| Scale effect is a significant problem in the study of scale,which is precisely one of the basic problems in geological research.In the previous work,scale effects have been deeply discussed in geographic information science,landscape ecology,social economics and many other fields.However,most of them primarily focused on the appearance illustration,and few people explained the underlying principles.To uncover the inside mechanism of scale effect,this thesis explores the evolvement course of the patch area varying with the scale changing based on the new land cover data at the resolution of 30 meter from GlobeLand 30.To begin with,a basic evolvement model is built to explain the relation between the patch area and the scale.In the level of patches,the shape of a patch is the key factor that influences the change of patch area.Then,relative support radius and filling degree are established as two parameters to describe the shape of patches.And the change of patch area from original scale to target scale is represented by the patch area ratio.Finally,an empirical model is proposed with these parameters based on Kriging interpolation.With different shapes of patches account may exhibit different area scale evolution laws.According to the patch morphology,the patches are(further)divided into spots,lines and surface patches.Among them,the "spot" patch will disappear directly on the scaling.Therefore,based on the basic model,this paper builds a line scaling model and a surface scaling model through 7219 line patches and 7265 surface patches and then validates the models by extracting 500 line patches and 500 surface patches.The verification results show that the average difference between the patch area ratio and the actual patch area obtained from the "line model" is 14.6%,while by the "surface model" is 17.3%.It also illustrates that the model can explain the estimating patch area in different aggregation resolutions.But analyzing these two models,it can be found that "line model" and "area model" can be unified modeled.So,it is possible to try a unified model to estimat patch area in various aggregation resolutions.Based on the basic model,this paper selects more than 30 thousand patches of training data,builds a unified patch area scaling model,and tests sample data to verify the unified model through 2000 patches.The results show that accuracy of the unified model is 20%,which is a little different from that of the "line model" and "surface model".So,a unified model can be used to explain the evolution of patch size with scale.But the precision of model with average area ratio of 20%still needs to be improved.It is found that the model accuracy is low may due to the presence of complex morphological patches.Therefore,this paper attempts to transform the complex morphological patches into the width of basic uniform patches,and then based on the width of the uniform patch model.Based on the above problems,this paper presents an algorithm for the homogenization of complex patches.The complex patches are segmented by the algorithm to obtain the homogeneous patches.Then,based on the basic model,using 9003 homogeneous patches to construct a uniform patch area scaling model,and 392 homogeneous patches were used to validate the homogeneous model,the model accuracy is 10.8%.Through 200 complex patches verified the unified model and the homogeneous model,the average area ratio of the uniform model is 18.8%,and the average area ratio of homogeneous model is 8.81%.It shows that the precision of model accuracy can be improved by the model of uniform patch area with scale evolution.To sum up,this thesis develops an effective model of patch area evolution with scale changing.Experiments shows that this model explain the mechanism of scale effect in patches of land cover data.The method and model can also be extended to the study of scale effect mechanism in other fields. |
PDF Full Text Request