The Study Of Land Use/Cover Change In The Nature Reserve Based On Kernel Density Estimation

The establishment of nature reserve plays an important role in protecting precious species resources and ecological environment.The scale of nature reserves in China has expanded significantly.However,due to the development of economy,the livelihood needs of residents etc.,the ecological environment in the reserve is generally affected by the interference of human activities,such as tourism development,animal husbandry,farming,mining,deforestation etc.The situation can be reflected by land use/cover change(LUCC).The change of land type can directly reflect the interaction and influence of human and land system.This topic has always been hot for scholars.In the study of LUCC in nature reserves,most scholars at home and abroad first extract the land types of nature reserves by classification of remote sensing images,then combine with GIS or landscape ecology related theories and mathematical statistical methods to carry out quantitative analysis,and explore the driving factors.In this paper,based on KDE(Kernel Density Estimation),it is proposed to monitor the hot spots of land type changes in nature reserves to reflect the impact of human activities on the ecosystem and to provide advice for planning decisions on the management of reserves.KDE(Kernel Density Estimation)is a common method for spatial aggregation analysis.It can reflect density of distribution,spatial aggregation and hotspots.This method takes the sample point as the center and produces a circle according to the search radius.The grid density at the center of the circle is the highest.The farther away from the center of the circle,the lower the density,until the boundary of search radius,density value is reduced to zero.The result of KDE is influenced by kernel function and search radius.The selection of kernel function is not the key factor.The search radius determines the smoothness of the result.The larger the radius is,the smoother the result is.The smaller the radius is,the lower the smoothness is,but themore details about the hot spots are shown.In this paper,KDE is applied to the land use change in reserves because its result can clearly and directly reflect the aggregation of input land blocks.It can also visualize the location of hotspots(higly-aggregating places).For changes in land types that reflect human interference in reserves(e.g.,water to farmland),the KDE result can indicate higly-aggregating areas where should be especially payed close attention to,provide spatial information for analyzing the causes and monitor the land use change in reserves.Taking Henan Danjiang Wetland National Nature Reserve,China as an example,this paper has analyzed the change of land use types from 2007 to 2017 in the reserve.It is found that a small part of the land types change in the reserve is related to natural factors:(1)Land types changing from vegetation to water,from farmland to water and vegetation are related to the development of river valley,which are mainly located at the southwest boundary of the reserve and the core area of mountain.(2)Land type changing from water to vegetation is related to the periodic fluctuation of water level,and his mainly concentrated at the junction of the land mountain and the river channel in the southeast of the reserve.Most of the changes are closely related to human activities,and the reserve has always been widely influenced by human activities:(1)Land types changing from bare area and water to farmland are related to seasonal tillage,which are widely distributed in the reserve except the core area of the mountain,and the residential area is also in this region.(2)Land types changing from bare areas to water and vegetation are related to sand mining and recovery.Hot spots are mainly concentrated in the southeastern part of the water area and near the river channel.(3)Land type changing from water to bare areas is related to development of tourism traffic,construction of wharf facilities and passage of vessels,which is mainly located in the core region of the water area and the southern boundary of the reserve,and the hotspot area is obvious and in large number.It is suggested that thereserve should pay more attention to the monitoring of land use changes.This paper also proposes a raster data classification method.The prime number generated by Eratosthene sieve method is used to classify the raster so that the land type changes can be uniquely identified by raster values through arithmetic operation without confusion.The method proposed in this paper is operable and is implemented by open-source GIS software platform.It can reduce the research cost and is worth popularizing and applying.