Relationship Between Dynamic Changes Of Large-Scale Pig Farms And Peripheral Vegetation Area Changes In Plain River Network Area Based On Multi-Temporal GF-2 Imagery

Pig breeding pollution has become one of the important sources of agricultural non-point environmental pollution in our country,which is attracting more and more attention from all sectors of the society.Accurate prevention and efficient supervision of pig breeding pollution demand relevant departments to quickly grasp the accurate spatial distribution information of pig farms and ascertain the relationship between its change and the change of surrounding vegetation area.Using fourteen GF-2 scenes acquired in six phases from 2015 to 2018 of Shizhuang town and Jiangan town,Rugao city,Jiangsu province,in tandem with corresponding six phases field data,a streamlined framework was developed to extract large-scale pig farms' spatial distribution and ascertain it's changes,which were the main objectives of this study,followed by the proposition of some scientific and reasonable recommendations to guide the local pig breeding industry.Specifically,taking 2018-05-09 as the baseline phase,the large-scale pig farms located in Shizhuang town and Jiangan town were taken as sample and verification objects respectively.Based on the sample objects,the constituting components of large-scale pig farms were firstly defined as highlighted breeding greenhouses,normal pigsties,fecal sewage pools,reservoirs and green vegetation.Then,a chessboard segmentation and BAI(Building Area Index)were introduced to remove the clustered factories and other buildings from GF-2 imagery by selecting suitable BAI threshold,thus the interference on later pigsty extraction was eliminated.Next,the multiresolution segmentation was implemented based on the clustered buildings removed GF-2 imagery.Finally,based on the segmented image objects,the above-mentioned five kinds of ground objects' spectral,geometric and textural features were selected to build up the extraction rule set.Following same extraction rules,the above-mentioned five kinds of ground objects in Jiangan town were firstly extracted.Next,the spatial distribution of verification objects were acquired by using spatial overlay and distance analysis processes.Ultimately,the spatial agreement of the extracted results was reckoned based on corresponding field spatial distribution data to demonstrate the reliability of the extraction rule set.Other five phases' spatial distribution of large-scale pig farms in Jiangan town was extracted by using the same extraction rule set.Thus,the spatial distribution change of pig farms was obtained and the relationship between its change and peripheral vegetation area change was ascertained.The extraction result showed that:1)for the specific plain river network area,the chessboard segmentation with suitable segmentation object sizes and appropriate BAI thresholds could effectively remove the clustered buildings on imagery and laid a good foundation for later pigsty extraction.In this study,the chessboard segmentation object sizes of the six phases images of 2015-08-02,2016-03-21,2016-07-27,2017-09-14,2017-12-22 and 2018-05-09 were set as 400,510,424,553,516 and 580,and the BAI thresholds were determined at 0.085562,-0.023243,0.080496,-0.269791,-0.150923 and-0.059799,respectively.2)Based on the clustered buildings removed GF-2 imagery,the multiresolution segmentation and object-oriented extraction method were able to better extract the spatial distribution of pig farms.The spatial agreement of the extracted results of the above six phases pig farms in Jiangan town was estimated at 88.01%,82.89%,85.19%,88.14%,80.26% and 85.73%,respectively.In conclusion,integration of high spatial resolution GF-2 imagery and the proposed streamlined framework is an effective strategy to quickly and accurately extract the spatial distribution of large-scale pig farms in the specific plain river network area.The change analysis result showed that:1)There was an increasing trend in pig farms' area in Jiangan town,increasing from 47.2568 hm2 on 2015-08-02 to 58.1462 hm2 on 2018-05-09.Especially,from 2016-07-27 to 2017-09-14,the overall area of them increased from 47.3420 hm2 to 60.4218 hm2,with an increase rate of 27.63% and a corresponding peak of area at 60.4218 hm2.2)The area change of pig farms in Jiangan town was very closely related to the area change of its surrounding vegetation.During the whole study period,the expanded part of pig farms was completely originated from the surrounding vegetation.In most cases,the dismantled part of pig farms was completely transformed into vegetation,and in rare cases,the dismantled part was transformed into other ground objects.The area change of pig farms was closely related to pig price,pig market trend,policies and regulations of relevant departments.The general positive pig market,the overall increase in pig price,the environmental protection supervision at all levels,the space layout optimization of local pig farms,the nationwide collection of environmental protection tax,the increase of agricultural green GDP proportion in Jiangsu province and other factors comprehensively led to the increasing trend of the pig farms' area in Jiangan town.This study puts forward some scientific and reasonable recommendations for the subsequent development of pig breeding industry in Jiangan town.