Weather classification algorithm

The first processing step is to identify segments along the radial where attenuation occurs. The radial is broken into one or more segments where each segment is either flagged as "liquid rain" or "no weather". The liquid rain portions of the segment are used to calculate the constrained PhiDP. This step is a hydrometeor classifier.

The simple weather classification ensures that changes in PhiDP from non-meteorological targets like ground clutter, point targets, refractivity gradients, and insects do not introduce errors in the attenuation estimate. Reflectivity and differential reflectivity values outside of the segments, identified as rain, are not corrected. Also, small ice particles add only a minute amount to the overall one-way path attenuation. Therefore, once the beam propagation along the radial is higher than the freezing level, additional attenuation is no longer calculated.

The weather classification algorithm relies on three principles about hydrometeor particles:

• Hydrometeors particles tend to have a high value of the co-polar correlation coefficient, RhoHV. A mean RhoHV is computed over a defined length. If the mean value becomes lower than the threshold, range bins are flagged to not use the data. The default value used within this classifier is 0.85.
• Quantitative measurements of PhiDP tend to be highly correlated, from range bin to range bin, when hydrometeor targets are present. This gives a "smooth" looking data field versus more "noisy" fields that are likely to have non-meteorological particles. Therefore, the texture over some area is used as a discriminator. The standard deviation of PhiDP is calculated over the same range as the mean RhoHV. This is used to represent texture along the radial.
• Convective and stratiform meteorological events have some size to them; typically more than a few kilometers. This rule is used to decide when to enter a rain segment and how long this segment must be before exiting to a no weather segment.

Each range bin is uniquely evaluated as being rain or no weather using Equation 1 and Equation 2 in Dual-polarimetric attenuation correction.