Change Of Abandoned Farmland In Northern Guangdong Province And Its Response To Targeted Poverty Alleviation Based On SA-PIO-BP Algorithm

Increased abandonment of farmland is a serious threat to national food security.Studying the distribution of abandoned farmland and exploring the influence mechanism of abandoned farmland is of great significance to the revitalization of rural areas in my country.Quick extraction of abandoned farmland is a prerequisite for effective research on abandoned farmland,and it is one of the important directions for data acquisition and dynamic changes.Since the“targeted poverty alleviation”was proposed in 2013,the implementation of relevant policies and capital investment in poverty alleviation will affect the state of abandoned land and will affect the country's food security and the continued development of rural areas.This paper defines the abandoned farmland as more than one year,including the cultivated land that has not been cultivated for a year,to extract the abandoned farmland data.chooses the BP neural network algorithm with a good classification effect,as the original algorithm framework.Aiming at the BP neural network algorithm's shortcomings of slow convergence and local optimization,the pigeon optimization(PIO)algorithm and the degradation simulation mechanism(SA)algorithm are used to optimize,to improve the classification accuracy of the BP neural network for remote sensing images as a change detection Extract the basis of abandoned farmland.To avoid the influence of the number of samples on the classification accuracy,when selecting the training area,the samples of each category are sampled according to the stratified systematic sampling method.The economic development of Guangdong Province is uneven from north to south.During 2000-2015,agricultural land marginalization occurred many times,and farmland was abandoned seriously.Therefore,this paper chose Dabu County,a provincial key poverty alleviation and poverty-stricken county located in northern Guangdong Province,as the research area.Use the optimized BP neural network to classify the remote sensing images of Dabu County to obtain land use data.After extracting abandoned land and researching its spatial distribution,abandonment rate,and reclamation rate.Finally,according to the statistical yearbook of Dabu County from 2014 to 2020,the factors of the abandonment of cultivated land from 2013 to 2019 are analyzed.The conclusions are as follows:(1)Compared with the original BP neural network algorithm model,the accuracy of the PIO-BP algorithm model is improved by about 3%?5%,while the accuracy of the SA-PIO-BP algorithm model is improved by about 1%?2%compared with the PIO-BP algorithm model.The classification accuracy is generally improved by about BP compared with the BP neural network algorithm.4%?7%;(2)The test results show that when the grid is designed as 3×2km for stratified systematic sampling,the classification accuracy of the three classification algorithm models is better than other sampling design effects,and the SA-PIO-BP algorithm model the classification accuracy is the best;(3)The abandonment rate gradually decreased from 2013 to 2019,and the abandonment rate in 2019 was the smallest,only 0.19%;The reclamation rate from 2013to 2019 was also relatively large,and the reclamation rate in 2018 reached 96.74%.Abandoned farmland is mainly distributed in the range of cultivated land with an altitude of 50?350m,and the slope is less than 25°.(4)Abandoned farmland in Dabu County is mainly abandoned for one year,and the area of continuous abandonment for 2 years and abandonment for 3 years accounts for a small proportion.The abandoned farmland in Dabu County was fragmented,abandoned patches smaller than 0.5 hm~2 accounting for 90.49%of total abandoned patches.At the same time,the nuclear density analysis of abandoned farmland shows that the distribution of abandoned farmland in Dabu county appears to be connected,with serious abandonment in the north;(5)In the poverty alleviation work in Dabu County,the return of agricultural workers and the accumulation of financial poverty alleviation funds have a certain degree of restraint on the growth of abandoned farmland.