GM(0,N)模型在湿地面积预测中的应用——以莫莫格湿地为例  

Application of GM(0,N)model to prediction of wetland area:A case study on Momoge wetland

作　　者：史文杰 李昱[1] 刘学智 张小丽[1] 张弛[1] 

机构地区：[1]大连理工大学建设工程学部水利工程学院,辽宁大连116024 [2]广州珠科院工程勘察设计有限公司,广州510611

出　　处：《南水北调与水利科技》2017年第6期101-107,共7页South-to-North Water Transfers and Water Science & Technology

基　　金：水利部公益性行业科研专项(201401014-2);国家自然科学基金(51279021)~~

摘　　要：预测未来年度湿地面积,对研究湿地生境变化趋势、保护湿地有重要作用。建立灰度GM(0,N)模型旨在提供一种简便的方法,预测湿地水面面积大小。首先应用灰色关联度分析模型,量化确定了对湿地面积影响程度较大的相关因素,分别是:莫莫格湿地年降水量、嫩江径流量、洮儿河径流量。利用这三个相关因素建立了GM(0,N)预测模型,对莫莫格湿地面积进行了模拟预测。为了提高精度,对GM(0,N)模型进行了修正。利用残差和后验差检验方法对模型作了可靠度分析,检测结果显示:修正的GM(0,N)模型平均相对误差9.1%,后验差检验等级为1级,多元线性回归模型平均相对误差15.5%,说明灰度预测模型对于莫莫格湿地水面面积预测具有一定优势。The prediction of wetland area in the coming years is critical for studying the change trends of the wetland habitat and preserving the wetlands. We developed a grey GM（0,N） model as a simple and convenient method to predict the water surface area of wetlands. First,we used the grey relational analysis model to quantitatively determine the main correlative factors that greatly influenced the wetland area. They were： annual rainfall of Momoge wetland, runoff volume of Nen River and Taoer River. These three factors were used to establish the GM（0, N） model to predict the area of Momoge wetland. To improve the prediction precision, we modified the GM（0, N） model. Then we conducted the residual test and posterior variance test to evaluate the reliability of the model. The average relative error of the modified GM（0, N） model was 9. 1%, and the posterior variance test grade was Grade 1,while the average relative error of the multiple linear regressive model was 15.5%. This suggests that the modified GM（0,N） model has an advantage in a practical application.

关 键 词：GM(0 N)模型 灰色关联度分析 湿地面积 预测 残差 

分 类 号：N941.5[自然科学总论—系统科学] Q149[生物学—生态学]