Correcting negative kriging weight in krigeST

Dear R-sig-Geo users,

I'm currently using krigeST for 3-d geostationaly sattlite data
interpolation, and I'm trying to estimate "Interpolation variance" in
Yamamoto et al, 2000.
(https://link.springer.com/article/10.1023/A:1007577916868)

My problem is that the variance is somtimes negative. I'm thinking
that it is due to the negative kriging weights. I found that there is
a solution in Deutsch (1995)
(https://www.sciencedirect.com/science/article/pii/0098300496000052#).

But I'm not sure there is already a option in krigeST fuction. Could
you let me know about the solution?

Thank you for your consideration.


Best regards,

Sojin