Casey M. Jelsema
Department of Statistics
Western Michigan University
While analyzing two or more spatial processes, along with spatial dependence, we need to take into consideration the cross-spatial dependence. In this research, we discuss how one can develop a flexible class of anisotropic nonstationary spatial and cross-spatial dependence for large datasets. In geostatistics, kriging is a procedure to make predictions at unobserved locations. Many environmental datasets exhibit lognormality and lognormal kriging equations provide optimal estimates. While punctual kriging provides predictions at point locations, block kriging seeks average predictions over a finite region. We extend lognormal block kriging for multivariable spatial processes and apply our methodology to measurements on clouds taken by NASA satellites. This is joint work with Prof. Rajib Paul and Mona Aldualij.
All statistics students are expected to attend.