Research On The Highland Barley Growth Monitoringand Yield Estimation Based On High-Resolution Remote Sensing Images

In the process of agricultural production,crop growth is important information for agricultural management,and crop yield is a direct indicator for assessing farm productivity and farmers’returns,the ultimate goal of farming.Scientific and efficient acquisition of crop growth information and timely and accurate yield simulation are of great significance for farm production management,agricultural production supervision,assessment of agricultural disasters,national agricultural decision-making,and food storage safety.In this study,highland barley was selected as the research object and Menyuan County of Qinghai Province as the research area.According to thephenological period of highland barley growth,high-resolution remote sensing image and ground measured data were obtained,and NDVI time series data were extracted.Combined with parameters closely related to crop growth,statistical monitoring method was used to conduct grading evaluation and dynamic monitoring of crop growth.NDVI value,meteorological factors and terrain factors were selected as the yield influencing factors,and the sensitive factors were determined after correlation analysis with the measured data.The regression analysis method was used to construct the remote sensing estimation model of highland barley yield.After accuracy verification,the highland barley yield in Menyuan County was estimated.As a result of the study,the paper draws the following specific conclusions.（1）Based on crop growth cycle of high image data,build the NDVI time series data set,and combined with crop measured data at the same time,cover degree and greenness of correlation analysis,the correlation coefficient of 0.75,indicating that NDVI can reflect the growth status of highland barley crop,and the highland barley crop has a strong response capability,which can be used for growth class classification and growth remote sensing monitoring.（2）By calculating the three NCGI_groundvalues and regression analysis with NDVI,the correlation between them is good and the R²can reach 0.90,which can be used in the calibration calculation of remote sensing growth index and used as the"truth value"of growth to compare with NCGI_rs.After the accuracy verification of growth grade and growth index,the accuracy of NCGI_rsobtained based on NDVI inversion is not ideal.In the verification of growth grade,RMSE of phase I is0.26>0.20,which does not meet the accuracy requirements.In the verification of growth index,the correlation with the three phase NCGI_groundis poor.It is not suitable for grade classification and growth monitoring as a growth grade classification index in Menyuan County.Finally,the NDVI value with better fitting degree was selected as the growth grading index in this study.（3）To select the optimum parameters of the growth condition attribute values,the NDVI value and the measured output correlation analysis,the analysis found that single period NDVI value if there is negative correlation coefficient and irregular change,multiphase NDVI accumulation average correlation coefficient were greater than the single stage,the stage and NDVI accumulation average stronger correlation with the measured output and affected by the individual value of error is small,can better reflect the process of highland barley growth.In this study,the average accumulation value（correlation coefficient 0.865）of the six stages of NDVI on July26,which was the end point of the accumulation,was selected as the growth condition parameter of the highland barley to participate in the model establishment.（4）Correlation analysis was carried out on the sensitive factors,namely,terrain factor and meteorological factor and NDVI factor and its different combinations,comprehensive consideration of goodness of fit R²and the adjusted R²value model choice,choose three factor combination model respectively,with 10 groups of expression verifies the accuracy of the measured point data into the model,the results show that the most suitable Men Yuan county highland barley yield estimation model sensitive factors for July 26,the mean NDVI accumulation,R²=0.748,the adjusted R²=0.727,its production minimum relative error is 3.85,the maximum is 21.7%,The accuracy is 78.14%--95.85%,the maximum absolute error is 60 kg,the limit is 70 kg,both within the limit range.Therefore,this model can be used to estimate the yield of Menyuan County.