Research On Generation Of Virtual Forest Scenes Based On The Neutral Landscape Model

Geographic language is the medium of dissemination in geographic information.It plays an important role in the transfer of geographic information and the spatial cognition in human society.With the development and progress of science and technology,the forms of geographic language have experienced various stages such as languages,traditional maps,geographic information systems,and virtual geographic environments.As a new generation of geographic language,the virtual geographic environment uses computer graphics technology,virtual reality technology and network technology to construct a realistic,immersive and interactive virtual geographic scene,providing a way of space perception that is closest to nature.With the maturity of related technologies and theories,the virtual geographic environment is increasingly used in geography studies such as understanding the geographical environment,simulating geographic phenomena,and analyzing the laws of geology.The virtual geographic environment also provides a new method for urban management and other areas of research,evolving concepts such as digital water conservancy,smart city,and digital earth.The virtual geographic environment also has a wide range of applications in forestry research,providing a new visualization platform for forestry managers and researchers to make forest research more efficient and convenient.By expressing,analyzing,simulating and predicting forest objects in a three-dimensional visualization manner,it can help understand complex forest ecosystems and support the planning management and decision-making of forestry.At present,research on large-scale forests in virtual geographic environment focuses on visual representations(automated modeling and expression of individual trees,efficient rendering of large-scale forest scenes).There are relatively few studies on the construction of forest landscapes as a whole object.When constructing large-scale forest scenes,proper layout of different types of trees can enhance the realism of scene expression on the landscape scale,so it is necessary to carry out special research on it.In landscape ecology,large-scale forests are abstracted as landscapes inlaid with patches of the same species of trees.The statistical characteristics of patches in terms of size and shape can describe the spatial distribution pattern of forest landscapes.Existing methods of modeling forest scenes are still unable to express the features at landscape scale.It is necessary to propose a new method to express and simulate the spatial pattern of forest landscapes.This study starts with a neutral landscape model and uses specific landscape metrics to generate forest landscapes with similar statistical characteristics.Based on this,the distribution and properties of individual trees are simulated.Our research work mainly starts from the following aspects:1)This study summarized the existing neutral landscape models and analyzed the control parameters,characteristics,and deficiencies of each model.It is found that the existing models are difficult to generate complex landscapes,and it is impossible to differentiate the distribution characteristics of various types.Aiming at this problem,this paper proposes a neutral landscape model based on graph structure.Based on the commonly used landscape metrics,a metric representing the shape characteristics of the patch is added as a parameter of the model.Not only the size and shape of each patch can be controlled,but also the model is able to differentiate the spatial distribution characteristics of each type.The proposed neutral landscape model addresses the simulation of vegetation distribution at landscape scale.2)On the basis of the simulated forest landscape,individuals are assigned locations and their attributes in order to realize the simulation of individual attributes on the microscopic scale.Aiming at the landscape composed of single tree species and the landscape composed of mixed tree species,different random algorithms are proposed to quickly generate tree distribution points.Finally,the properties of individual trees are assigned using the characteristics of tree diameter distribution and the relationship between diameter and height,diameter and crown width.The proposed algorithm can guarantee the random distribution of trees without overlap between crowns.3)The actual vegetation distribution in the experimental area is used to verify the model and the simulation results and the main sources of error are analyzed.The simulation results are expressed in a three-dimensional visualization manner.The research explores the controllability of the model parameters for the simulation results and analyzes the factors affecting the model's operating efficiency.The experimental results show that the neutral landscape model proposed in this paper can simulate the spatial distribution of vegetation at the landscape scale and can express the complex landscape spatial patterns.At the same time,the method proposed to simulate the location distribution and attributes of individual trees can be used in forest scenarios to provide reasonable tree geometry information for three-dimensional realistic visualization.The overall idea used in this study is from the landscape scale to the micro scale and from the spatial pattern to the point distribution and can provide a new idea and new method for the modeling of large-scale forest landscapes.