filtering out known instrumental error from time series

Hello all,
I am working with high-frequency hydrological time series, from 
automatic loggers. These have a known instrumental error, determined 
from laboratory tests. Could anybody please advise on methods (an indeed 
R packages/functions) that I could use to remove part of this 
instrumental noise, making sure that the time series are not smoothed 
beyond the known instrumental error?
Many thanks
A. Duranel - UCL Department of Geography