Applications of Uncertain Canonical Process in Spatial Predictions

摘要：

　　The Quiver trees spread in arid Namibia and South Africa.In order to investigate the spatial distribution pattern associated with climate changes,from 2002 to 2008, Quiver trees observational data were collected.Due to the financial constraint, data set is not large enough and even uncompleted: only forty-three out of the originally designated fifty-three observational sites, which causes extreme difficulties in the spatial pattern analysis.In our preliminary analysis, we found that the growth of the Quiver trees is robust to climate variables, say, rainfall and temperature and further that even for close sites geographically, the growths can be very diversified,which implies conventional kriging prediction or the inverse distance approach not working in the circumstance.In this paper, we define signal to noise ratio as a habitat measure for the Quiver Trees and engage the uncertain interpolation and extrapolation approaches including the uncertain canonical process regression and uncertain kriging prediction, for the habitat measure of the ten ”missing” sites.

关键词： Quiver trees habitat measure habitat distance uncertain measure uncertain canonical process regression uncertain kriging prediction

作者: Danni Guo Renkuan Guo Yanghong Cui

作者单位: Kirstenbosch Research Center,South African National Biodiversity Institute,Private Bag X7,Claremont Department of Statistical Sciences,University of Cape Town,Private Bag,Rondebosch,7701,Cape Town,Sou

会议类型: 国内会议

会议名称: 第九届中国不确定系统年会、第五届中国智能计算大会、第十三届中国青年信息与管理学者大会

会议地点: 南京

会议语种:英文

页码: 444-461

在线出版日期: 2011-07-27（万方平台首次上网日期，不代表论文的发表时间）