 |
NEXRAD
Weather Radar Expert Speaks Out
An
Analysis of the So-called "Anomalous" Radar Images on
the Enterprise Mission Website
RE: Radar Images at http://www.enterprisemission.com/truth1.htm (The Truth is Out There - and We Found It.)
See also: Turret
Peak Event – Radar Analysis
One of the reasons I was reluctant at first to
go into great detail about this matter is the complexity of the
WSR-88D radar and the various weather and internal systems
phenomena that can cause weird looking things to appear. I knew
that if I came out with only partial explanations that the things
I couldn't fully explain would be considered "beyond the
explanation of scientific experts" and used to bolster the
points that the above web page tries to make. The risk for that
is still present, however, as I am skilled at weather radar
interpretation only, which does not necessarily make me any kind
of expert on the exact nature of patterns which could be
non-meteorological. Still, I should be able to eliminate many of
those patterns from the "unexplainable" column. Suffice
it to say that I've outright seen almost all of those radar
patterns before. In cases where I haven't, I believe I have seen
iterations of them and can therefore provide some plausible
insight.
I want to make it clear that I am not trying to
debunk that grander theory behind what the producers of the web
page are getting at. I have listened to Art Bell's show on numerous occasions, and have to say that some
of the findings there are too eerily coincidental to be pure
bunk. I am not an expert on any of the subjects and phenomena
discussed on the show and so can't comment on them with any
scientific plausibility. However, I am an expert at weather radar
interpretation and so will speak my piece accordingly. I will
also add, for those of you who might think I'm another government
plant to play spoiler, that I myself have seen unexplained
phenomena in the clear night skies of New Mexico. I do not
comment to debunk, merely to keep the conversation within the
realm of the unbiased scientific process. I suppose you could say
I'd like to play Scully to the above web page's apparent Mulder!
Complexities of Meteorological Radar
Systems
Prior to about 1990, the vast majority of radar
data in North America came from simple analogue radar systems
that were developed in the mid 1950s and updated in the mid
1970s. Simple blasts of EM energy were sent out, and the strength
and "rebound time" of the return pulses were analyzed
to determine target strength and range. "Anomalous"
patterns were frequent with these systems, owing usually to
military chaff drops, nearby mountain ranges, and atmospheric
density discontinuities which caused abnormal refraction of the
radar beam. NWS Meteorologists at the time played 100 % of the
role in differentiating between "anomalous" and real
returns. (By the way, "anomalous" in weather radar
speak simply means that the returns are not caused by
precipitation or clouds. Thus, returns from mountain ranges,
though completely valid and scientifically explainable, would be
considered anomalous.)
With the implementation of the WSR-88D NEXRAD system in the
early and mid 1990s, this changed completely. The new systems are
much more powerful, much more sensitive and almost completely
automatic. The new systems attempt to keep anomalous returns
(called AP for Anomalous Propagation of the radar beam) to a
minimum by using complex algorithms and filtering techniques.
These same algorithms, however, can also block out or impair
detection of meteorological targets if not applied properly,
causing some very unnatural looking patterns. The radars can also
measure the velocity of radar returns using the Doppler theory.
Because of this, full color images of target velocity (measured
inbound and outbound along each radial) are available for every
elevation slice. This is what allows us to actually see and
measure the "spin" within a tornado, as well as see
it's shape! Cool, huh!?
So now we have a huge, new nationwide radar
network with incredible capability. Things began to be seen on
radar that had never been seen before. With time, they have
almost all been fully explained. Swarms of bees, clouds of dust,
migrating birds, airliners, highway traffic, subtle atmospheric
boundaries (fronts), inversions, hail "spikes", cloud
layers, virga (evaporating rainfall) and sunrise/sunset returns
are just a few examples of phenomena that were initially
unexplained. Those are just the things that have made it through
the signal processing intact. There is an array of other AP
phenomena that occur purely due to the automatic processing of
the return radar signal. These can occur due to noise/feedback
within the system and improper application of clutter suppression
and attenuation applications/algorithms, among other things.
So now we have two confirmed sources of AP to
start with: actual non-meteorological physical phenomena and
errors inserted by the radar system itself. Then you have to
consider what you're looking at on these Internet images. Every
network radar operates independently on a 5 to 10 minute cycle in
which in an entire volumetric sample of the atmosphere is
obtained within 248 nautical miles of the radar site. This is a
called a "volume scan". A series of providers (WSI/Intellicast is one) take all these individual pictures and run
their own algorithms on them to make a composite picture. Since a
lot of radars overlap, I can only wonder at how the providers
handle conflicting data between radars. It gets even more complex
when you attempt to produce a "composite loop" of radar
images. Imagine over 150 independent, complex systems meshed
together by a computer application to produce a regional radar
loop. (If you ever see a loop on the Weather
Channel, you will note that the picture
seems to shimmy and move irregularly around each radar. That is
the result.) Curiously, regional loops from Intellicast on the Internet don't seem to do that nearly as much.
Perhaps that company has even further processed the signal to
make it seem like all the radars are acting as one? As you can
see, by the time you get your Internet or TV composite or loop,
chaos has reigned supreme! You've taken a signal with inherent
AP, processed it through an individual radar that may then add
(or remove) more AP to the equation, and from there sent it into
the hands of private companies who do god-knows-what to it to
make it look prettier for us!!! So, on the one hand, this would
make analysis of any of the images on the web page dubious at
best. On the other hand, the distinct patterns in some of the
images should be associated with something, as they either made
it through several processing phases unscathed or were modified
as a whole right before dissemination to the world. That last
part is the unknown to me here. It seems necessary to examine
further if any of the patterns shown could have come from final
processing on the provider's end before final display of the
product.
Analysis of Images at http://www.enterprisemission.com/truth1.htm as of 1/8/99...
Note: The following analyses were made without
distinct knowledge of local weather conditions, topography,
internal radar systems calibrations and settings or even simple
"knowns" such as the exact nature of the image
displayed. The WSR-88D produces at least a dozen separate
products for as many as 14 separate elevation slices every 5 to
10 minutes. That's on the order of 1,000 distinct radar generated
products every hour, but we're seeing only one per hour with the
images on this page. Looking at just one unknown image for a
detailed analysis is like having a mechanic listen to your car
idle to assess engine trouble, without ever opening the hood. The
composite maps shown on this web page and by most media outlets
are usually either Base Reflectivity (0.5 degree beam elevation)
or Composite Reflectivity (maximum reflectivity at each data
point in entire volume scan). I will use the information provided
with the image, as well as my personal knowledge of the area, to
provide the most detailed scientific analysis that I possibly can
given those resources.
*First Image - US Composite Image (no clue on
specifics)
Date/Time: Dec. 7, 1998 / Time Unknown
Caption: "Note small partial ring over central Arizona"
Analysis: When I clicked on the picture to blow
it up I got the 404 error message. No conclusion.
*Second and Third Images - US Composite Image
with blowup over northwest Washington
(Looks like Base Reflectivity)
Date/Time: Nov. 12, 1998 / 0906 GMT (106 AM PST)
Caption: "Note partial radial emanating from Bangor,
Washington"
Analysis: To tell the truth, I'm not quite sure
what the caption here is referring to. (It would probably be
better if I sat down with someone to have them explain to me in
detail what it was they thought was unusual.) I'm not sure
exactly where the radar is here either, but it appears to be
somewhere northwest of Seattle and Everett, maybe towards Port
Townsend? (See: NEXRAD Map) Considering the
location, time of day, time of year and the other radar returns
from the west coast, I'd guess that there was layer of marine air
seeping eastward into the west coast states. This happens quite
frequently, especially in northwest Washington. What I'm seeing
here is a typical pattern depicted on a WSR-88D when there is a
marine inversion present. The radar beam, which normally slopes
upward logarithmically from its origin due to atmospheric
refraction, is likely being "reflected" off the base of
the inversion/moisture discontinuity (and hitting the ground)
when it should be shooting off into space. Since the internal
settings of the radar assume normal refraction of the radar beam,
these returns are AP since the beam refraction is not normal.
What I believe you're seeing is a combination of returns from low
clouds, fog, drizzle and topography. The solid, red returns to
the east of the apparent radar site are most likely the west
slopes of the Cascade range. Under normal beam refraction, most
of the beam would probably pass over the Cascades and they
probably wouldn't make such a dramatic appearance. Now, of
course, the question is: Why then is there a distinct lack of
returns to the southwest of the radar site? The most likely
explanation for this is that the Olympic Mountains are simply
blocking the radar beam. Also, since the Olympics are obviously a
constant factor in radar images from this radar, it's likely that
the clutter suppression algorithm has been enabled to remove
false radar returns that are created by them. This might also
explain why the Cascades show up much more than the Olympics in
this image. The Olympics are considered more heavily in the
clutter suppression scheme under normal beam refraction
conditions than the Cascades.
Short and Sweet: This image shows conditions
know to be associated with a marine inversion, that is cooler and
more moist air at the surface with warmer and drier air aloft.
Under these conditions, AP is always encountered in the form of
spurious ground returns and enhanced topographic features. A
combination of beam blockage and clutter suppression algorithms
are most likely responsible for the lack of returns to the
southwest of the radar site.
*Fourth Image - South-central US Composite Loop
(most likely .5 degree slice)
Date/Time: Dec. 20, 1998 / 2007-2307 GMT (207 - 507 PM CST)
Caption: "Dissipating radial pattern over
Brownsville, Texas. A "Radar anomaly?" Or ...?"
Analysis: Indeed, at first glance at this would
appear to show a dispersion of EM energy or some type of
reflector in a doughnut shaped area with a radius about 95
nautical miles out from the radar. At a .5 degree beam angle,
that would place the phenomena somewhere between 10,000 and
15,000 feet. As to the apparent cause for the returns, it appears
unlikely that is any kind of usual meteorological target. This is
especially true because the adjacent radars at Corpus Cristi,
Austin, and (I think) Laredo show no signs of it. The distinct
circular appearance of the radar returns make it obvious that
this image was detected and/or produced solely by the Brownsville
radar. If there had been actual scattering bodies aloft over
southern Texas at this time, some of them would have certainly
been picked up by the adjacent radars, and the pattern would not
nearly be so perfectly circular. Therefore, I must deduce that
this was a glitch within the processing system of the Brownsville
radar. Similar patterns have occurred before when there have been
calibration errors, system noise, etc., and they are well
documented. Either this was the case here, or every other radar
system in southern Texas at this time was not operational!
Short and Sweet: This image most likely shows an
internal glitch or processing anomaly distinct to the Brownsville
radar during that particular time frame. The absence of returns
from any of the adjacent radars and the purely circular pattern
further support this conclusion, assuming that the other south
Texas radars were in operation at the time.
*Fifth Image - North-Central US Composite Loop
(most likely .5 degree slice)
Date/Time: Dec. 21, 1998 / 0706-1106 GMT (106 - 506 PM CST)
Caption: Partial radial ring over South Dakota. Do
"anomalies" only occur over 260° of a radar sweep?
Analysis: The answer is...YES! If that anomaly
is being produced by the radar itself, then almost anything goes.
You can get spikes, circles, partial radials or even a radar
screen completely filled when there are clear skies. That leads
to another of the unknowns I mentioned in my opening words. One
of the things I see in this picture and the one from Brownsville
is that there appeared to be fair weather at both locations when
these loops were captured. What people unfamiliar with the system
might not know is that this is when maintenance is routinely
performed on these systems by staff electronics technicians. It
is routine to have a bad or scan or two either at the beginning
or end of the maintenance period, if not several. Of course, it
is unlikely that this radar was being worked on at 2 am in the
morning, so that was most likely not the case here! I have to say
that I cannot in any way be sure about what exactly caused this
image. I can say that I've seen this pattern many times before
under different circumstances and in different locations.
Possible reasons include: bad volume scan, change in detection
mode (there are more sensitive and less sensitive modes for
different applications), internal systems errors, etc.
Short and Sweet: Similar to the Brownsville
image, this one most likely shows an internal glitch or
processing anomaly distinct to the radar (Rapid City, SD?). The
fact that it occurred in only one volume scan and that similar
patterns have been proven to be attributed to internal processing
problems further supports this conclusion.
*Sixth Image - National Composite Image (Can't
tell if Base or Composite Reflectivity)
*Seventh and Eighth images - Midwest and Northwest Composite
Loops at same time frame as sixth image (These appear to be Base Reflectivity)
Date/Time: Dec. 23, 1998 / 1306-1706 GMT (706 - 1106 AM CST / 506
- 906 AM PST)
Captions:
6) Parallel streaks nationwide on the 23rd. Radar chaff or
"the fleet" dropping by?
7) Midwest close-up of the images above.
8) And over the Pacific northwest.
Analysis: You're going to like this one! First
off, I see nothing unusual at all with the National shot (sixth
image), so that about wraps that up! None of the streaks
exhibited in the following two pictures are on this one, and
you'll soon know why. Notice in both pictures 7 and 8 how the
streaks mysteriously appear from east to west across the country
during the time frame 1306 - 1706 GMT. These times correspond
directly to the hours just before and after sunrise from east to
west. What you're seeing here is the glory of the rising sun as
it blasts its low-angle EM energy down the pipes of the network
radar system! This again is another frequent happening in the
world of weather radar. During this time of year, it usually
happens between 7 and 8 AM LST as the sun rises and corresponds
directly to the angle of the rotating radar antennae for just an
instant. The radar is pulsing out its power into clear air and
getting few or no returns, when the sun suddenly slams it with a
full dose of energy back down the radial. Seen on a single radar,
it's just a single frame in the loop that flashes by each morning
and is ignored. Put all together in these regional composite
loops, it does look pretty freakish! If you check the times on
the loops relative to when the flash occurs at each radar site,
then do the math to change GMT to LST, you'll see that they ALL
occurred at or between 7 and 8 AM LST. And yes, the same thing
happens when the sun sets as well. Similar streaks developing
from east to west on a radar loop just before sunset, but
oriented westward from the radars, would indicate this. Oh, and
the reason there were no streaks on the national image from 1 PM
GMT (1306 GMT) is because that correlates to 7 AM CST, 6 AM MST
and 5 AM PST...just at or before sunrise in the central time zone
and well before sunrise in the mountain and pacific zones.
Short and Sweet: These images depict the rising
of the sun as seen by the WSR-88D system when skies are mainly
clear. This happens when the sun's energy is instantaneously
detected as it rises to an angle coincident to that of the radar
antennae. The fact that this phenomena was observed
"flashing" from east to west across the country at
around the time of sunrise lends strong support to this
conclusion. Instances of this phenomena have been recorded at
every single radar location in the US.
*Ninth Image - Central US Composite (most
likely .5 degree slice)
Date/Time: Nov. 10, 1998 / 2007 GMT (207 PM CST)
Caption: Streaks criss crossing at 90° angles. Do meteors do
this? (no). And
note the stationary "streak." A "transient radar
anomaly" that isn't transient?
Analysis: The only thing I see at all unusual
with this image is the strange west-east oriented streak over
southern Missouri, seemingly emanating from the radar over
southwest Missouri. (Those thin north-south streaks over OK are
obviously bogus and probably associated with the post-processing
of the data by Unisys.) The southern Missouri return looks like
it could be a sun spike, but that's not possible at 2 in the
afternoon. The only thing I can even draw on to make a conclusion
is the fact that a similar line appeared in the same area in the
seventh image (loop) above. The nature of the return in that
loop, a weird looking line with precipitation obviously moving
through it, lends itself to processing problems with Unisys in
trying to provide that composite loop. In this image, it could
just be a bad radial in a scan, but it would be most helpful to
be able to see the images from before and after that one.
Short and Sweet: No concrete conclusion possible
due to lack of data. My best educated guess would be that it's
either a post-processing problem with Unisys or a simple bad
radial scan from the radar itself.
*Eleventh Image - Northeast US Composite Loop
(most likely .5 degree slice)
Date/Time: Nov. 11, 1998 / 0915-1008 GMT (415 - 508 AM EST)
Caption: The unmoving Michigan/Iowa storms and "radar
blanks." If this is just "ground clutter," why do the clouds circulate around
the circles?
Analysis: This is a pretty clear case of either
clutter suppression gone mad or poor post-processing by the radar
image vendor. Again, without knowing the settings of those radars
at that particular time, I can't be sure. However, it is known
and has been frequently seen before. And the clouds are not
"circulating around the circles." What you're seeing is
lake effect snow in the wake of a cold upper level low centered
somewhere over Quebec in this loop. The actual echoes are moving
east-southeast in off the lakes, but are dissipating as they move
further inland and constantly re-forming back over the lakes.
Hence, you're seeing movement within the echo pattern, but the
area of precipitation as a whole is pretty much stationary. The
precip over northwest Illinois looks a little funky. I'd surmise
that it was actually snowing across a good portion of northern
Indiana and Illinois at this time, but it appears that at least
one radar there is out of commission as we're only seeing partial
returns. Impossible to say if these data gaps were caused by the
radars themselves or by vendor post-processing.
Short and Sweet: Data gaps in this loop were
mostly likely caused by a combination of misapplication of radar
clutter suppression algorithms, problems with vendor
post-processing and a radar outage over portions of Indiana and
Illinois. While no concrete evidence can be shown, a comparison
of the images from the individual radars against this loop and
investigation into the status of the radars in Illinois and
Indiana would be enlightening.
*Twelfth Eleventh Image - .5 Degree Loop
(apparently composite) centered on Greenville, SC
Date/Time: Aug. 5, 1998 / 1038 - 1147 GMT (638 - 747 AM EDT)
Caption: A "direct hit" on Greenville, South Carolina
during the Persied Meteor shower. But of what, by what?
Analysis: First off, this is probably a
composite loop owing to the circular echo patterns around the
adjacent radars at Columbia, SC, Charlotte, NC and the site in
far northeast Georgia. (Also, note the sun spike caused by the
rising sun!) The nature of the circular patterns and the fact
that this is a .5 degree slice indicates that we're seeing
something rather uniform lifting from very near the ground to
several thousand feet into the air before dissipating. (Something
exploding aloft would create very wide rings which would either
contract around the radars with time or be swept away by
prevailing winds.) Given the time of year and day, I believe
we're either seeing the lifting and dissipation of low clouds and
some fog as the sun rises (common in this area in August). This
pattern isn't occurring directly around the
Greenville-Spartanburg radar probably because the fog/low clouds
may have developed only on the Piedmont and lower elevations,
while the radar (I think) is at 1500 - 2000 feet elevation.
Short and Sweet: Loop shows lifting and
dissipation of low level phenomena with the sunrise. It is most
likely low clouds and fog. This pattern is the exact opposite of
the one you'd see if something were exploding aloft and falling
to the ground.
Update on Analysis of this Loop - 1/24/99: I was
contacted by a noted biologist who informed me that my analysis
of this loop was incorrect! What is seen here are not clouds and
low fog lifting uniformly, but thousands of Purple Martins taking
off at sunrise. (But the sunrise spike is still real!) This kind
of thing has been documented on numerous occasions. After doing
some checking, I realized that loop IS only the GSP image. The
other locations, which I thought might have been radar sites,
were instead roosting sites for the birds. They must fan out in
those patterns after the mass take-off! (See: http://www.ametsoc.org/ams/supplement/waf/13.2/Gauth/gauth.htm)
Overall Conclusion: The images
on http://www.enterprisemission.com/truth1.htm show no glaring signs of anything at all unusual or
unexplainable. There are a few that are not fully explainable by
me, but I think I could nail them all down given more data to
work with.
Well, that about wraps it up for this version!
I plan to further examine another loop on this page which claims
to focus on Turret Mt. in Arizona (HERE is the
analysis). I'll ship that separately when I can get to it. I hope
this effort will inspire those who are looking to do so with a
bit more of a scientific method in the future. There is a whole
heck of a lot out there that we don't know, but let us not
disregard what we do know for the sake of a cause.
Questions and comments are more than welcome. Let us enlighten
each other!
Charlie Maxwell
NWS, New Mexico
Email: Chasman0@aol.com
DISCLAIMER: The opinions and analyses offered
by me are in no way those of the NWS or any other part of the US
government. (standard disclaimer).
Site Designed by
