(See other posted articles at ).
EXECUTIVE SUMMARY:
- In a similar way to modeling UFO interplanetary movements to
"explain" the rise and fall of sightings over time, we can use the
distance from putative bases on earth to predict the sighting
density of objects across e.g. the USA.
- Assuming the sighting density per state population is related to
some power of the distance from given points on earth, we can locate
a small number of points where these models are both statistically
relevant at high confidence and also explain a large part of the
state-to-state differences in sightings.
- For Triangle UFO's (as an example) the pre 2006 and post 2006 data
from the NUFORC database allows us to double check the relevant
"best models". They are found to both include regions along the
Antarctic coast as well as small areas along the Arctic coast. No
other points on earth explain all the data.
- We find the power law in the "best case" relating straight-line
("great circle") distance from the Antarctic to each US state is a
power of around 4. I.e. the states *further* from the particular
site in the Antarctic see more sightings per cap than those closer.
- This not only implies the objects are not appearing randomly over
the US like a bunch of tourists, but there appears to be a
considerable amount of loitering over the relevant destination
areas.
- The data is consistent with Triangles flying direct to certain
areas they find interesting and loitering longer on average in
regions further from their "home base" than those closer.
In an earlier post we looked at trying to match up UFO sightings of
different types against planetary movements. Given certain sightings
seemed to rise and fall depending on the distance from Earth of
various planets, we tried to create an "interplanetary flight model"
that predicted UFO sightings (across the US) as a function of starting
from different planets during or prior to a (sometimes annual) close
approach, flying at a constant speed along the relevant distance,
arriving and probabilistically being spotted by a N American.
We found there were certain sets of parameters that seemed to match up
against UFO sightings quite well, even allowing us to predict the trip
time for the chief suspects.
We can perform the same kind of calculations for suspected "bases"
here on planet dirt.
We have looked before at using "local conditions" at each point on
earth to see whether UFO sightings retain an imprint of weather or
other events occurring at points distance from the location where
various objects are sightings. And, again, that was found.
But this time we will use another property of distant places -- it's
distance from the location where the relevant object is sighted.
We might posit that the further an object has flown from its suspected
start point (aka "base") the less likely it will be observed given the
possible number of locations it can appear rises sharply as the
distance increases. (And there are also strange consequences from
large distances where tracks across the sphere of the Earth
criss-cross more direct routes creating even more possibilities).
So we can perform this kind of calculation:
for each spot on the earth presume UFO's start from there and fly
to the centre of each US state. Does the density of UFO sightings
in each state fit some kind of power-law function of the distance
from the given start point?
Well of course I wouldn't be asking the question unless I knew most of
the answer. :)
Let's look at the results for a particular type of object -- the
"Triangle". As usual I'm using the NUFORC database up to the end of
2021. This presents some idiosyncratic problems that might now
present as opportunities.
In Mar 2006 the NUFORC moved from telephone and letter reports of UFO
sightings to a web-based system. (I haven't done any guessing as to why
this happened -- but go ahead yourself :). This not only increased
the number of reports 10-fold in a matter of months it also changed
the type of reports and the types of people that submit them. This
makes using the full period of the dataset ~1900-2021 problematic.
But what we can do here is split the data into pre 2006 and post 2006
and essentially run out modeling twice. Now the differences in the 2
sections of the data are a benefit. If the same kinds of models are
found in each part it makes the results more robust -- the same type
of assumptions seem to predict the appearance of the objects in quite
different eras of UFO watching.
The distribution of Triangle UFO sightings looks like:
Sightings per mn capita
State <2006 >=2006
Alabama 3.49868 16.0528
Alaska 4.06266 43.3351
Arizona 4.10072 26.5082
Arkansas 2.35041 23.5041
California 2.12033 19.2363
Colorado 3.29877 35.5534
Connecticut 4.17724 29.5192
Delaware 2.11431 35.9433
Florida 1.52926 20.4723
Georgia 1.86004 15.6635
Hawaii 1.39704 17.4629
Idaho 2.41702 45.9234
Illinois 2.09953 20.14
Indiana 2.87023 21.9044
Iowa 4.80169 29.1303
Kansas 3.09104 24.7283
Kentucky 2.71181 26.4401
Louisiana 1.7128 16.0575
Maine 3.7613 54.1627
Maryland 1.33191 19.4792
Massachusetts 2.20769 24.1374
Michigan 3.02341 23.7841
Minnesota 2.36812 22.2239
Mississippi 2.33931 17.3777
Missouri 3.45186 31.8886
Montana 5.80861 53.2456
Nebraska 2.10949 16.8759
Nevada 2.07552 21.447
New.Hampshire 2.25461 66.8867
New.Jersey 2.00937 17.9727
New.Mexico 6.23469 30.2142
New.York 1.97012 12.7805
North.Carolina 2.58892 21.0101
North.Dakota 2.64226 33.0283
Ohio 2.66932 23.0767
Oklahoma 2.04534 22.7544
Oregon 3.72303 40.7051
Pennsylvania 2.57762 23.8235
Rhode.Island 3.78681 23.6676
South.Carolina 3.47212 30.0236
South.Dakota 2.32973 25.627
Tennessee 1.96961 23.9383
Texas 1.56539 12.7052
Utah 2.00272 25.0341
Vermont 3.19467 54.3095
Virginia 1.31218 19.5634
Washington 3.62604 40.4443
West.Virginia 2.71131 30.9089
Wisconsin 2.59905 25.644
Wyoming 3.41235 35.8296
When we take the straight line ("great circle") distance between the
enter of each state and every 10x10 degree region across Earth we find
the following "top 10" best matches between some power of the distance
and the relevant state sighting density per capita:
2006 data:
Lat Lng R2
-85 85 0.29533058
-85 95 0.29302384
-85 105 0.28977774
-85 -105 0.28635705
-85 -115 0.28605394
-85 115 0.28602370
-85 -125 0.28497686
-75 85 0.28371493
-85 -135 0.28324145
-85 65 0.28298575
2020 data:
-75 55 0.44798858
65 -125 0.44221503
75 -115 0.44117760
65 -115 0.44076530
-75 -55 0.43486416
-75 -45 0.43042487
-75 65 0.42431659
15 -95 0.42115346
-75 45 0.42062016
-75 35 0.42054592
We immediately see a big chunk of the Antarctic coast seems to be
roughly in common between the 2 different (sub-)datasets. The 2020
data has a better R2 -- meaning the model of using distance between
[continued in next message]
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
