Remote-sensed Prediction And Analysis Of Grain Protein Content In Winter Wheat Based On HLM Model

With the continuous launching satellites,remote sensing data have been widely used for crop monitoring and researching,combining remote sensing technology with modern agriculture to achieve intelligent and efficient agricultural production and management,which is the inevitable trend of modern agricultural development in China.With the improvement of national living standard,while pursuing wheat yield,the demand for wheat quality is also increasing.Grain protein content(GPC)of winter wheat is an important index to evaluate its quality.At present,protein monitoring of conventional winter wheat is mostly obtained by field sampling and indoor measurement,and its monitoring results are limited by time span,manpower,material resources,etc,and the differences of climate and ecological conditions in different years in winter wheat production areas lead to the large variation of grain quality in time,so conventional quality monitoring cannot achieve the monitoring and prediction of large-scale winter wheat grain quality.The multivariate remote sensing detection technology can obtain the large-scale "surface" ground object spectrum information periodically and quickly in real time,which provides a new way and method for the quantitative monitoring and prediction of regional crop quality.In this paper,the goal is to realize the prediction of regional winter wheat protein content.Taking the field data of winter wheat quality,remote sensing images and meteorological data as data sources,one hierarchical linear model(HLM)is adopted.The influence of environmental factors on wheat growth is fully considered.The changes of data between different levels are embedded in the model parameters,and the region based on HLM is constructed.The remote sensing estimation model of winter wheat quality at scale can realize the remote sensing estimation of winter wheat protein quality at regional scale.The main research work and achievements in the paper are as follows:(1)The selection of GPC influencing factors of winter wheat was carried out.The correlation analysis of GPC,remote sensing and meteorological data obtained synchronously from sampling points in the study area is carried out.The analysis results show that GPC has good correlation with most spectral parameters and meteorological factors;On this basis,the most relevant spectral parameters(enhanced vegetation index,EVI),accumulated temperature from early March to early June(Tem3s6s),sunshine from the mid-May to early June(Rad5s6s)and rainfall from late-May to early June(Pre5x6s),were selected as the key factors affecting GPC of winter wheat.(2)The GPC quantitative prediction model of winter wheat was established.Based on the selection of key influencing factors of winter wheat GPC,the quantitative prediction model of winter wheat GPC was constructed by using unitary,multivariate and hierarchical linear regression statistical methods.The comparative analysis of the modeling results showed that the model constructed by the same method with different number of independent variables can achieve higher accuracy by selecting the appropriate number of multivariable parameters;GPC-HLM model is superior to other regression models in inversion accuracy for different models with the same number of independent variables,which also shows that the nested relationship between spectral parameters reflecting crop growth and meteorological factors is considered in GPC-HLM model,which can more accurately retrieve the protein content of winter wheat grain,and the model has high reliability.(3)The remote sensing quantitative prediction of GPC of Winter Wheat in the study area was carried out.According to the difference of MODIS-NDVI index between winter wheat and other vegetations during late November and mid May in the study area,winter wheat planting area was classified and extracted.Based on the characteristics of the regional consistency of winter wheat planting,the continuous grid data of winter wheat quality gluten type in the study area were obtained by using the nearest point spatial interpolation method;Using the best spectral parameters(EVI),quality gluten type(Glu)and meteorological factors(Tem3s6s,Rad5s6s,Pre5x6s)raster data as auxiliary variables,based on the GPC-HLM model,the GPC prediction results of winter wheat in the study area are obtained.Compared with the overall investigation,the GPC prediction results are more reliable.The research results obtained in this paper can effectively explain the impact of different levels of influencing factors on the formation of winter wheat grain protein content,and realize the rapid remote sensing quantitative monitoring and prediction of regional winter wheat GPC,which is a beneficial attempt for precision agriculture strategy,and has important practical significance and value.