Identification By Remote Sensing Based On Machine Learning And Suitable Area Prediction Of Amygdalus Persica

With the development of niche theory,computer science,statistical model and other sciences,it is possible to combine various environmental variables to predict species distribution in geographical space.The application of various species distribution(niche)models is born,and it is widely used in the research of agricultural ecology and other fields.They are often used to simulate the relationship between species niche and environment Based on the growth pole theory,relying on the core characteristic industries to drive the regional development,and then radiating to drive the characteristic agricultural industry layout in a larger region,gradually forming a fusion pattern,trying to increase farmers’ income and create a cohesive development,at the same time,it is of great significance to the ecological protection of species resources and the realization of Rural Revitalization Strategy.Among many species distribution models,no species distribution model can be perfectly applied to all species.In this case,a large number of original models have emerged,which improve the models and algorithms.They provide the possibility for more accurate and reasonable simulation results.This paper selects Max Ent model,and then adopts three categories with high accuracy classification results The classification methods are SVM,objectoriented decision tree and random forest(RF);the high resolution hyperspectral image of sentinel II is selected by phenology to extract Amygdalus persica tree features.The Amygdalus persica tree distribution area generated by the best time image remote sensing is generated,and random sample points are generated to collect sample points in the field The results of the experiment were used to simulate the suitable environment variables of Amygdalus persica trees in the whole country.Then,the data generated by the latest cmip6 model will be used to simulate the distribution in 2050 2070 corresponding to rcp4.5,rcp6.0SSP245 and SSP460 of the fifth IPCC Assessment Report(AR5),rcp6.0SSP245 and SSP460 The typical concentration of species represents the distribution of suitable region under the path.In addition,combined with the R language open source package biomod2,four of the better and commonly used algorithms in the package,GLM,GBM,Ann and RF,are called to analyze the potential geographical distribution of huangtao under the current environmental conditions,SSP245 in 2050 and SSP460 in 2050.The results show that Max Ent is the best to simulate the current suitable area of yellow peach,and the effect of the future climate scenario model simulated by random forest model in biomod2 is better.Although Max Ent maximum entropy model is slightly better on the whole,the prediction result of the future climate scenario model is slightly lower than that of the random forest model in biomod2 integrated model;From the distribution of yellow peach,the suitable area of yellow peach will increase slightly in the future.On the whole,the suitable area of yellow peach will migrate to the Northeast under SSP245 scenario,and not only to the Northeast high latitude areas,but also to the northeast and southwest areas under SSP460 scenario.The results provide a certain basis and reference for the subsequent introduction and cultivation,biodiversity protection and resource allocation.