I dont know if this will work or not,
first calculate cum. infiltration for each storm using the CN method and construct infiltration time series. Then look for cross correlation between base flow time series and infiltration time series and come up with a statistical model.
i think this will...