Some modification of it is needed when stations do not have enough data in the reference period, and as E.M.Smith describes, GISS calculates
anomalies at grid points, which involves a bit of local aggregating.
But indeed, as I said above, GISS still uses the Hansen / Lebedeff method of calculating
anomalies at grid points.
Second, and here again I must apologize, the NASA GISS record uses a different method of forming
the anomalies at grid points.