| Forest landscape marks forest ecosystem as its main body.The research and analysis of forest landscape can reveal the spatial structure and function of forest,which is the basis of optimizing the allocation of landscape resources and serves for forest resource management and decision-making.Based on the theory of forest management and landscape ecology,with ArcGIS 10.2,Fragstats 4.2,ENVI 5.3 and IDRISI 17 as the technical software support,this paper chooses Ningxiang city as the research object,and takes Landsat remote sensing image and forest resource survey data as the data source,and interprets the forest landscape classification map in 2009,2013 and 2017.On this basis,using transfer matrix and landscape index method,this paper analyzes the dynamic changes of forest landscape pattern in the study area from 2009 to 2017,and constructs CA Markov model to predict the forest landscape pattern of Ningxiang city in 2021.The main conclusions are as follows:(1)Classification of forest landscape types.With reference to the land classification system of Forest Resource Planning Designing and Surveying,the forest landscape types in the study area are divided into two levels and eight categories.Arbor forest land,bamboo forest land,shrub forest land and other forest land are classified as forest landscape,while cultivated land,water area,construction land and other non forest land are classified as non forest landscape.According to the forest landscape classification map obtained by interpreting remote sensing images,it is found that the forest landscape is mainly distributed in the mountains in the West and southeast of Ningxiang City.In the non-forest landscape,the cultivated land landscape is mainly distributed in the flat area in the northeast of Ningxiang City,and the construction land landscape is mostly distributed in the flat area in the northeast of Ningxiang city.(2)Based on the forest landscape characteristics and transfer matrix,analyze the dynamic changes of forest types,and study the conversion rules between various landscape types.In terms of quantity,forest landscapes account for about 44 to 45%,less than non-forest landscapes.Among forest landscapes,arbor forest landscapes account for the highest proportion,accounting for about 35-37%.Among the non-forest land landscapes,the cultivated land landscape area is the largest,up to 51%.From a spatial point of view,the landscape transfer is dominated by cultivated land landscape and woodland landscape.After 2013,the transfer rate of cultivated land landscape decreased significantly,indicating that the study area has taken certain measures to strengthen the protection of cultivated land landscape and ease the decline of cultivated land landscape.At the same time,thanks to the implementation of afforestation and afforestation of barren hills and wasteland,all types of forest landscapes in the forest landscape are transformed into each other,and a large number of other forests are converted into arbor forests land.(3)Based on landscape index,the forest landscape pattern of Ningxiang city is analyzed.Seven indexes are selected from the patch level and landscape level to study the dynamic change and trend of forest landscape pattern.The results show that,on the patch level,the fragmentation degree of arbor forest landscape is the largest,and the cultivated land landscape has high aggregation.From the perspective of landscape level,the forest landscape in the study area shows a trend of balanced distribution,landscape aggregation degree decreases,and all kinds of forest landscape heterogeneity develops towards diversity.(4)Build CA-Markov model.Based on the CA-Markov model in IDRISI 17.0 software,firstly,the forest landscape classification data in 2009 and 2013 are used to predict the 2017 forest landscape classification data,and the comparison analysis with the actual 2017 forest landscape classification data is carried out to test the accuracy of the model.It is found that the kappa coefficient of 2017 forest landscape classification prediction value and 2017 forest landscape classification actual value is 0.83,which meets the accuracy requirements.Then on the basis of 2017 forest landscape classification data,combined with six factors such as elevation,slope,river distance,road distance and GDP,the suitability Atlas of each forest landscape type is calculated in MCE module of IDRISI software,and the CA-Markov model of the final study area is obtained.(5)Simulate the forest landscape pattern of Ningxiang city in 2021.Based on the forest landscape classification data of Ningxiang city in 2017,the suitability atlas is imported to predict the forest landscape classification of Ningxiang city in 2021 in CA-Markov model.It is found that the area of cultivated land and construction land will increase,the area of forest land will decrease,and the heterogeneity of landscape will decrease.It is necessary to further optimize the landscape pattern and ensure the coordinated development of the region. |
