From wlee@stout.atd.ucar.edu  Thu Mar  7 16:31:11 1996
From: wlee@stout.atd.ucar.edu (Wen Chau Lee)

Subject: VORTEX message #5 & #6


                              VORTEX Message #5


Dear VORTEX ELDORA data users:

This message follows up the unfolding problems stated in message 1.
The following is a copy of Message 1.

************************ Message 1 **************************

A bug has been identified in calculating unfolded velocity VR from VL
and VS (velocities from individual PRFs). This error only affects 
VR when abs(VL-VS)> 31.2 (fore) and 31.866 (aft) which is caused by 
improperly handled angle differences near pi and -pi. This error affected 
velocities between VLny and VSny (nyquist velocities of large PRF and 
small PRF) or -VLny and -VSny. In the usual operation mode (Nyquist 
velocity of +-79 m/s), the PRFs are 2500 and 2000 (VLny=19.502 m/s and 
VSny=15.6 m/s for fore radar, VLny=19.916 m/s and VSny=15.933 m/s for 
aft radar). The slight differences in Nyquist velocities between fore 
and aft radars are caused by the differences in frequencies. So this 
problem occurred roughly 16<VR<19.5 and -19.5<VR<16 that created some
speckles in the data.

The solution is to identify abs(VL-VS) > 31.2 and 31.866, and recalculate 
VR from VL and VS. An editing function will be added in solo to deal with 
this problem. It will take a few days to complete this addition. Stay tune. 

This problem is different from the speckles caused by high spectrum 
width. However, the solo editing function may be able to correct part
of the high spectrum width speckles. If you know the correct Nyquist 
interval, you can recalculate VR using the appropriate equation.

**********************************************************************

The unfolding problem in paragraph 1 and 2 will be referred as problem
I while the unfolding problem in the last paragraph will be referred
as problem II. Additional editing functions have been included in
the SOLO package to deal with these unfolding problems. The details 
are discussed as follows.

The theory to unfold dual-PRT velocities is described in Loew and
Walther (1995)(27th Conf. on Radar Meteorology, p710-712). The
examples listed in their Table 1 are calculated in the phase domain
for the ratio of PRF1/PRF2=5/4. Then, the difference between the
V(PRF1) and V(PRF2) is one of nine constants in theory. However,
the difference is not a constant due to the non-zero spectral width
in the real data. Therefore, each region (in table 1) is identified
from the phase difference (column 2) which falls into a certain 
region (eg -pi/4 < delta < pi/4).  Unfolding problem I resulted from
improperly handling the last interval. The correct interval should
be 7/4 pi < delta < 9/4 pi. However, 7/4 pi < delta < 2pi was used
in VORTEX (same situation happened in the negative side). So, 
velocities with VL-VS exceeds +- 2pi were put in the wrong intervals.
The example given in Message 1 is using PRFs 2500 and 2000. This problem
exists for other PRFs also (eg, 3500 and 2800). Fortunately, the solution
is the same.

Only one of the actions needs to be taken.

SOLUTIONS:     
 
I. In translator script, use

   setenv DERIVED_FIELDS "VS < VQ VQ < VT VT < VG"

   instead of the normal 

   setenv DERIVED_FIELDS "VR < VT VT < VG"

   VS < VQ will trigger the translator to correct the problem I 
   discussed above and leave the VR field unchanged.

II. In SOLO editor:

    select fix-vortex-velocity in <field>
    and    remove-aircraft-motion in <field>

    The <field> can be VG. This will trigger the editor to use
    VL and VS to calculate a new velocity field then remove the
    aircraft ground speed component.


 
The unfolding problem II results from the high spectral width data.
They usually can be found inside tornados or near the storm top of 
supercells where the difference in phase between VL and VS exceeds
the difference between two catagories. At this point, visual 
examination of data continuity is the best way to identify these points. 
This can be done in the EXAMINE widget in SOLO. When the correct 
interval is identified, the new velocity can be calculated using the
formula in table 1 (third column) and the original value can be 
replaced by the user determined value.  

Note that problem I ONLY occurs in a few gates per ray. For example, in the
anvil region is quite often to find this type of folding problem.


Wen-Chau