and tried to reduce the interference of meteorological change on soil moisture monitoring by arithmetic.Finally,the soil water monitoring model based on thermal infrared image data was established.The results showed that meteorological factors and crop growth affected the uniformity of gray scale(temperature)distribution of thermal infrared images,the correlation between texture characteristics and vegetation coverage was slightly higher,and the correlation between temperature histogram characteristics and vapor pressure deficit was strong.Normalizing or averaging the index can effectively weaken the influence of the above factors,which was negatively correlated with the time scale selection,and the best time scale was daily.However,there was an obvious shift phenomenon when estimating the extreme value of soil water content in the daily scale normalized index.The partial least squares regression model constructed from the daily scale mean index had a good performance in estimating soil water content(Calibration:R~2=0.765,NRMSE=9.8%,RPD=2.06,n=96;Validation:R~2=0.718,NRMSE=11.3%,RPD=1.88,n=96).(3)At present,thermal infrared image has poor recognition of complex ground objects,which have problems of inaccurate canopy and soil temperature collection The results showed that the user accuracy of the Otsu-EXG-Kmeans algorithm combined with EXG index was as high as 95.9%,and the algorithm had better extraction accuracy of canopy and soil temperature(canopy temperature:r=0.79,RMSE=1.90℃,n=60,P<0.001;soil temperature:r=0.77,RMSE=3.47℃,n=60,P<0.001).The daily average variation trend of CWSI obtained by Otsu-EXG-Kmeans algorithm can accurately monitor the change of soil water content(r=-0.738,n=60,P<0.001).EXG index was negatively correlated with CWSI,canopy temperature,soil temperature and mixing temperature(P<0.001),and was positively correlated with soil water content,vegetation coverage,leaf area index and plant height(P<0.001),which can respond well to soil water deficit.