Integration
Data Smoothing

Data smoothing applies a digital filter to sample data to reduce signal noise and helps improve chromatogram appearance and reproducibility of peak baselines. Data smoothing is performed in the integration window. (For MS chromatograms, data smoothing is defined in the PGM File or during Mass Trace extraction.) After smoothing, the smoothed chromatogram is displayed overlaid over the original chromatogram. The original sample data file is not altered and the smoothed data file is stored separately.

Filter Types

The Savitzky-Golay filter smoothes to least-squares fit, using a weighting function based on second-degree and third-degree polynomials. Savitzky-Golay smoothing is useful for reducing high-frequency noise of a data set that is continuous (such as a chromatogram) without significantly degrading the underlying signal.

The Moving Average (= Boxcar) filter is a simple algorithm that produces a set of output values in which each output value is equal to the average of n points centered around the corresponding input value, where n represents the filter size. Because the Moving Average filter equally weights each point, its ability to discriminate between noise and signal is limited.

The Olympic filter is very similar to the Moving Average filter, except that the maximum and minimum points of each input data set are rejected before the average is calculated. This provides better rejection of impulse noise (spikes) than the moving average filter.

In addition, the Gaussian filter is available for acquiring MS chromatograms and extracting a mass trace (in the Mass Spectra window). This filter applies the Gaussian distribution for chromatogram smoothing.

Filter Size

Filter size is the number of input data points used to generate each output data point. The filter size is an odd number between 5 and 999. Use a narrow filter size if the desired peaks are narrow, and a wider filter size for wider peaks. As a rule of thumb, select a filter size that approximately equals the peak's half width. Note that too narrow a filter results in insufficient smoothing while too wide a filter can lead to distorted data.

 Tip: 

Distortion of data during data smoothing mainly affects the peak height. Therefore, it is generally better to evaluate smoothed chromatograms by area rather than height.

Iterations

If a filter is applied several times, by far the highest smoothing result (>95%) is achieved when the filter is applied the first time. Thus, normally a single smoothing step is sufficient. However, applying a narrower filter multiple times often provides improved noise reduction without the signal degradation that can occur when using a wider filter size. This requires additional processing time, however, so a wider filter size may be preferable if its results are acceptable.

For additional details, refer to  Working with Chromatograms  Performing Data Smoothing.

For MS Chromatograms, refer to Using Mass Spectrometers  Extracting Mass Traces afterward.