Академический Документы
Профессиональный Документы
Культура Документы
distorted BTO
Paul Smithson
Work in progress, comment welcome version 20
April 2016
Original theory
www.findMH370.com
There should have been debris and it should have been detectable by
satellite
All but one sat detection (16-28 March) located in single zone comprising
2% of [then] search area [in fact, just outside it suggesting wider area
was scanned by satellites]
The only reason that objects at 45S have been dismissed out of
hand (without, even, the need to provide an alternative
explanation for their existence) is that they dont match the BTO.
The BTO adherents say we must prove them wrong for alternate
theory to be considered. The reverse, apparently, does not apply.
Nobody has offered a credible explanation for existence of >400
reflective objects, many >10m, clustered at 45S, sighted by
separate hi-res multispectral satellites in an area close to
expected end point.
If we didnt have the BTO predictions, would this not be the first
place to look? It is the best and only candidate for an MH370
debris field.
18000
16000
predict
observe
14000
12000
10000
18:00:00
3500
20:24:00
22:48:00
1:12:00
3000
2500
2000
1500
1000
18:00:00
20:24:00
22:48:00
1:12:00
Similar shape
curve but
BTO predicted
>> than BTO
observed
Error magnitude
~1300 to ~3300
microsecs.
Errors smaller in
middle, greater
at the each end
This example for turnback
2 mins after IGARI. Passes
over WMKK 17:59:16 HDG
208.38
IGARI +0s
19000
18000
17000
R = 1
2241
0011
16000
0019
2141
1825
15000
1828
BTO Observed
2041
R2 polynomial
correlation of
0.99983
14000
13000
12000
1941
11000
10000
12000
17000
22000
BTO predicted
27000
Time
Observed
Model
18:25:30
18:27:00
18:28:15
19:41:00
20:41:00
21:41:30
22:41:15
00:11:00
00:19:30
12,520
12,520
12,480
11,500
11,740
12,780
14,540
18,040
18,400
11,500
18,400
13,836
12,514
12,487
12,465
11,663
11,788
12,691
14,390
18,084
18,438
11,663
18,438
13,836
Minimum
Maximum
Mean
Residual
6
33
15
-163
-48
89
150
-44
-38
-163
150
0
Time
18:25:3
0
18:27:0
0
18:28:1
5
19:41:0
0
20:41:0
0
21:41:3
0
22:41:1
5
00:11:0
0
Observed
Model
Residual
12,520 12,535
-15
12,520 12,504
16
12,480 12,479
11,500 11,560
-60
11,740 11,705
35
12,780 12,733
47
14,540 14,562
-22
18,040 18,069
-29
127,000
126,000
126,000
R = 1
125,000
125,000
124,000
AES from obseved BTO values
124,000
123,000
122,000
123,000
122,000
121,000
121,000
120,000
120,000
120,000
125,000
120,000
130,000
123,102
123,101
123,081
122,569
122,704
123,267
124,214
126,099
126,293
model
123,102
123,088
123,076
122,650
122,725
123,218
124,135
126,121
126,313
R = 1
Residual
0
14
5
-81
-21
48
78
-22
-20
125,000
130,000
observed
123,102
123,101
123,081
122,569
122,704
123,267
124,214
126,099
126,293
model
123,112
123,096
123,082
122,598
122,684
123,241
124,224
126,113
126,279
residual
10
-6
2
29
-19
-25
10
15
-14
Is it a fluke [1]?
The quality of fit between distorted BTO and path model
is as good or superior to - most BTO-derived path models
(cf sk999s summary of residuals various path models)
Those path models were specifically generated to achieve
best fit with the BTO data and speed, heading, timing finetuned accordingly. This path was not
This path starts from a totally different place, with magnetic
heading and no turns or tweaks to improve the fit
As any BTO path-modeler knows, it is difficult to obtain a
good BTO fit across all data points without changing
speed/heading. Errors/residuals are extremely sensitive to
small changes in heading or speed
The p-value on the linear correlation is 1.53E-11. p-value on
polynomial correlation is smaller still
Is it a fluke [2]?
Distorted How?
Four broad logical options:
1. Method error. BTO has been misinterpreted from
the outset and does not represent vector
distance/time delay for round trip LES-sat-AES
2. Distance error. Vector distances/TD are faithfully
represented by BTO. But bias constant k and/or
vector distances differ from assumption
3. Algorithm disrupted. BTO normally faithfully
represents vector distances but stopped doing so
after disappearance. Something has screwed up
the algorithm output
4. Data changed. Somebody changed the data
Sequence of events
Factual
17:19:26 KL-ATCC handoff
17:19:30 Good night"
Scheduled to begin
17:20:15 IGARI turn
17:20:36 Mode S signal loss
Scheduled end IGARI
17:20:47 turn
Assumption/ration
ale
Later turn poor BTO
correlation
Tighter turn (>25) not
poss.
Turn before IGARI not seen
(and too quick)
Inference
Diversion initiated during the IGARI
turn, 17:20:15 17:20:47
Emergency commenced after good
night but before IGARI turn
Explains failure to check in with HCM
(working the problem)
Power lost (or removed) => Mode S
signal loss
Pilot incapacitation shortly after
diversion (no comms.)
Great circle to WMKK, ETA 17:56:05;
route discontinuity, magnetic trackhold 207.0(M) to fuel exhaustion
Second engine flame out 00:17:30,
EP 44.90S, 89.41E
Testable predictions
Evidence of this path should be obtainable in
multiple radar recordings [assuming they exist]
Reverse drift model of objects sighted ~45S will
have origin nr ~44.9S, 89.4E
Bifurcated debris drift: some E (circumpolar);
some N then W (SIO gyre)
Plane wreckage will be found very close to end
of track predicted (207.0 0.1)
Somebody will figure out how the BTO got
distorted
Radar
Name
1 Bukit Puteri
2 Bukit Ibam
3 Western Hill
4 Bukit Lunchu
5 Khok Muang
6 Singapore
7 Bukit Nenas
8 Pekanbaru
9 WMKK
10 en route
11 WMSA
time
17:23:06
17:23:11
17:31:25
17:37:04
17:23:06
17:33:23
17:35:45
18:11:08
17:48:47
17:45:29
17:47:08
lat
6.79
6.78
6.67
5.07
6.79
5.53
5.24
0.88
3.64
4.04
3.84
End
long
103.78
103.77
103.72
102.90
103.78
103.13
102.98
100.77
102.16
102.36
102.26
time
17:59:59
18:14:46
18:10:19
18:16:36
17:54:55
18:22:58
18:33:47
18:24:04
18:03:20
18:00:01
18:00:57
lat
2.26
0.43
1.89
0.21
2.88
-0.58
-1.92
-0.72
1.85
2.25
2.14
long
101.46
100.54
101.27
100.43
101.77
100.04
99.37
99.97
101.25
101.45
101.39
Duration
Distance
NM
Time
304.3
425.5
320.9
326.2
262.5
409.0
478.8
106.7
120.0
119.9
114.0
00:36:53
00:51:35
00:38:54
00:39:32
00:31:49
00:49:35
00:58:02
00:12:56
00:14:33
00:14:32
00:13:49
a/c
HDG(T)
207.2
207.1
207.1
206.9
207.2
206.9
206.8
206.5
206.9
207.0
206.9
But drift
from 45S
area to
Reunion is
feasible
White tracks of
model
particles
arriving
Reunion July
2015
Copyright CSIRO
http://www.marine.csiro.au/~griffin/MH370/O
M_af_MH370_IO_tp3sl15ds_splash3932/20140320
.
http://www.cmar.csiro.a
u/ofam1/om/OM_af_MH
370_IO_tp3sl15dsr_Reu
nion3/OM_af_MH370_IO
SUPPLEMENTARY SLIDES
Time/HDG at WMKK
Time over WMKK and Final Heading
211.0
IGARI +240s
210.0
209.91
209.13
209.0
208.0
208.35
207.60
207.0 206.97
206.0
205.0
IGARI +0s
Geometry of
turnback
produces
linear
relationship
between time
at WMKK and
the HDG
SENSITIVITY ANALYSIS
IGARI +60s
19000
18000
17000
19000
18000
R = 1
17000
16000
16000
15000
15000
14000
14000
13000
13000
12000
12000
11000
11000
10000
12000
14000
16000
18000
20000
22000
24000
10000
12000
IGARI +120s
17000
16000
18000
20000
22000
24000
19000
18000
R = 1
17000
16000
16000
15000
15000
14000
14000
13000
13000
12000
12000
11000
11000
10000
12000
14000
IGARI + 180s
19000
18000
R = 1
14000
16000
18000
20000
22000
24000
10000
12000
R = 0.99
14000
16000
18000
20000
The difference between IGARI+0s and +180s is diversion to KLIA initiated 180 seconds later.
+0 overflies WMKK at 17:56:04 HDG207.0(M). +180s overflies WMKK at 18:01:06 HDG
209.2(M)
22000
Sensitivity test
Array of R2 values
206.6
206.7
206.8
206.9
207 207.1
207.2
207.3
207.4
207.5
207.6
17:53:0
0 99,904 99,915 99,925 99,934 99,943 99,951 99,957 99,964 99,968 99,973 99,976
17:53:3
0 99,922 99,931 99,940 99,948 99,955 99,962 99,967 99,972 99,975 99,978 99,979
17:54:0
0 99,938 99,946 99,953 99,960 99,966 99,971 99,980 99,978 99,980 99,981 99,981
17:54:3
0 99,951 99,958 99,964 99,969 99,974 99,977 99,979 99,981 99,981 99,981 99,979
17:55:0
0 99,962 99,968 99,973 99,976 99,979 99,981 99,982 99,982 99,981 99,978 99,975
17:55:3
0 99,971 99,975 99,979 99,981 99,982 99,983 99,982 99,980 99,977 99,973 99,968
17:56:0
0 99,978 99,980 99,982 99,983 99,983 99,981 99,979 99,975 99,970 99,965 99,958
VERY
narrow band of heading/timing combinations that
17:56:3
0 99,982 99,983
99,980 99,977 99,973 99,968 99,961 99,953 99,944
produce
max 99,983
R-sq 99,982
association.
17:57:0
For
any 99,983
given99,982
HDG,99,979
timing
must
match
30s
to achieve
0 99,984
99,975
99,970
99,964+/99,956
99,948
99,938 99,927
17:57:3
max R-sq
0 99,983 99,980 99,977 99,972 99,966 99,959 99,951 99,942 99,932 99,920 99,907
For
17:58:0 any given Timing, HDG must match +/- 0.1 degree
0 99,978 99,974
99,969 99,962 falls
99,955away
99,946quickly
99,936 99,925
99,913
99,899
99,884
Strength
of association
if you
move
away
17:58:3
>60
seconds
or >0.2
degrees
from99,918
the sweet
spot
0 99,971
99,965 99,958
99,950
99,940 99,930
99,905 99,890
99,875 99,858
211.0
207.4
210.0
207.3
207.2
209.0
207.1
208.0
207.0
206.9
207.0
206.8
206.7
206.0
206.6
205.0
206.5
0.74618055555555607 0.74713541666666716 0.74809027777777815
207.2
Time/HDG
from turnback
model
207.0
206.8
Time/HDG
with max R-sq
206.6
206.4
0.74618055555555607
0.74713541666666716
0.74809027777777815
In other words
The exact time over WMKK and HDG
derived from model of diversion turnback at IGARI without any reference
to BTO data, just happens to produce
perfect correlation with BTO data.
Recalling sensitivity analysis, to do
this the timing must be correct to
30 seconds and HDG 0.1
4. Radar
Sacred cows
No intent to bash ISAT analysts. IG has done the
best data-driven analysis out there. There was
every reason to expect them to be right.
Growing evidence that for some reason its
predictions are not being fulfilled
Difficult (publicly) to abandon position. Difficult
(psychologically) to drop assumptions when you
have convinced yourself that everything else lines
up with it.
That does not make it sensible to come up with
ever-more-unlikely explanations for unfulfilled
predictions. Better to put the spotlight on the
fundamental premise.