Remote Sensing Evaluation Of Citrus Orchard Planting Expansion In Nanfeng County

Nanfeng County is the main planting area of tangerines in China,with a long history of planting.At present,citrus planting in Nanfeng County has become a scale,forming an industrial chain of planting,sales and eco-tourism.The citrus fruit industry has developed into an agricultural pillar industry in Nanfeng County.This article uses Landsat-8 OLI remote sensing image data in 2018 to construct classification feature sets including spectrum,texture,terrain,etc.,compare the accuracy of results of support vector machines,random forests,and neural network algorithms,select the optimal algorithm,and extract 1988-2018 The spatial distribution of citrus orchards in Nanfeng County is analyzed,and the characteristics of the land occupied by citrus orchards in Nanfeng County are expanded year by year.The following main results are obtained:(1)Comparison of Different AlgorithmsThrough the ENVI platform,taking the 2018 Landsat-8 OLI winter image as an example,pixel-level minimum cloud amount synthesis is performed.Based on the same number of classification samples,through the optimal parameter combination of different algorithms,the image classification in 2018 was completed,and the accuracy of the classification results was evaluated.The results show that the overall accuracy and KAPPA coefficient of support vector machines are higher than that of random forests and neural networks,which are 89.86%and 0.86,and the average difference between the classification producer accuracy and user accuracy of each object is the smallest,15.03%and 9.47%;Different algorithms have large differences in the classification and extraction of different features.Among them,the support vector machine has a high classification accuracy for water bodies and woodlands.The producer accuracy is 95.77%and 98.97%,and the user accuracy is 90.62%and80.99%.Therefore,this paper uses a relatively stable support vector machine to extract citrus orchards in Nanfeng County.(2)Dynamic Monitoring of Citrus Orchard Expansion in Nanfeng CountyUsing Landsat long-term sequence image data to construct a classification feature set including spectrum,texture,and terrain features,through the support vector machine algorithm,and using automatic adaptive feature algorithm to extract historical samples,the dynamic expansion of citrus orchards in Nanfeng County from1988 to 2018 Process monitoring.The results show that the average overall accuracy of the image classification results is 85.96%,and the average KAPPA coefficient is0.82.The citrus orchard in Nanfeng County has increased from 64.46 km~2 in 1988 to524.66 km~2 in 2018.Citrus orchards are mainly planted in the range of elevation 0-300m and slope 0-25°.The planting phenomenon in low altitude and gentle areas is obvious;The townships for citrus planting are Baishe Town,Shishan Town and Qincheng Town.The main sources of land for orchards in the county are forest land and cultivated land..(3)Landscape Pattern Change and Evaluation in Nanfeng CountyUsing patch number,average patch area,patch density and patch shape fragmentation index as landscape pattern evaluation indicators,quantitative analysis of the main vegetation landscapes in citrus orchards and research areas showed that:from 1988 to 2018,South The expansion process of citrus orchards in Fengxian County is fierce,the degree of fragmentation is intensified,and the citrus landscape shows a discrete to aggregate change;the average patch area of citrus orchards continues to increase overall,the area of citrus landscapes continues to expand,and the citrus orchards show a continuous trend;The degree of fragmentation of cultivated land and forest land is high,and the overall change is not obvious;the overall patch density of citrus orchards shows a decreasing trend.The patch shape fragmentation index of citrus orchards and the shape fragmentation index weighted by patch area generally show an increasing trend.The plaque shape fragmentation index fluctuated greatly,showing an increasing trend from 1988 to 2006,a small decline from 2006 to 2008,and remained stable after 2008.