Backward Linear Prediction (LN)
The first few data points in an FID can become corrupted due to such things as probe ring-down. This results in an undulating baseline. While this can be corrected using NUTS tools such as polynomial correction of the spectrum (FB), it can also be corrected using backward linear prediction.
See also: Forward Linear Prediction
Note that linear prediction is not included in NUTS Lite.
Looking at the sinusoidal fluctuation of this FID, it is easy to see that the first point is not at the "correct" position.
After FT of the above FID, the baseline clearly needs to be corrected.
Taking the same FID, type LN (not available on menus) to bring up this screen for setting the necessary parameters.
The values shown here are the default values and are usually appropriate for baseline correction.
We are going to correct the first 2 data points, based on the periodicity of the next 64 points. Most linear prediction algorithms need to know how many different frequencies are present in the spectrum, which is usually not known. By setting this value to -1, NUTS will determine this value.
We have chosen backward preciction.
After executing the backward linear prediction, the first data point is clearly in the correct place.
After FT, the baseline is much flatter.
Description of Linear Prediction
Last updated: 11/3/97.