Study On Site-specific Management Zones In Field Using Unmanned Aerial Vehicle And Satellite Images

Cultivated land provides the material basis for the survival and development of people.Cultivated land quality is directly related to the crop yield and final quality.Improving the using efficiency of cultivated land and the sustainable utilization of agricultural resources are of great importance for ensuring food security and increasing farmers’ income.In China,each cultivated field is fertilized as a whole under the unified management,therefore this unified fertilizer method has not considered the spatial difference of soil nutrients within fields.The imbalance of nutrient supply has reduced the utilization rate of fertilizers,causing pollution to the ecological environment of cultivated land,and the economic benefits of cultivated land and the quality of agricultural products have been greatly affected.The precise management zone can be invested in fertilizer according to the site-specific management zones within field,so it can improve the economic benefit of cultivated land,protect the ecological environment of cultivated land,and promote the sustainable development of agricultural economy..The site-specific management zones aim to divide the whole field into several independent and small units according to the growth of crops at field.Each unit has similarity and there is spatial difference between different units.The variable fertilization management is then carried out according to different units.However,the traditional field grid sampling method needs to be sampled according to the requirements in the study area.It is difficult to popularize in practical application,because this method not only consumes a lot of manpower and material resources,but also has low operational efficiency.Compared with the traditional field grid sampling,the new site-specific management zones method by using remote sensing technology is more efficient and can be popularized in larger area,and it has become a hot research topic at present.In this paper,we selected an experimental field of Hailun cooperative in Heilongjiang province as the study area,and the crop growth in this field has obvious spatial differences.We collected the GF-1 images and unmanned aerial vehicle UAV multi-spectral images,and the vegetation index was extracted from UAV multi-spectral images.The method of object-oriented multi-scale segmentation was used to accurately manage the field and the crops in growing period were sampled by grid within the field.We collected the Chlorophyll SPAD value,plant height and used vegetation index to evaluated segmentation precision.Firstly,the site-specific management zones of cultivated land was divided by using the images from different periods of GF-1.Then the images of different periods of unmanned aerial vehicle were used as the data source of the precisemanagement of farmland.Finally,the normalized vegetation index(NDVI)was used to classify the precision management of farmland.Red edge normalized vegetation index(NDRE),red edge reflectivity(RE),red edge ratio vegetation index(RERVI),vegetation index(RVI)were used to divide the data sources into site-specific management zones.The results are as follows:(1)By using object-oriented multi-scale segmentation algorithm,the accurate management division of GF-1 image in late June and late July 2018 is divided,and we found there is obvious difference of variable coefficient between intra and sub interval after the division of accurate management partition.There is a large spatial difference among the management sub-units,and the SPAD value,plant height,ratio vegetation index RVI,vertical vegetation index DVI,in the partition based on the GF-1 image in late June.The interval variation coefficients of normalized vegetation index(NDVI)are higher than those of late July,The coefficient of variation in the region is lower than that in late July.Therefore,the accurate management partition based on GF-1image is feasible,and the precision of late June is higher than that of late June.(2)Using object-oriented multi-scale segmentation method,by taking UAV multi-spectral image as data source,to divide precise management partition and evaluate the optimal segmentation scale is practicable.The variable coefficients between the regions in late June and late July were obviously changed after dividing precise management zones.The precision of precise management zones in June and July was evaluated by SPAD value,plant height and NDVI,NDRE,RERVI,RVI,and SPAD value,plant height and NDVI,NDRE.The variation coefficient of RERVI,RVI interval is larger than that of late July,The coefficient of variation in the region is lower than that in late July.It could be seen that the precise management partition based on UAV multi-spectral image is feasible,and the precision of image in late June is higher than that in late July.(3)The result of accurate management based on vegetation index shows that the precision of red edge vegetation index is higher than that of normalized vegetation index and ratio vegetation index.(4)After comparing and analyzing the precision of multi-spectral image of UAV in late June late July,the precision of multi-spectral image management shows that: The precision of Site-specific Management Zones based on UAV multi-spectral image is higher than that of GF-1,and the precision of selecting image Site-specific Management Zones in late June is higher than that in late July.