Remote Sensing Monitoring Of Winter Wheat Sowing Date Based On High Spatial And Temporal Resolution Data

Sowing date is an important factor affecting wheat yield and quality.Monitoring sowing date in the early stage of winter wheat growth is of great significance for remote sensing yield estimation and classification management of agricultural production.The rapid development of remote sensing technology provides a new method for low-cost monitoring of winter wheat sowing season in large regions.There are two main problems in the existing monitoring methods of sowing season.First,the method of phenological monitoring based on remote sensing data of the whole growing period of winter wheat is used.Secondly,the spectral monitoring method is only used in the early stage of winter wheat growth,which cannot be guaranteed due to the weak precision of vegetation signal in the early stage.n this paper,the planting area of winter wheat in Hebei province was taken as the experimental area,and based on the curve variation law of time series in the early stage of winter wheat growth,the remote sensing monitoring model of sowing date was constructed by using the multi-temporal timing information of the early stage of winter wheat growth,so as to improve the monitoring accuracy and advance the monitoring time,so as to provide a new research idea for sowing date monitoring.The main research contents and conclusions are as follows:(1)In the process of data processing in the study area,the use of spatio-temporal data fusion method(FSDAF),the MODIS(high resolution)and Landsat8 OLI(high spatial resolution image data fusion processing,get the winter wheat growth prophase 8 d temporal resolution and spatial resolution of 30 m high space-time resolution time sequence images,solve the winter wheat growth prophase vegetation signal is weak,high temporal resolution and high spatial resolution remote sensing data of winter wheat sowing date monitoring resulted from the trade-off between the problem of limited timeliness.(2)Combined with the national land use map,two types of features of cultivated land and non-cultivated land were extracted.Based on the HSV color space transformation of MIR,NIR and RED bands,the spatial distribution differences between wheat growing areas and non-wheat growing areas on "H-NDVI" in the image were analyzed,and the area of winter wheat planting in the research area was obtained by mask treatment with S-slope.(3)Based on winter wheat processing time sequence image smoothing filtering results in the study area,using stepwise regression method to select the best phase characteristics of the data,establish the sowing date with ground survey data of remote sensing monitoring model,the measured data and model test results of the determination coefficient R2=0.70,reached the extremely significant correlation,so the use of winter wheat growth prophase long phase of planting the time-series data can realize remote sensing monitoring.(4)On winter wheat growth prophase vegetation index time series to deal with the envelope curve,to envelope area and depth characteristics as independent variables respectively sowing-date remote sensing monitoring model is established,in which to envelope depth of the independent variable to establish the sowing date monitor model accuracy is better than to go to the envelope area of sowing date monitoring model is established for the independent variable,the forecast of planting the sowing date and the actual decision coefficient between R2=0.60,and combined with the feature of envelope area and depth of sowing date higher precision of remote sensing monitoring model(R2=0.67),The determination coefficient of remote sensing monitoring model based on single de-enveloping feature is 11.67%higher than that based on single de-enveloping feature..(5)A remote sensing monitoring model for sowing date was established by combining the multi-temporal data at the early stage of winter wheat growth with the feature removal parameters of the continuum.Its determination coefficient R2=0.74,which was 5.71%higher than the remote sensing monitoring model for sowing date based on the multi-temporal data and 10.45%higher than the remote sensing monitoring model for sowing date based on the feature removal of the continuum.To sum up,the optimal remote sensing monitoring model was established by combining the multi-temporal data at the early stage of winter wheat growth with the continuous removal of characteristic parameters.