Prediction Of Water Status And Determination Of Spectral Threshold Of Winter Wheat Based On Hyperspectral Date

Water,as an important index and main component of crop growth,directly affects the physiological and ecological process of plants,which is of great significance to wheat production.In North China,winter wheat is the main food crop,but due to the influence of climate factors,the precipitation during the growth period lower than the water requirement of crops,so the irrigation has become an important way to replenish its water.However,there are some problems in irrigation,such as water shortage and low water use efficiency.Therefore,hyperspectral technology is used to monitor the water status of winter wheat in real time to adjust irrigation system is very important.Based on this,this paper mainly through the field plot experiment in 2017-2019 to obtain the data of water status and spectral parameters of winter wheat under different irrigation treatments.And then carry out correlation analysis and empirical model construction to find the parameters that can characterize the water status of winter wheat;Combined with the spectral data,the sensitive bands of water status of winter wheat to spectral reflectance were selected,and the prediction model of water status of winter wheat was established by using spectral parameters.The optimal spectral model was proposed to predict the water status of Winter Wheat by introducing growth days and leaf area index,and the most suitable spectral threshold was obtained by combining with the suitable soil moisture threshold,which was used to guide the actual farmland irrigation and provide a theoretical basis for the follow-up remote sensing monitoring of water status of winter wheat.The main conclusions are as follows:1.Through the correlation analysis between plant water parameters and soil water parameters,it is determined that plant water content and 0-50 cm soil water content can be used as parameters to characterize the water status of winter wheat,and combined with spectral reflectance,the sensitive bands of water status are 470-520 nm,620-690 nm and 740-800 nm;2.Using the related data to build the traditional prediction model based on the spectrum,it is found that the prediction accuracy of the model is low and the validation effect is not good.However,the introduction of multiple growth days into the prediction model based on spectral parameters eliminates the shortcomings that the original model can't change dynamically,which greatly enhances the applicability of the model and improves the prediction accuracy at the jointing stage,and the verification error is reduced.Among them,the optimal equation is the soil water content prediction model based on spectral index Rg/Rr,y=3.87x-0.11d+17.22,which can be used as a model to predict the water status of winter wheat;3.Through the introduction of leaf area index optimization prediction model,it can also improve the prediction accuracy of the model in jointing stage and enhance the correlation between reflectance and water parameters except for sensitive bands.Meanwhile,the prediction accuracy of spectral parameter model is improved obviously.Among them,the prediction model of soil water content based on NDVI,y = 354.619 x + 3.507l-269.067,and the fitting accuracy can reach 0.722,the verification effect is better than other;4.By combining the prediction model with the suitable threshold of soil moisture,the spectral index OSAVI and Rg/Rr are selected as the spectral indexes to guide irrigation,and their spectral threshold lower limit equations are OSAVI=0.0312d-0.709 and Rg/Rr=0.0101 d + 1.681.These two equations can be extended to practical application: when the measured spectral index is below the corresponding equation line,it indicates that there is a deficit of soil water,and irrigation can be started to ensure the crop yield grow normally.