The Gaussian filter used here is included with the standard GEMPAK distribution. It is a moving-average filter with a normally-distributed set of filter weights determined by the degree of filtering. The degree of filtering determines the standard deviation of the distribution used to calculate the filter weights. Larger degrees of filtering produces more damping at all wavelengths - focused at shorter wavelengths - and a larger window area for the averaging.
For the QG diagnostics shown here, a filter weight of 25 was used on the raw height, temperature, and vertical motion fields prior to computing the QG diagnostics. With a grid spacing of approximately 100 km (1.0x1.0 degree spacing), this means that 67% of the power of waves with wavelengths of 2500 km (25*dx) were eliminated. Damping of all wavelengths <=2500 km helps extract the synoptic-scale forcing from the QG diagnostic calculations.
For more information about the Gaussian filter, see Ron McTaggart-Cowan's notes about smoothing the 32-km NARR dataset.
Back to the QG Diagnostics webpage