Hello Amirhossein,

No, without any data from one source/vector, there's no foolproof way to estimate how scores from that source would or would not correlate with scores from any other source.

What other options might you have?

1. Use previous months' data to estimate the relationships of interest and use these correlations as your best guess as to what current degree of association might be.

2. If one of your gauging stations is so similar to the one with missing data that you could consider it a proxy indicator, you could use the observed correlations of that station with the others as estimates. Of course, you would have to decide what sorts of characteristics justified designating another station as a proxy case (e.g., geographic proximity, altitude proximity, shade/sun exposure similarity, etc.).

Good luck with your work.

Dear David Morse ,

Thanks a lot for your perfect answer

I have read about the Ordinary Kriging method introduced by Isaaks, E.H., Srivastava, R.M., 1989. Applied Geostatistics. Oxford University Press, New York to estimating unknows quantities but it was an old and laborious method.

So I'm looking for a novel method similar Kriging but with more practicality.

best wishes

There is quite a body of literature on the topic of "imputation" which should be considered. The biomedical world uses it across a range of tasks.