Abstract
This paper shows how to derive nonstationary spatio-temporal covariance functions via spatio-temporal stationary covariances and intrinsically stationary variograms. Three closely related kernels are employed for this purpose: (i) 2{ϕ( s 1;t 1)+ϕ( s 2;t 2)}−ϕ( s 1+ s 2;t 1+t 2)−ϕ( s 1− s 2;t 1−t 2) , (ii) ϕ( s 1+ s 2;t 1+t 2)−ϕ( s 1− s 2;t 1−t 2) , (iii) ϕ( s 1;t 1)+ϕ( s 2;t 2)−ϕ( s 1− s 2;t 1−t 2) , where ϕ( s ;t) is an intrinsically stationary variogram. Typical examples of covariances generated by kernel (iii) are those of the Brownian motion and fractional Brownian motion. Many new nonseparable spatio-temporal covariance functions are obtained via kernels (i) and (ii).
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
