Subset a data set based on first non-NA values

Question

I have a data frame like this:

 Year S1  S2  S3
 1699 1   NA  NA
 1700 5   23  5
 1701 6   1   6
 1702 7   13  9

I want to keep only those co