Extraction And Change Analysis Of Winter Wheat Distribution In Henan Province Based On Multi-source Remote Sensing Data

Agricultural production is the foundation of national social and economic development and the key to land and resources management and food security.Winter wheat is the most widely planted crop in the world.Its planting area and yield are of great significance for the country to formulate economic development plans,regulate crop planting structure and ensure social stability.As a major agricultural province,Henan Province is an important main grain producing area in China,providing 1/4 of the wheat for the whole country,and its output has been ranked first in the country for many years.Timely and accurate access to the distribution of winter wheat in Henan Province is of great significance to formulate agricultural policies and ensure food security.Taking Google Earth engine as the data processing platform and Henan Province as the study area,this thesis determines the synthesis time window of remote sensing images based on MODIS vegetation index(NDVI/EVI),synthesizes Landsat series data and Sentinel-2 images,obtains the images before and after overwintering,and analyzes the high-quality remote sensing images in each observation period.The decision tree,support vector machine and random forest algorithm are used to identify the optimized feature combination image(spectral feature + vegetation index + principal component feature + texture feature + terrain feature),and the most suitable algorithm is selected to produce the time series winter wheat data products in Henan Province in recent 20 years.The changes of winter wheat planting area and planting space and the driving factors affecting the change of winter wheat planting area were analyzed.The main research contents and conclusions of this thesis are as follows:(1)Research on spatial distribution information extraction of Winter Wheat in Henan Province Based on machine learning1)Determination of synthetic time window of remote sensing image.In order to determine the synthetic time window of remote sensing images in each observation stage,this study analyzed the winter wheat vegetation index curve in each observation period.The results show that the winter wheat vegetation index curves in each period and region have obvious characteristics of “two peaks and one valley”,which is easy to distinguish from other ground objects.The intersection of the time period of winter wheat vegetation index increase in three regions is selected as the synthetic time window for the synthesis of Landsat series images and Sentinel-2 images,and the remote sensing median images before and after overwintering are obtained respectively,and the results are more representative.2)Availability analysis of remote sensing images in each observation period.According to the image synthesis time window determined by MODIS time series vegetation index curve,the images before and after overwintering and the number of images of various sensors in each observation time period are counted respectively.At the same time,the observation frequency map of high-quality remote sensing images in Henan Province is drawn.The results show that before 2016,the number of high-quality remote sensing images in each observation period was less than 320,and the number of high-quality images increased sharply after Sentinel-2 was added,with the highest number of 40-60.The amount of remote sensing images is sufficient,which provides a guarantee for the extraction of winter wheat spatial distribution information.3)Production of time series winter wheat data products based on machine learning algorithm.Decision tree,support vector machine and random forest algorithm are used to classify the optimized feature combination images,and the random forest algorithm with the best classification accuracy is used to make a 30 m resolution winter wheat time series product.The results show that the overall accuracy and Kappa coefficient of time series products produced by random forest algorithm are more than 85% and 0.80 in each year.(2)Analysis on temporal and spatial changes and driving factors of Winter Wheat in Henan Province1)Analysis on the changes of planting area and planting distribution of winter wheat.Based on the temporal data products of Winter Wheat in Henan Province and geographic information technology,the changes of winter wheat planting area and winter wheat planting distribution were analyzed.The results showed that the planting area of winter wheat showed an increasing trend from 2002 to 2019.Except for some mountainous areas in northern and central Henan and Shangqiu City,the planting area of winter wheat increased to varying degrees,and the center of gravity of the planting area moved from Xuchang to Luohe.2)Analysis on driving factors of winter wheat planting area change.Combined with statistical yearbook data and meteorological data,Pearson correlation analysis and significance test were used to analyze the driving factors of planting area change.The results showed that the total rural population and mean precipitation were negatively correlated with winter wheat planting area,and other driving factors were positively correlated with winter wheat planting area.It can provide some scientific basis for the estimation of winter wheat yield,the change of planting frequency and the formulation of planting policy in the study area.There are 32 figures,27 tables and 82 references.