Embed Size (px)
Citation preview
NASA Technical Memorandum 100707
High Altitude
Atmospheric Modeling
Alan E. Hedin
Laboratory for Atmospheres
Goddard Space Flight Center
Greenbelt, Maryland
National Aeronautics andSpace Administration
Sclentiflc and TechnicalInformation Office
1988
https://ntrs.nasa.gov/search.jsp?R=19890002769 2020-07-12T10:56:51+00:00Z
Contents
I ,
2.
2.1
2.2
Introduction ................................................
Data sets studied ...........................................
General characteristics .................. ,...............
Coverage .................................................
.
3.1
3.2
3.3
Model comparisons ............................................
Model descriptions .......................................
Residual plots.. ................................ ,........
Results ..................................................
I
I
I
4
4
4
7
8
4. Summary of model strengths and weaknesses ................... 28
5. Conclusion .................................................. 29
References ...................................................... 31
Appendix A.
Appendix B.
Appendix C.
Appendix D.
Description of coverage files ...................... 35
Description of histogram files ..................... 37
Description of data comparison and coverage plots.. 41
Data set comparison summaries ...................... 77
PRECEDING PAGE BLANK NOT _I, MED
Ill
I. Introduction
The purpose of this study is to compare existing total density data
with with several recent empirical models in order to assess current data
coverage, the accuracy of current empirical models, and where improvements may
be possible. The altitude range included in this study is 120 to 1200 km.
While techniques based on atmospheric drag effects provide total density
rather directly, the in situ mass spectrometer techniques provide densities of
individual constituents that must be su_ed to give total density, and this
was done for this study.
Several comparison studies have been made in the past (Marcos et al.,
1978; Hickman et al., 1979; Prag, 1983; Marcos, 1987) with somewhat differentdata sets and models, although there is some overlap in data and models with
the current study. A general result of previous studies was that density
models have an accuracy of around 15%, and this has not improved much over the
last 20 years. This study attempts to refine and quantify this assessment.
It produced extensive plots of data minus model residuals as a function of
various parameters to allow detailed comparisons and assessments.
This report describes the various products produced and gives some
general highlights of results. However, any study of this kind produces rich
detail and apparent anomalies that cannot be easily summarized, and interested
readers should consult the more detailed plots.
2. Data sets studied
2.1. General characteristics
The data sets utilized in the current report and areas of useful data
are summarized in Table I.
The Jacchia Drag data are total densities, deduced by the Smithsonian
Astrophysics Observatory (Jacchia and Slowey, 1965; 1970; 1972; 1975) from the
change of sateilite orbital elements as a result 6f air drag. The particular
data set used in this study was originally sent on tape by Jacchia to
F. Barlier in France, who kindly made a copy available to this author. These
data are believed to constitute the major proportion of the data used by
Jacchia in the generation of his models. The time (and thus spacial)
resolution of density derived from orbit change is generally rather coarse
with densities determined no more often than every three hours, and often one
day in contrast to tens of seconds or better for in situ data. Absolute
densities depend directly on the assumed drag coefficient. Jacchia (1977)
used 2.2 for atomic oxygen and a dependence on composition (altitude) as
specified by Cook (1976). Unfortunately, the Cook paper has a number of
possible algorithms and selectable parameters. The exact algorithm for drag
coefficients used by Jacchia is not specified in any known publication. This
situation could be important for determining drag and/or density at lower and
higher altitudes where atomic oxygen is not the major constituent, and theJacchia models (which were utilized to determine the composition dependent
drag coefficient) do not always give the correct composition. The data were
gatheredduring the decline of solar cycle 19 and the rise of solar cycle 20
over a wide range of latitudes.
Table I. Summaryof data sets.
Name Method Dates Altitude FI0.7 Max Lat NumPts
Jacchia Drag 61001-70365 244-1200 70-176Barller Drag 64033-73058 123-735 68-176OGO-6 MS 69159-71177 394-1090 108-170AE-CMESA Accel 73353-76271 129-250 70- 93AE-COSS MS 74001-75161 130-837 70- 93AE-DMESA Accel 75280-76029 140-250 72- 79AE-DOSS MS 75291-76029 140-550 72- 78Cactus Accel 75178-79021 226-600 69-197AE-E MESA Accel 75335-78032 134-250 69-130AE-E OSS MS 75343-79049 134-544 69-197AE-E NACE MS 75335-81155 134- 69-225
DE-2 NACS MS 81220-83047 199-864 128-230
MSIS Comb MS 69178-83047 135-963 69-230
Drag - indicates density by orbit decay.
90 22124
90 11647
82 292470
68 1111O5
68 133405
9O 36743
90 49875
30 106933020 7O24O
20 181990
20 242931
90 292830
90 33181
Accel - indicates density by in situ accelerometer.
MS - indicates in situ mass spectrometer.
The Barlier Drag data are total densities analogous to the Jacchia
densities but derived independently (Barlier et al., 1973) for generally
different satellites. These data are primarily useful for studying density
variations, as absolute values were normalized to Jacchia (1971) in an overall
sense. The data were taken during most of solar cycle 20 over a wide range oflatitudes.
The OGO-6 satellite (Carignan and Pinkus, 1968; Hedin et al., 1974)
was the first to obtain a long time series of fairly reliable mass spectro-
meter measurements. Perigee for most of the mission was near 400 km, and so
this satellite obtained more mass spectrometer data above 400 km than other
missions before or since. Total mass density was calculated as a summation of
measured atomic oxygen, helium, and molecular nitrogen. Measurements were
taken during the peak and decline of solar cycle 20 at all latitudes except
the extreme polar region.
The Atmospheric Explorer AE-C, -D, and -E satellites each carried an
in situ accelerome_er (MESA) (Champion and Marcos, 1973), open source mass
spectrometer (OSS) (Nier etal., 1973), a closed source mass spectrometer
(NACE) (Pelz et al., 1973) for composition measurements, and a closed source
mass spectrometer (NATE) (Spencer etal., 1973) for temperature and wind
measurements. The closed source geometry was thought to provide a better
determination of absolute density, while the more open source could better
detect reactive species. Absolute values from all three instruments are
comparable within the original target calibrations of about 15%. The NACEinstrument failed early in the AE-C flight. The entire AE-D satellite failed
after only about 3 months in operation. _All three instruments were highlysuccessful on AE_E with more data reduced for NACE. Total massdensity from
the mass spectrometers was calculated as a summation of measured atomic
oxygen, helium, argon, and molecular nitrogen. Total density from MESA was
based on a drag coefficient of 2.2 (Marcos etal., 1977). Measurements were
taken during solar cycle minimum at all latitudes and during the rise of cycle21 at low and midlatitudes.
The Castor satellite launched by CNES carried an accelerometer
(CACTUS) (Boudon and Barlier, 1979; Villain, 1980; Berger and Barlier, 1981)
for in situ density measurements. Measurements were taken during the rise of
solar cycle 21 at.low latitudes.
The Dynamics Explorer 2 satellite carried a mass spectrometer (NACS)
(Carignan etal., 1981) for in situ density and composition measurements, a
second mass spectrometer (WATS) (Spencer et al., 1981) for in situ temperature
and zonal wind measurements, and a Fabry-Perot spectrometer (FPI) (Hays et
al., 1981) for F-region temperature and meridional wind measurements. The
absolute density calibration of the mass spectrometers was not adequate due to
a failure of laboratory equipment and cannot be given much reliance. Total
mass density was calculated as a summation of measured atomic oxygen, helium,
argon, and molecular nitrogen. Measurements were taken during the peak and
decline of solar cycle 21 at all latitudes. They provide theonly set of high
solar activity high latitude mass spectrometer measurements available.
The MSIS combined density data set consists of separate data sets for
each atmospheric constituent that were formed from subsets selected from mass
spectrometer measurements on the OGO-6, AE-C, -D, -E, and DE-2 satellites
described above, as well as the ESRO-4 (Trinks and yon Zahn, 1975), San Marco-
3 (Newton et al., 1974), and AEROS-A (Spencer et al., 1974) satellites and
numerous rockets. Data subsets were selected to provide the widest possible
coverage in geographical and geophysical parameters. These combined data sets
were used in the generation of the MSIS-86 model. While there is no parallel
set of total densities because the data were not originally selected with this
goal in mind (and thus without simultaneity in time for the various
constituents), the distribution and coverage of oxygen density are assumed to
represent the overall coverage in total density that could be obtained using
the mass spectrometer data. There is a similar combined data set for
temperature combining subsets from the AE-C, -D, -E, and DE-2 satellites,
rockets, and Millstone Hill, St. Santin, Arecibo, Jicamarca, and Malvernincoherent scatter stations.
2.2. Coverage,
More details on data coverage in terms of latitude, local time, day of
year, longitude, and UT can b% found in the coverage plots and data comparison
plots described in Appendix C, which are provided on microfich for each dataset.
The coverage of the MSIS combined data set for atomic oxygen, which is
essentially the coverage for goodpredictlon of total density from the MSIS-86
model, has been examined in more detail by sorting the data into bins based on
latitude, local time, day of year, universal time (UT), daily magnetic index
(Ap), mean FI0.7, daily minus mean FI0.7, and altitude. The results of this
binning procedure are computer readable files described in Appendix A. In
general, coverage is relatively poor at high magnetic activities, at altitudesbelow 200 km and above 800 km, and at high latitudes for low solar activity.
3. Model comparisons
3.1. Model descriptions
A large number of empirical models have been developed over the years.
The genealogy of these models is illustrated in Figure I. Hickman et al.
(1979) have given a brief but thorough description of the most common models
available at that time. Five specific models are selected for comparison with
data in the current study.
MSIS-86
The MSIS-86 model (Hedin, 1987) is the latest in the series of
empirical models of neutral temperature and density in the thermosphere (above
85 km) and lower exosphere based on in situ mass spectrometer and incoherent
scatter data. The model is dependent on user-provided values of day, time
(UT), altitude, latitude, longitude, local solar time, magnetic index (Ap), a
three solar rotation average I0.7-cm radio flux (FIO.7), and previous day
FI0.7. A history of three hour magnetic indices (ap) can be used for somewhat
better detail during magnetic storms. The model calculates neutral tempera-
ture, total density, and densities of N2, O , O, N, He, Ar, and H. The modelis based on a fit of in situ composition an_ temperature data from eight
scientific satellites (OGO-6, San Marco-3, Aeros-A, AE-C, -D, -E, ESRO-4, and
DE-2) and numerous rocket probes, as well as five ground-based incoherent
Year
1960
1965
1970
1975
1980
1985
ICAOARDC
HistoricaJ Development of Empirical Thermosphere Models
Primarily Total Density D=t= Primarily Temp. & Comp. Datafrom Satellite Drag from Ground and In-situ Instr.
I
DENSEL _ J60
LOCKHEED- (Jacchia}JACCHIA
USSA62
USSASuppl.
USSA76
J65
JACCHIA _L/"
WALKER-BRUCE
J70
GRAM(Justus)
ClRA61I
i HARRIS-J PRIESTER
'/I
II
CIRA65
J77
GRAM3
II
ii
I
Iil
t
I
'0, 72
ClRA86 (
OG06(Hedin)
M 1,M2 MSIS77 ESR04
(Thuillier) / (Hedin) (yon Zahn)DTM - r -'7 /
(Barlier) / l
AEROS MSIS(Kohnlein) UT/Long_
C(Kohnlein)
MSIS83
JMSIS86
Fig. i. History of empirical models.5
scatter stations (Millstone Hill, St. Santin, Arecibo, Jicamarca, and
Malvern). The model supersedes the MSIS-83 model by inclusion of high
latitude, high solar activity data from the Dynamics Explorer satellite, and
the addition of atomic nitrogen to the gas species included in the model. The
MSIS-86 model was selected by COSPAR for inclusion in the next CIRA which has
yet to be published.
MSIS-8_
The MSIS-83 model (Hedin, 1983) is an empirical model of neutral
temperature and density in the thermosphere (above 85 km) and lower exosphere
based on in situ mass spectrometer and incoherent scatter data. The model is
dependent on user-provided values of day, time (UT), altitude, latitude,
longitude, local solar time, magnetic index (Ap), a three solar rotation
average I0.7-0m radio flux (FI0.7), and previous day FI0.7. A history of
three hour magnetic indices (ap) can be used for somewhat better detail during
magnetic storms. The model calculates neutral temperature, total density, and
densities of N2, 02, O, He, At, and H. The model is based on a fit of in situcomposition and temperature data from seven scientific satellites (OGO-6, San
Marco-3, Aeros-A, AE-C, -D, -E, and ESRO-4) and numerous rocket probes, as
well as five ground-based incoherent scatter stations (Millstone Hill, St.
Santin, Arecibo, Jicamarca, and Malvern). The model supersedes the MSIS-77
model by inclusion of data from the AE-D, AE-E, and ESRO-4 satellites, as wellas additional data from incoherent scatter stations that cover a wide range of
solar activities and the inclusion of longitude/UT variations. The MSIS-83
model extends the previous description of neutral parameters below 120 km to
the base of the thermosphere in a continuous manner.
MSIS-77
The MSIS-77 model (Hedin et al., 1977a; 1977b) is an empirical model
of neutral temperature and density in the thermosphere (above 120 km) and
lower exosphere based on in situ mass spectrometer and incoherent scatter
data. The model is dependent on user-provided values of day_ altitude,
latitude, local solar time, magnetic index (Ap), a three solar rotation
average I0.7-cm radio flux (FI0.7), and previous day FI0.7. The model
calculates neutral temperature, total density, and densities of N2, 02, O, He,Ar, and H. Themodel is based on a fit of in situ composition and temperature
data from five scientific satellites (AE-B, OGO-6, San Marco-3, Aeros-A, and
AE-C), as well as four ground-based incoherent scatter stations (Millstone
Hill, St. Santin, Arecibo, and Jicamarca). The model supersedes the OGO-6
model, which was based on data from only one satellite. The OGO-6 model is
now totally obsolete.
The MSIS-77 model used in this Study specifically did not include the
longitude/UT variations (Hedin et al., 1979), which are sometimes associatedwith this model name.
J77
The J77 model (Jacchia, 1977) is the latest in a series of empirical
models of neutral temperature and density in the thermosphere (abov e 90 km)
and lower exosphere based on atmospheric drag effects on satellite orbits.
The model is dependent on user-provided values of altitude, latitude, sun6
declination, hour angle of sun, fraction of tropical year, invariant or
geomagnetic latitude, magnetic index (kp), a six solar rotation average of
FI0.7, and previous day FI0.7. The model calculates neutral temperature,
total density, and densities of N , O , O, He, At, and H. The model is basedprimarily on total densities deri_ed _rom changes in satellite orbits
(approximately 16 satellites during the 1960's) with an attempt to represent
changes in composition observed by OGO-6 and ESRO-4. The model differs from
the J71 and J70 models by the inclusion of elaborate formulations to describe
composition changes with local time and magnetic activity.
The code used for this model was generated from the original publica-
tion without later unpublished modifications.
JT__fio
The JTO model (Jacchia, 1970) is an empirical model of neutral
temperature and density in the thermosphere (above 90 km) and lower exosphere
based on atmospheric drag effects on Satellite orbits. The _odel is dependent
on user-provided values of altitude, latitude, sun declination, hour angle of
sun, fraction of tropical year, magnetic index (kp), a three solar rotation
average of FI0.7, and previous day FI0.7. The model calculates neutral
temperature, total density, and densities of N2, O2, O, He, At, and H. Themodel is based on total densities derived from changes in satellite orbits
(approximately 16 satellites during the 1960's). The model essentially
differs from the later J71 (Jacchia, 1971) by a smaller value of the 0/02
ratio at 150 km. The J70 model superseded the earlier J65 model (Jacchi_,
1965) by extending the model calculations below 120 km.
This model is the thermosphere end of the Marshall Space Flight Center
GRAM model, and the code for this model w_s obtained from MSFC.
3.2. Residual plots
Each of the data sets described above was divided into subsets
according to altitude and then compared to the five models in order to
calculate data residuals by taking the logarithm (base e) of the ratio of the
measured total density to model total density. The mean residual, standard
deviation of the residuals (square root of the sum of squares of each residual
minus the mean residual), and RMS (square root of the sum of squares of the
residuals) were calculated, and the residuals used to generate histogram plots
of the residuals and a large number of plots of residuals versus various
coordinates. The detailed description of the plots is given in Appendix C,
and the plots themselves are largely in microfich format. The binned data
used for the histogram plots (Figures CI and C2 of Appendix C) are also in a
series of ASCII computer files as described in Appendix B.
The total number of data points in some data sets was very large
(Table I). For handling and plotting convenience, a subset of points was
usually selected at random to bring the total points under 20,000. Previous
experience and statistics suggest this should be adequate for the current
comparisons. In any large data set, there are almost always a number of
points that are erroneous because of occasional problems somewhere in the
electronic or data-handling systems. Thus data points whose residuals from
both MSIS-86 and J70 were more than 10 times the estimated experimental error
were discarded. Normally, only a few percent, and in no case more than 10%,of the points were dropped in this test. For each data set the exact samesetof points was finally comparedwith each of the five models and thus shouldprovide a reasonable relative comparison of the models. Judgments as to therelative value of various data sets is more problematic without considerableindividual attention to the nature of the problem points.
The model densities were calculated using the same (three solar cycle)average FI0.7 and one-day lag for each model. Also, most of the comparisonswere done using the daily Ap/Kp rather than the three-hour index. Limitedtests indicated a possible improvement of a percent or two (for all modelsexcept MSIS-77, which was not designed to use the more detailed index) inoverall standard deviations using the appropriate three-hour index.
3.3. Results
A sucmmry of the mean residuals for each of the data subsets under
magnetically quiet (Ap<=10) conditions is given in Table 2 as the logarit_m
(base e), along with the rank (I best to 5 worst) of each model with respect'
to this data subset in parenthesis after the residual. At the bottom is an
accumulation of how many data sets were ranked I to 5 for each model. These
rankings are plotted in Figure 2 and show that, except for MSIS-77, there isan approximately equal chance that any one of the models would give the lowest
mean residual for any of the data subsets. Also in the table is the number of
points and, in parenthesis after the number of points, the fraction of the
original data set used in the calculation.
The average over all data subsets for the models is -.02, -.O1, -.12,
-.01, -.04 for MSIS-86, -83, -77, J77, and J70, respectively. In other words,
the average difference in absolute densities between the existing data sets
and models is generally only a few percent. Averages based on mass spectro-
meter data are about 4% (.04 in logarithms) less than averages based on drag
data only. This is well within any a priori estimates of calibration errors.
Examination of specific data sets like Jacchia drag (used for generating theJacchia'models but not the MSIS models) or AE-E NACE (used for generating MSIS
models but not Jacchia models) confirms that there is little difference in
absolute values overall across models. The exception is MSIS-77, which was
generated with a database which did not have complete solar activity coverage,
but not surprisingly does well with, for example, AE-C data that were used in
generating this model
A summary of the overall standard deviations of the residuals for eachof the data subsets under magnetically quiet (Ap<=10) conditions is given in
Table 3 in the same format as Table 2, and the rankings are plotted in
Figure 3. Here we see a systematic trend toward lower standard deviations in
the later MSIS models. Taking a given data subset at random, one is more
likely to obtain the smallest standard deviations in the residuals using MSIS-
86. There are, of course, specific exceptions to this generalization such as
the'high altitude Jacchia drag data, where MSIS-86 and JTO are equivalent at
400-800 km and J70 is best above 800 km, and the counter example of OGO-6,
where MSIS-86 is best in the 400-800 km range. Small differences in ranking,
such as between J77 and J70, may not be significant, and it should be
remembered that in many cases the difference between best and worst is only a
few percentage points. Also, the data subsets are not entirely independent
8
Table 2. Overall mean (magnetically quiet).
Jacchiadrag
Barlierdrag
OGO-6ms
AE-C MESA
accel
AE-C OSS
ms
AE-D MESA
accel
AE-D OSS
ms
AE-D NACE
ms
Cactus
accel
AE-E MESAaccel
AE-E OSS
ms
AE-E NACE
ms
DE-2 NACS
ms
Alt Pts MSIS-86 MSIS-83 MSIS-77 J77 J70
200-400 3197 -.05 (3) -.05 (3) -.21 (4) -.03 (1) -.04 (2)4O0-800 6516 -.05 (1) -.06 (2) -.23 (3) -.05 (1) -.O6 (2)800-1200 3386 .O4 (2) .O4 (2) -.04 (2) -.O3 (1) -.05 (3)
120-200 1050 .02 (1) -.03 (2) -.02 (1) -.04 (3) -.07 (4)200-400 4761 .04 (3) .02 (I) -.06 (4) -.04 (3) .03 (2)
400-800 1447 .10 (2) .I0 (2) -.04 (I) .I0 (2) .15 (3)
200-400 1978(.20) .20 (2)
400-800 12863(.07) .20 (2)
120-200 5746(.25) .11 (3)
200-250 6101(.55) .02 (I)
.20 (2) .04 (1) .09 (2) .15 (3)
.20 (2) .04 (1) .21 (3) .25 (4)
.11 (3) .05 (2) .05 (2) .00 (1)
.04 (3) -.08 (4) .03 (2) -.03 (2)
120-200 6447(.65)-.01 (I) -.01 (I) -.05 (3) -.02 (2) -.07 (4)
200-390 6279(.2) .02 (2) .06 (3) -.08 (4) .09 (5) .01 (I)
120-200 11024(.7) -.03 (3)
200-250 4399 -.08 (3)
120-200 11787 -.01 (I)
200-390 I0923(.75)-.06 (3)
.00 (I) .00 (I) .02 (2) -.02 (2)
-.04 (2) -.11 (4) .01 (I) -.04 (2)
120-200 4614
200-400 5591
.01 (1) .01 (1) .04 (2) .01 (1)-.02 (1) -.20 (4) .03 (2) -.03 (2)
-.12 (3) -.09 (I) -.I0 (2) -.15 (4) -.18 (5)
-.21 (3) -.14 (I) -.33 (5) -.17 (2) -.22 (4)
200-400 11276(.I0)-.05 (2)
400-600 9808(.I0)-.02 (2)
120-200 11455(.45) .02 (2)
200-250 8363 -.01 (2)
120-200 11812(.62)-.01 (2)
200-400 9741(.12)-.03 (2)
120-200 7533 -.05 (1)200-400 10158(.14)-.09 (3)400-600 I0243(.20)-.17 (4)
200-400 6079(.08)-.23 (3)
400-800 3149(.20)-.17 (4)
-.05 (2) -.26 (4) -.03 (I) -.O8 (3)
-.O1 (I) -.36 (4) .01 (I) -.O7 (3)
.02 (2) -.01 (1) -.01 (1) -.03 (3)
.00 (1) -.06 (4) .05 (3) -.01 (2)
.00 (I) -.07 (4) -.05 (3) -.07 (4)
-.03 (2) -.15 (3) -.01 (I) -.03 (2)
-.05 (I) -.05 (I) -.05 (I) -.07 (2)
-.09 (3) -.22 (4) -.02 (2) -.07 (I)
-.14 (3) -.32 (5) -.12 (2) -.08 (1)
-.23 (3) -.33 (4) -.15 (1) - 22 (2)-.15 (3) -.38 (5) -.09 (1) -.14 (2)
Rank summary: I-6 1-11 I-8 1-11 I-5
2-11 2-10 2-3 2-11 2-12
3-10 3-8 3-3 3-5 3-74-2 4-0 4-12 4-I 4-5
5-3 5-I
n.,
p.,
m
-.. Q
e.
----"'T 7 I f
0 0 00 _ Q_
F
0 0 o'q" t_;
_o a_etuao_a d
0D-."3
D,-D-."3
D-.D,-
I
O0
O__r
¢0O0
I
GO
O0
C,OOO
!
GO
Q;
O"
0
.i-)
bO
Q;"0
bO
p.
0
bOr-
w
C_S-,
Q)"00
s-
r..
10
Table 3. Overall Standard Deviation (magnetically quiet).
Jacchia
drag
Barlier
drag
OGO-6
ms
AE-C MESA
accel
AE-C OSS
ms
AE-D MESA
accel
AE-D OSS
ms
AE-D NACE
ms
Cactus
accel
AE-E MESA
accel
AE-E OSS
ms
AE-E NACE
ms
DE-2 NACS
ms
Aft Pts MSIS-86 MS15-83 MSIS-77 J77
------m_m----
200-400 3197 .15 (I) 16 (2) .18 (3) .15 (I)
400-800 6516 .26 (2) .26 (2) .29 (3) .25 (I)
800-1200 3386 .27 (3) .26 (2) .30 (4) .29 (3)
J70
.16 (2)
.26 (2)
.22 (I)
120-200 1050 .22 (1) .23 (2) .22 (1) .22 (1) .23 (2)200-400 4761 .20 (1) .20 (1) 21 (2) .21 (2) .21 (2)4OO-8OO 1447 .31 (1) .31 (1) .33 (3) .32 (2) .31 (1)
200-400 1978(.20) .14 (I)
400-800 12863(.07) .17 (I)
.14 (I) .14 (I) .16 (2) .17 (3)
.18 (2) .18 (2) .20 (3) .21 (4)
120-200 5746(.25) .16 (2) .16 (2) .15 (1) .18 (3) .18 (3)200-250 6101(.55) .21 (1) .21 (1) .21 (1) .21 (1) .21 (1)
120-200 6435(.65) .12 (2) .12 (2) .10 (1) .13 (3)200-390 6279(.2) .14 (1) .14 (1) .14 (1) .15 (2)
.13 (3)
.16 (3)
120-200 11024(.7) .i5 (2) .15 (2) .14 (1) .15 (2) .14 (1)200-250 4399 .18 (1) .19 (2) .18 (1) .18 (1) .18 (1)
.11 (1) .11 (1) .13 (2)
.18 (2) .18 (2) .17 (1)120-200 11787 .11 (1)200-390 10923(.75) .17 (1)
.11 (1)
.17(1)
120-200 4614 .17 (2) .17 (2) .16 (1) .16 (1) .16 (1)200-400 5591 .21 (2) .22 (3) .23 (4) .20 (1) .21 (2)
.15 (2) .18 (4) .16 (3) .16 (3)
.25 (2) .28 (5) .26 (3) .27 (4)
.12 (li .12 (1) .14 (2) .14 (2)
.18 (1) .19 (2) .20 (3) .20 (3)
.14 (2) .13 (1) .16 (4) .17 (5)
.18 (1) .21 (4) .20 (3) 21 (4)
.11 (1) .15 (2) .17 (4) .16 (3)
.19 (1) ..20 (3) .20 (2) .20 (2)
.19 (2) .22 (4) .18 (1) .20 (3)
200-400 11276(.I0) .14 (I)
400-600 9808(.IO) .24 (I)
120-200 11455(.45) .12 (1)200-250 8363 .18 (1)
12'0-200 11812(.62) .15 (3)
200-400 9741(.12) .19 (2)
120-200 7533 .11 (1)200-400 10158(.14) .20 (2)400-600 10243(.20) .19 (2)
200-400 6079(.08) .14 (1) .16 (2) .23 (5) .20 (4) .19 (3)400-800 3149(.20) .17 (1) .21 (3) .22 (4) .21 (3) .19 (2)
1-18 1-11 1-12 1-9 1-82-9" 2-16 2-5 2-8 2-83-2 3-2 3-4 3-9 3-94-0 4-0 4-6 4-3 4-3
5-2 5-I
]]
u
cor_
4)
O"
>,
>,...,
m m
m (/1
_0 t"
(J 4D
4.m
m
co
(0 O
X [--
O O O O QO _ G) 'q" C_1m
s_,_S e_e G _o a_e_uaoJa d
0
_J
O"
,.-I
o
4-)
b0
E
0
"0
"o
"o
_J
ob_
°,_
w
c.
'00
12
because of the grouping of instruments on the same satellite. Yet, for
reasons not understood, the various instruments on a given satellite do not
always give the same density, and it is not usually clear which one is right.
A summary of the mean residuals for each of the data subsets under
magnetically active (Ap>10) conditions is given in Table 4 in the same format
as Table 2, and rankings are plotted in Figure 4. The features are much the
same as for quiet conditions. The average over all data subsets for the
models iS -.03, -.01, -.07, .01, -.08 for MSIS-86, -83, -77, J77, and J70,
respectively. The MSIS-77 and J70 models have the most change in mean value
(about 4%) with increasing magnetic activity.
A stm_nary of the overall standard deviations of the residuals for each
of the data subsets under magnetically active (Ap>10) conditions is given inTable 5 in the same format as Table 2, and the rankings are plotted in
Figure 5. The features are much the same as for magnetically quiet conditions
except J77 has systematically higher standard deviations.
The models were also ranked for several specific types of variations
in Tables 6-10 and Figures 6-10 using data for all magnetic activities. Herethe standard deviation of the means of the binned data were calculated and
used for ranking. Table 6, for example, was taken from the plots showing
average residuals in l-hour bins as a function of local time (e.g., Figure C4
of Appendix C). If the data and model variations as a function of latitude
were the same, and the coverage for other important parameters, such as local
time, is either the same for each bin or correctly modeled, then the plotted
averages should be equal, and the standard deviation of these average values
zero. Coverage in other parameters is rarely perfect, but this plot and
ranking still emphasize selected types of variations. The local time ranking
(Table 7) corresponds to Figure C5, the mean FI0.7 ranking (Table 8) to Figure
C6, the daily FI0.7 ranking (Table 9) to Figure C7, and the magnetic activity
(Ap) ranking (Table 10) to Figure C19. The models are about equal in
predicting overall latitude variations (Figure 6), except for J77, which issomewhat worse. For local time variations (Figure 7) and daily FI0.7
variations (Figure 9), the later models are systematically better. For mean
FI0.7 variations, the models are about equal, except MSIS-77 is worse. For
magnetic activity variations (Figure 10), the later models are only slightly
better, and J77, worse.
A summary of the data comparisons on a data-set by data-set basis is
given in Appendix D with the attempt to identify variations that could
fruitfully be examined further. The most frequently noted residual trends (in
order) involved altitude, daily FIO.7 (daily minus mean), magnetic activity,
and annual (or semiannual) variations. The altitude trends could be a
combination of measurement problems and model problems. All of these trends
deserve careful Study by looking for similarities and differences between thedifferent data sets. Since these trends are handled by the model(s) in an
overall sense, the presence of a trend in a particular data set presumably
indicates that the magnitude of that variation (like the semiannual variation)
depends on some other unidentified factor. Both this data set summary and the
preceding discussion of overall means and standard deviations provide only ahint of the rich detail, exceptions, and anomalies that can be found in the
comparison plots.
]3
Table 4. Overall mean (magnetically active).
Alt Pts MSIS-86 MSIS-83 MSIS-77 J77 J70
Jacchia
drag
200-400 3197
400-800 6516
800-1200 3386
-.05 (3) -.04 (2) -.16 (5)-.04 (I) -.04 (I) -.16 (3).04 (2) .06 (3) .02 (I)
.02 (I) -.12 (4)
.06 (2) -.16 (3)
.07 (4) -.12 (5)
Barlier 120-200 681
drag 200-400 2761400-800 824
.o0 (I) -.06 (3) -.03 (2) -.14 (5) -.09 (4)-.02 (2) -.04 (3) -.09 (5) .o0 (I) -.06 (4):04 (2) .05 (3) -.05 (3) .11 (4) .O0 (I)
OGO-6
ms
200-400 757(.20) .00 (I) -.03 (2) - 17 (4) - 05 (3) -.03 (2)
400-800 6557(.07) .15 (4) .14 (3) .02 (I) .23 (5) .12 (2)
AE-C MESA 120-200 5746(.25) .14 (5)
accel 200-250 6101(.55) .02 (I)
.13 (4) .10 (2)
.05 (3) -.03 (2)
.11 (3) .07 (I)
.07 (4) -.05 (3)
AE-C OSS 120-200 6447(.65) .01 (2) .01 (2) .00 (I)
ms 200-390 6279(.2) .06 (3) .10 (4) .02 (2)
.01 (2) -.02 (3)
.16 (5) .O0 (I)
AE-D MESA 120-200 11024(.7) -.03 (3) .00 (I) .04 (4)
accel 200-250 4399 -.08 (4) -.03 (I) -.05 (2)
.03 (3) -.02 (2)
.06 (3) -.06 (3)
AE-D OSS
ms
120-200 11787 -.01 (2)
200-390 10923(.75)-.06 (2)
.02 (3) .06 (4)
.O0 (I) -.12 (5)
.06 (4) .00 (I)
.09 (4) -.08 (3)
AE-D NACE 120-200 4614
ms 200-400 5591
-.09 (4) -.05 (2) -.02 (I) -.06 (3) -.13 (5)-.19 (3) -.11 (2) -.22 (4) -.07 (I) -.23 (5)
Cactus
accel
200'400 8701(.I0)-.02 (I) -.O2 (I) -.15 (4) -.03 (2) -.14 (3)
400-600 8348(.10) .OO (I) .03 (2) .23 (5) .15 (3) -.21 (4)
AE-E MESA 120-200 11455(.45) .06 (3)
accel 200-250 8363 .03 (2)
.07 (4) -.03 (2) -.06 (3) -.02 (I)
.O5 (4) .01 (I) .04 (3) -.03 (2)
AE-E OSS
mS'
120-20011812(.62) .02 (2) .02 (1) -.03 (4) -.11 (3) -.06 (4)200-400 9741(.12)-.03 (1) -.03 (1) -.12 (3) -.03 (1) -.09 (2)
AE-E NACE 120-200 7533 -.04 (I) -.04 (I) -.08 (2) -.15 (4) -.09 (3)
ms 200-400 10158(.14)-.O9 (3) -.07 (I) -.18 (5) -.08 (2) -.11 (4)
400-600 10243(.20)-.18 (3) -.15 (2) -.29 (4) -.14 (I) -.14 (I)
DE-2 NACS 200-400 12286(.08)-.28 (3) -.28 (3) -.35 (4) -.15 (I) -.26 (2)
ms 400-800 90,77(.20)-.17 (3) -.16 (2) -.31 (5) -.02 (I) -.22 (4)
Rank summary:
]4
I-8 I-9 I-5 I-7 I-62-8 2-8 2-7 2-4 2-6
3-9 3-8 3-3 3-9 3-7
4-3 4-4 4-8 4-6 4-7
5-I 5-6 5-3 5-3
B
C
am_
C
rv
U w_
.¢.m
1.0
!
I
t_
o
m
O)
QO
m
m m
OJ
Q O Q (_
sl._$ el,e a jo a_e_uaoJa d
Q
0
!
OlI-..4
CO
CO(DOI
(.0m.-m¢
Ol
O0I
6Y)
(Y)
.,-q
p'q
.,-.4
(or-
c_
(D
r_0c_
boc-
=:
.,.-(
15
Table 5. Overall Standard Deviation (magnetically active).
Jacchia
drag
Barlier
drag
@30-6
ms
AE-C MESA
accel
AE-C OSS
ms
AE-D MESA
accel
AE-D OSS
ms
AE-D NACE
ms
Cactus
accel
AE-E MESA
accel
AE-E OSS
ms
AE-E NACE
ms
DE-2 NACS
ms
Aft Pts MSIS-86 MSIS-83 MSIS-77 J77
200-400 2464 .18 (1) .19 (2) .20 (3) .21 (4)400-800 4007 .29 (2) .30 (3) .31 (4) .29 (2)800-1200 2549 .29 (2) .30 (3) .33 (5) .32 (4)
120-200 1050 .22 (2) .22 (2) .22 (2) .23 (3)
200-400 2767 .20 (I) .21 (2) .22 (3) .23 (4)
400-800 824 .30 (2) .29 (I) .32 (4) .31 (3)
200-400 757(.20) .19 (I)
400-800 6557(.07) .20 (I)
120-200 12359(.25) .16 (2)200-250 11585(.55) .21 (2)
120-200 12370(.65) .13 (2)200-390 6279(.2) .15 (1)
J70
.18 (I)
.28 (I)
.23 (1)
.21 (I)
.21 (2)
.30 (2)
.19 (1) .20 (2) .21 (3) .21 (3)
.21 (2) .20 (1) .22 (3) .21 (2)
.16 (2) .15 (1) .21 (4) .18 (3).21 (2) .20 (1) .23 (3) .21 (2)
.13 (2) .11 (1) .16 (4) .14 (3)
.16 (2) .17 (3) .19 (4) .16 (2)
120-200 8907(.7) .15 (1) .15 (1) .15 (1) .17 (2) .15 (1)200-250 3723 .19 (1) .19 (1) .19 (1) .20 (2) .19 (1)
120-200 7608 .13 (1) .13 (1) .13 (1) .15 (3)200-390 7476(.75) .22 (2) .21 (1) .21 (1) .23 (3)
.14 (2)
.21 (1)
120-200 4016 .13 (1) .13 (1) .13 (1) .14 (2) .14 (2)200-400 5566 .22 (2) .22 (I) .23 (2) .24 (3) .22 (I)
200-400 8701(.10) .17 (1) .17 (1) .19 (2) .21 (3) .19 (2)400-600 8348(.10) .28 (1) .29 (2) .30 (3) .31 (4) .30 (3)
.13 (1) .13 (1) .16 (3) .15 (2)
.18 (2) .18 (2) .20 (4) .19 (3)
120-200 8289(.45) .t3 (1)200-250 5582 .17 (1)
120-200 7868(.62) .16 (2)200-400 7854(.12) .19 (1)
.15 (1) .15 (1) .19 (4) .18 (3)
.20 (2), .22 (4) .23 (5) .21 (3)
120-200 6766 .13 (1) .13 (1) .17 (3) .19 (4)200-400 9079(.14) .23 (2) .22 (1) .24 (3) .23 (2)400-600 7987(.20) .23 (1) .23 (I) .27 (2) .23 (1)
.16 (2)
.23 (2).23 (1)
200-400 12286(.08i 16 (I)
400-800 9077(.20) ,24 (I).17 (2) .21 i5) .20 (4) .18 (3)
.27 (3) .28 (4) .26 (2) .24 (1)
Rank summary: 1-18 1-14 1-11 1-1 1-102-11 2-12 2-6 2-6 2-113-0 3-3 3-6 3-10 3-84-0 4-0 4-4 4-11 4-0
5-2 5-I
]6
J_
¢0
>_t
> ••.., -_,4=t ,O_
u G3
m m
U
.- r-_
1--.
i
I.O
0
0qm
' I
CO
,.q-
O)
)
m
I
I
,0)
CO
I i
0 0(I0 CD
1 , !o o'q" C_i
aae_uao_a d
Q
I
co
co.s"
coCOI
O3
CO
COO0I
co
o3
0
>
p'4
,--4
cJ
boa_E
0
,,..4
r_
j..)co
0
E_
17
Table 6. Latitude Standard Deviation.
Jacchia
drag
Barlier
drag
@30-6ms
AE-C MESA
accel
AE-C OSS
ms
AE-D MESA
accel
AE-D OSS
ms
AE-D NACE
ms
DE-2 NACS
ms
Alt Pts MSIS-86
200-400 5661 .03 (I)
400-800 10529 .05 (2)
800-1200 5935 .O8 (3)
120-200 1731 .07 (2)
200-400 7528 .03 (I)
400-800 2271 .08 (3)
200-400 2735(.20) .04 (I)
400-800 19420(.O7) .03 (2)
120-200 18105(.25) .04 (1)200-250 17686(.55) .03 (1)
120-200 19050(.65) .05 (2)200-390 18897(.2) .04 (3)
MSIS-83 MSIS-77 J77 J70
.04 (2) .O5 (3) .04 (2) .03 (I)
.O7 (3) .08 (4) .04 (I) .04 (I)
.06 (2) .08 (3) .08 (3) .05 (1)
.07 (2) .06 (1) .10 (3) .07 (2).03 (1) .03 (1) .03 (1) .04 (2).o8 (I) .o7 (2) .08 (3) .o8 (3)
Rank summary:
.O6 (2) .07 (3) .O8 (4) .06 (2)
.03 (2) .02 (I) .O5 (3) .O6 (4)
.o4 (I) .04 (I)
.o5 (3) .04 (2)
.12 (3)
.o7 (4).09 (2).04 (2)
.04 (I) .04 (I) .O8 (3) .08 (3)
.01 (2) .03 (2) .04 (3) .03 (2)
120-200 19931(.7) .04 (2) .05 (3) .04 (2) .O8 (4)
200-250 8122 .O5 (I) .07 (3) .05 (I) .06 (2)
120-200 19365 .04 (2)200-390 18399(.75) .06 (3)
.03 (1)
.05 (1)
.04 (2) .02 (1) .07 (3) .04 (2)
.05 (2) .04 (1) .06 (3) .o6 (3)
120-200 8630 .10 (4) .09 (3) .06 (1) .07 (2) .11 (5)200-400 11157 .06 (1) .09 (4) .06 (1) .08 (3) .07 (2)
200-400 18365(.08) .03 (1)400-800 12226(.20) .03 (1)
I-92-6
3-4
4-I
.05 (2) .05 (2) .06 (3) .05 (2).04 (2) .04 (2) .08 (3) .03 (1)
1-42-10
3-54-I
1-10 1-2 1-62-6 2-3 2-9
3-3 3-12 3-34-I 4-3 4-I
5-O 5-O 5-I
]8
4"*
C4;r_
m
oF-
50"q"
m
CGram
C
n-
com
4,a
m
&.
>
"o
a.%
.J
00 •m
_D
(5
m
0(D
s_$
J , J
O_
CXl
OJ
0 0 0
0D,."9
D-.D,-"9
D,.
t".!
OO
O9
COO0!
O_
OO
m
CJ)
OO!
OO_M..'4
"w
0
r-
Q
.,-4
m>
QJ-o
4.3
J.J
m,..M
o
b_
J._r-m
"0
o
,d
19
Table 7. Local Time Standard Deviation.
Jacchia
drag
Barlier
drag
OGO-6
ms
AE-C MESA
accel
AE-C OSS
ms
AE-D MESA
aecel
AE-D OSS
ms
AE-D NACE
ms
Cactus
accel
AE-E MESA
accel
AE-E OSS
ms
AE-E NACE
ms
DE-2 NACS
ms
Alt Ptso
200-400 5661
400-800 10529
800-1200 5935
b_m_
MSIS-86 MSIS-83 MSIS-77 J77 J70
.03 (1) .04 (2) .03 (1) .05 (3) .o_ (2)
.06 (2) .07 (3) .07 (3) .07 (3) .03 (I)
.07 (2) .07 (2) .14 (4) .12 (3) .05 (1)
120-200 1731
200-400 7528
400-800 2271
.06 (I) .07 (2) .06 (I) .08 (3) .o9 (4)
.02 (I) .03 (I) .02 (2) .04 (2) .03 (2)
.04 (I) .04 (I) .06 (3) .06 (2) .04 (I)
200-400 2735(.20) .13 (2)
400-800 19420(.07) .04 (I)
120-200 i8105(.25) .03 (2)
200-250 17686(.55) .04 (2)
120-200 19050(.65) .04 (1)200-390 18897(.2) .02 (1)
.12 (1) .12 (1) .12 (1) .12 (1).04 (1) .04 (1) .06 (2) .06 (2)
.03 (2) .02 (I) .07 (3) .07 (3)
.04 (2) .03 (I) .07 (3) .07 (3)
.04 (1) .04 (1) .05 (2) .06 (3).03 (2) .02 (1) .05 (3) .05 (3)
120-200 19931(.7) 10 (1) .11 (2) .12 (3) .15 (4) .10 (1)200-250 8122 .09 (2) .I0 (3) .I0 (3) .09 (2) .08 (I)
120-200 19365 .10 (2)200-390 18399(.75) .11 (2)
.09 (1) .09 (1) .09 (1) .10 (2)
.11 (2) .10 (1) .13 (4) .12 (3)
120-200 8630 .10 (4) .09 (3) .06 (1) .08 (2) .10 (4)200-400 11157 .12 (1) .13 (2) .13 (2) .12 (1) .13 (2)
200-400 19977(.10) .02 (1)400-600 18156(.10) .03 (1)
120-200 19744(.45) .03 (1)200-250 13945 .05 (1)
120-200 19680(.62) .02 (1)200-400 17595(.12) .02 (1)
120-200 14290 .03 (1)200-400 19237(.14) .03 (1)400-600 18230(.20) .04 (2)
200-400 18365(.08) .06 (1)400-800 12226(.20) .08 (1)
.03 (2) .05 (3) .09 (5) .o7 (4)
.04 (2) .o9 (4) .o8 (3) .o8 (3)
.04 (2) .04 (2) .04 (2) .08 (3)
.06 (2) .06 (2) .10 (3) .10 (3)
.03 (2) .04 (3) .10 (5) .09 (4)
.03 (2) .04 (3) .09 (4) .09 (4)
.03 (I) .06 (2) .I0 (4) .09 (3)
.04 (2) .06 (4) .07 (5) .05 (3)
.03 (I) .I0 (4) .04 (2) .05 (3)
.06 (1) .10 (3) .07 (2) .07 (2)
.09 (2) .10 (3) .09 (2) .09 (2)
Rank summary: 1-19
2-9
3-04-I
I-9
2-173-34-0
1-112-53-94-45-0
1-32-10
3-94-4
5-3
I-6
2-7
3-11
4-5
5-0
2O
es,m.
tn
¢o
ot--
_=J_
¢0
03c-o
E.m>.
M
=ram
oJ
m
q.
Ioc3m
I i I
co
C_J
! I Io 0GD CD
s".=S e _.e13
!
! i I0 o_r ¢_1
_o a_e_uaaJa d
0t".-
.-a
o
r"..I
03
03_0
I
03
03:Z
¢DO0
!
O3
CO3E
0
o_,_L,
:>
.t,.4
,--I
0o
o
e..
21.
Jacchia
drag
Barlier
drag
OGO-6ms
AE-C MESA
accel
AE-C OSS
ms
Cactus
accel
AE-E MESA
accel
AE-E OSS
ms
AE-E NACE
ms
DE-2 NACS
mS
Table 8. Mean FI0.7 Standard Deviation.
Alt Pts
200-400 5661
400-800 10529
800-1200 5935
120-200 1731
200-400 7528
4O0-800 2271
200-400
,ill illiliJll II
MSIS-86 MSIS-83 MSIS-77 J77 J70
.05 (2) .04 (1) .05 (2) .06 (3) .05 (2)
.11 (2) .11 (2) .13 (3) .09 (1) .11 (2)
.08 (2) .08 (2) .11 (3) .13 (4) .07 (1)
.05 (3) .04 (2) .06 (4) .03 (1) .03 (1)
.03 (1) .04 (2) .04 (2) .04 (2) .03 (1)
.07 (2) .07 (2) .08 (3) .07 (2) .06 (I)
2735(.20) .04 (1) .06 (3) .05 (2) .05 (2)
400-800 19420(.07) .I0 (2)
120-200 18105(.25) .01 (2)
200-250 17686(.55) .04 (2)
120-200 19050(.65) .05 (3)200-390 18897(.2) .01 (1)
200-400 19977(.IO) .09 (4)
400-600 18156(.I0) .11 (3)
.05 (2).10 (2) .08 (1) .13 (3) .10 (2)
.01 (2) .00 (I) .02 (3) .04 (4)
.04 (2) .05 (3) .02 (I) .04 (2)
.05 (3) .03 (1) .04 (2) .06 (4)
.02 (2) .O5 (3) .02 (2) .02 (2)
.09 (4) .08 (3) .06 (2) .06 (1)
.10 (2) ,11 (3) .07 (1) .11 (3)
200-250 13945 .09 (1) .09 (2) .09 (2) .09 (3) .04 (3)
200-400 17595(.12) .06 (I)
200-400 19237(.14) .07 (2)400-600 18230(.20) .06 (2)
200-400 18365i.08) .07 (I)
400-800 12226(.20) .08 (I)
.06 (I) .06 (I) .06 (I) .06 (I)
.06 (I) .o9 (4) .o8 (3) .08 (3)
.05 (I) .08 (4) .O7 (3) .O8 (4)
.o8 (2) .15 (4) .09 (3) .08 (2)
.10 (2) .11 (3) .12 (4) ,10 (2)
Rank summary:I-7 I-4 I-4 I-5 I-6
2-9 2-13 2-4 2-6 2-8
3-3 3-2 3-8 3-7 3-34-I 4-I 4-4 4-2 4-3
22
t.)r_
m
4-)
oi--
n_
om
L
t_
0m
[.I,
u
0
m
CO
0
I ' I
O_ Cq
, I , ! , !0 0 0CD _r C_
B_,BQ j0
0D,.
D-.D,-"9
D-.D,.
!
O3
CO"5"
C_
!
I--4
-s"
I
C_OD
I
O3
",s"
0
t-o
.,,-(
c_°,.-Ic_c_>
D,.-
c_r,.
f-c_Q)p_
o
_D
e-c_f,.,
Q;
o
r..
23
Table 9. Daily Minus Mean FI0.7 Standard Deviation.
Jacchia
drag
Alt Pts MSIS-86 MSIS-83 MSIS-77 J77
200-400 5661 .06 (2) .06 (2) .05 (I) .08 (4)
400-800 10529 .09 (3) .08 (2) .13 (5) .06 (1)800-1200 5935 .07 (1) .07 (1) .11 (4) ,09 (2)
Barlier 120-200 1731 .04 (I) .04 (2) .04 (I) .05 (3)
drag 200-400 7528 .01 (I) .O1 (I) .03 (2) .O8 (3)400-800 2271 ,08 (I) .07 (I) .06 (3) .15 (2)
OGO-6 200-400 2735(.20) ,07 (2) .07 (2) .06 (1) .13 (3)ms 400-800 19420(.07) .03 (I) .04 (2) .05 (3) .11 (4)
AE-C MESA 120-200 18105(.25) .06 (3)accel 200-250 17686(.55) .09 (3)
J70
.07 (3)
.10 (4)
.lO (3)
.05 (4)
.09 (4)
.18 (I)
•14 (4)
•12 (5)
.05 (2) .04 (I) .05 (2) .08 (4)
.08 (2) .05 (I) .05 (I) .I0 (4)
AE-C OSS 120-200 19050(.65) .06 (4) .05 (3) .03 (1) .04 (2)ms 200-390 18897(.2) .05 (2) .05 (2) .04 (I) .05 (2)
.06 (4)
.04 (I)
AE-D MESA 120-200 19931(.7) .10 (3) .11 (4) .10 (3) .06 (2) .05 (1)accel 200-250 8122 .12 (3) .14 (4) .08 (2) .06 (I) .06 (I)
AE-D OSS 120-200 19365 .01 (1) .02 (2) .02 (2) .01 (1)ms 200-390 18399(.75) .04 (2) .05 (3) .01 (I) .01 (I)
AE-D NACE 120-200 8630 .06 (4) .06 (3) .05 (I) .04 (2)
ms 200-400 11157 .09 (I) .I0 (2) .04 (2) .06 (I)
Cactus 200-400 19977(.10) ,11 (4) .08 (3) .08 (3) .06 (1)accel 400-600 18156(.I0) .I0 (2) .08 (I) .12 (4) .11 (3)
AEmE MESA 120-200 19744(.45) .02 (2)
accel 200-250 13945 .02 (I)
.02 (2)
,01 (1)
AE-E oss 120-200 19680(62) .03 (2)ms 200-400 17595(.12) .03 (1)
.04 (4)
.05 (2)
.07 (2)
.11 (3)
.01 (1) .02 (2) .04 (3) .05 (4)
.02 (1) .02 (1) .05 (3) .04 (2)
.02 (1) .03 (2) .05 (3) .06 (4)
.03 (1) .04 (2) .10 (4) .08 (3)
AE-E NACE 120-200 14290 .01 (1) .01 (1) ,03 (3) ,03 (3)ms 200-400 19237(.14) .08 (2) .08 (2) .09 (3) .09 (3)
400-600 18230(.20) .06 (I) .07 (2) .07 (2) .08 (3)
.07 (2) .08 (3) .08 (3)
.09 (3) .06 (I) .I0 (4)DE-2 NAGS 200-400 18365(.08) .06 (I)
ms 400-800 12226(.20) .08 (2)
.02 (2)
.07 (1)
.11 (4)
.i0 (4)
.11 (5)
Rank summary:1-12 I-9 1-11 I-7 I-6
2-9 2-13 2-8 2-7 2-5
3-5 3-5 3-7 3-11 3-4
4-3 4-2 4-2 4-4 4-125-I 5-0 5-2
24
I
_/'GO
I=m
r_
m
I
I
= I0
--'_ "O'q" I
° I__ J
o
"- ]0 m
0o
I ' i
I I1 I0 0
III
1 , I'qr oJ
j0 a_e_ua0Ja d
O
D',..!
I
cr)..s-,
I
O')
r.l)
o
4O
>
r,..
E
I
"M(
r.o
o
fT.
25
Table 10. Ap Standard Deviation.
Jacchia
drag
nm_
Alt Pts
200-400 5661
400-800 10529
800-1200 5935
ii O iilli
MSIS-86 MSIS-83 MSIS-77 J77
.10 (1) .11 (2) ,14 (3) .16 (4).19 (3) .18 (2) .21 (5) .20 (4).21 (3) .19 (2) .22 (4) .24 (5)
Barlier 120-200 1731
drag 200-400 7528400-800 2271
J70
.lO (1)
.17 (I)
.16 (I)
OGO-6
ms
.09 (2) .08 (1) .08 (1) .12 (3) .08 (1).08 (1) .09 (2) .11 (3) .12 (4) .09 (2)
.11 (2) .09 (I) .11 (2) .12 (3) .13 (4)
200-400 2735(.20) .12 (2)
400-800 19420(.07) .12 (I)
AE-C MESA 120-200 18105(.25) .07 (2)accel 200-250 17686(.55) .10 (2)
120-200 19050(.65) .09 (3)200-390 18897(.2) .08 (2)
.12 (1) .14 (2) .15 (3) .14 (2)
.13 (2) .14 (3) .16 (4) .14 (3)
.07 (2) .06 (I) .17 (4) .11 (3)
.I0 (2) .O9 (I) .14 (4) .12 (3)
.08 (2) .06 (I) .15 (5) .I0 (4)
.07 (I) .I0 (3) .12 (4) .08 (2)AE-C OSS
ms
AE-D MESA 120-200 19931(.7) .O7 (I) .07 (I) IIO8 (2)
accel 200-250 8122 .09 (I) .09 (I) .09 (I)
AE-D OSS
ms
120-200 19365 .10 (I)
200-390 18399(.75) .16 (2)
Cactus
accel
200-400 19977(.10) .11 (3)400-600 18156(.10) .18 (2)
AE-E MESA 120-200 19744(.45) .07 (I)
accel 200-250 13945 .13 (2)
AE-E OSS
ms
120-200 19680(.62) .13 (3)200-400 17595(.12) .11 (2)
AE-E NACE
ms
120-200 14290 .08 (I)
200-400 19237(.14) .17 (3)
400-600 18230(.20) .11 (2)
.11 (3)
.lo (2)
.10 (t) .10 (1) .10 (1)
.15 (1) .15 (1) .19 (3)
.o8 (2)
.09 (1)
.11 (2),16 (2)
.09 (I) .I0 (2) .12 (4) .I0 (2)
.16 (I) .17 (2) .25 (4) .17 (2)
.07 (1) .07 (1) .17 (3) .12 (2)
.12 (1) .12 (1) .13 (2) .12 (1)
.12 (2) .11 (1)
.10 (1) .11 (2)
.08 (1) .09 (2).16 (2) .14 (1).12 (3) .14 (4)
DE-2NACS I200 -400 18365(.08) .09 (I)
ms 400-800 12226(.20) .15 (I)
.13 (3i
.16 (3)
.17 (4)
.16 (2)
.io (I)
.11 (3) .13 (4) .16 (5)
.20 (3) .23 (5) .21 (4)
.16 (4)
.11 (2)
.12 (3)
.20 (4)
.I0 (_)
.10 (2).18 (2)
Rank summary:1-10 1-14 1-11 I-2 I-7
2-11 2-10 2-7 2-3 2-12
3-6 3-3 3-4 3-8 3-44-0 4-0 4-3 4- 11 4-4
5-2 5-3 5-0
26
m
ON
r,.,
>
>,
_,.-.
-" U
u tj
m
r0 _"'QD
O C0
F- _"
LD
oc_m
O"3
CO
{ , {O OG) (D
O4
s_s,e_eQ _0
I
0"3
{ , {O O'q" c_]
0
Op,....-)
{.,,.p,..
I
I
Z
I
0o,-¢
.,-¢
>
.,.-(
o,-¢
0.,-{
0.)
0
¢..
,.-"W
0z
2'7
The ap history (in the case of MSIS-86 and "83) or lagged ap/kp (in
the case of J77 and J70) was tried for the Cactus data (low inclination orbit)
and DE-2 NACS data (polar orbit) in place of the more convenient daily Ap/Kp.
For Cactus the result for the magnetically active data (Ap>10) was a .02 (2%)
decrease in the overall standard deviation for MSIS-86 and -83, and a .01 (1%)
drop for J77. No change for the other models. For DE-2 there was a .O1
decrease for MSIS-86 and no change for the others. Plots against ap were a
bit smoother. It is apparent that on an overall statistical basis using the
three-hour ap's provides only a small improvement and no change in relative
rankings of the models. During a magnetic storm, of course, use of the three-
hour ap/kp can give dramatically better results, particularly for the timingof the onset. However, since high magnetic activity levels occur less
frequently than low magnetic activity levels, the use of the three-hour
indices has only a modest effect on overall statistics.
4. Summary of model strengths and weaknesses
MSIS-86
The MSIS-86 model is systematically better overall than any of the
other models in modeling of total density variations, and equivalent to theothers in absolute values. The standard deviation With respect to the Jacchia
drag data (200-400 km)is .15 (15%) under magnetically quiet conditions and.18 (18%) under active Conditions. The median standard deviation in
comparison to all the data sets under magnetically quiet conditions is about
.17 (17%), and the mean absolute error is -.02 (-2%). Some data sets, usually
with limited coverage, have lower standard deviations. The most obvious
deficiency is at high altitude, where _he database for generating this model
is weak, but where the accuracy of the data from both mass spectrometers and
drag techniques also needs careful assessment. The model is significantly
better than the Jacchia models in local time variations, particularly at lower
altitudes. Although not the focus of this study, a ranking of models for
predictions of temperature and composition variations based on the MSIS
combined data set would show a very strong preference for this model.
MSIS-83
The MSIS-83 model is overall slightly worse than MSIS-86. The
standard deviation with respect to the Jacchia drag data (200-400 km) is .16
(16%) under magnetically quiet conditions and .19 (19%) under activeconditions. The median standard deviation in comparison to various data sets
under magnetically quiet conditions is about .18 (18%) and the mean absolute
error is -.01 (-1%). However, this study did not emphasize, nor does an
overall assessment give a large weight to, the polar variations, which were
the focus of the changes between MSIS-83 and MSIS-86.
MSIS-77
The MSIS-77 model is slightly worse than MSIS-83 in overall standard
deviations, but distinctly worse in absolute densities. The standard
deviation with respect to the Jacchia drag data (200-400 km)is .18 (18%)
under magnetically quiet conditions and .20 (20%) under active conditions.The median standard deviation in comparison to various data sets under
magnetically quiet conditions is about .18 (18%) and the mean absolute error
28
is -.12 (-12%). The greater errors here are largely a consequence of the morelimited data base, lacking coverage at high solar activities, as well ascoverage at high latitudes. The lack of longitude terms in the model shows upmost clearly in comparisons with DE-2 data. This model does as well or betterthan other models in comparisons with the OGO-6and AE-Cdata on which it washeavily based, but not as well as other models in comparisons with data setshaving wider coverage such as the Jacchia drag data. Although _itting the lowaltitude (below 200 km) Atmospheric Explorer variations satisfactorily, themodel formulation was not as general as used in the later models and cannot beextrapolated below the AE altitudes (about 140 km).
J77
The J77 model has the highest standard deviations under magneticallyactive conditions, but also with respect to latitude and local timevariations. The standard deviation with respect to the Jacchia drag data(200-400 km) is .15 (15%) under magnetically quiet conditions and .21 (21%)under active conditions. The median standard deviation in comparison tovarious data sets under magnetically quiet conditions is about .18 (18%), andthe mean absolute error is -.01 (-1%). The attempt to include bettercomposition and temperature variations through the cumbersome, time consuming,and nonphysical pseudo-temperature technique was only partly successful andappears to have detracted from the description of total densities.
J7__q0
The J70 model is slightly worse overall than MSIS-77 in total density
comparisons, but better than J77. The standard deviation with respect to the
Jacchia drag data (200-400 km) is .16 (16%) under magnetically quiet
conditions and .18 (18%) under active conditions. The median standard
deviation in comparison to various data sets under magnetically quiet
conditions is about .19 (19%), and the mean absolute error is -.04 (-4%).
This model is distinctly better than other models at higher altitudes (above
800 km) and usually worse at the lowest altitudes (less than 200 km). Between
200 and 800 km, J70 and MSIS-86 have equivalent standard deviations in
comparison to the drag related data sets, except Cactus, where J70 is slightly
worse. J70 is by far the worst model for temperature and composition
variations.
5. Conclusion
Five empirical models were compared with 13 data sets, including both
atmospheric drag based data and mass spectrometer data. The products of this
study included plots and ASCII files describing database coverage, extensive
comparison plots of data and models, and ASCII files of binned residuals.
Although the most recently published model, MSIS-86, was found to be
the best overall model, the general conclusion of previous studies wasreaffirmed: the best current accuracy is around 15%. A definite, but small
(few percent), improvement in total density accuracy of newer over older
models was discernible in this study.
It is clear that a model (like MSIS-77), which was generated from a
limited database, can be as good or better than other models for some data
29
sets, but not as good for a wider range of data. Similarly, a later model
(like MSIS-86) may do worse than the earlier MSIS-77 on an earlier data set
(e.g., AE-C) because measurement errors or unidentified geophysical factorsmake the newer data sets not completely consistent with the older data sets.
This illustrates the obvious conclusion that a model based on or optimized for
a limited set of data can be very good for that data, but poor for other
data. Likewise, the addition of new data may make the model worse for older
data. However, unless we have reasons to discount certain measurements, the
broader based model is more likely to be best in comparison to a new
independent data set (or for a random prediction), and this is illustrated by
comparisons with the Cactus data, which were used for neither the MSIS orJacchia models.
The excellent overall agreement of the mass spectrometer based MSIS
models with the drag data, including both the older data from orbital decay
and the newer accelerometer data, suggests that the the absolute calibration
of the (ensemble of) mass spectrometers and the assumed drag coefficient in
the atomic oxygen regime are consistent to 5%. While the time may soon be at
hand to base a model on both the mass spectrometer and drag data, puzzling
disagreements in detail still remain at both high and low altitudes.
This study illustrates a number of the reasons for the current
accuracy limit. There appear to be sizable differences (order 20%) in overall
absolute values between some mass spectrometer missions (e.g., OGO-6 and DE-2)and most of the models and differences on the order of 10 to 15% between mass
spectrometers and drag measurements under certain conditions. This
illustrates the importance of reliable calibration techniques for mass
spectrometers and for the drag coefficient (especially when atomic oxygen is
not the major constituent) and the need for measurement accuracy at the level
desired for the model. There are trends still existing between certain data
sets and models with respect to FI0.7, Ap, and annual/semiannual variations.
These trends are apparently different from data set to data set and are
presumably the result of ancillary geophysical factors, which may or may not
be easily found and taken into account.
The largest variations in total density in the thermosphere are
already accounted for to a very high degree by existing models. In
statistical terms, more than 90% of the original variance in latitude, local
time, FI0.7, etc., are explained by existing models. The primary variations
were already well known at the time that J65 (Jacchia, 1965) was formulated
and this must explain in large part why progress has not been rapid in the
ensuing two decades (except in related areas like temperature, composition,and wind). Progress will likely 'continue to be modest in the future, although
there are areas with greater potential for improvement such as where we stillhave insufficient data (like the lower thermosphere or exosphere), where there
are disagreements in technique (such as the exosphere) that can be resolved,
or wherever generally more accurate measurements become available.
3O
References
Barlier, F., J. L. Falin, M. Ill, and C. Jaeck, Structure of the neutral
atmosphere between 150 and 500 km, Space Res., 13, 349-355, 1973.
Berger, C., and F. Barlier, Response of the equatorial thermosphere to
magnetic activity analysed with accelerometer total density data.Asymmetrical structure, J. Atmos. Terr. Phys., 43, 121-133, 1981.
Boudon, Y., F. Barlier, A. Bernard, R. Juillerat, A. M. Mainguy, and J. J.
Walch, Synthesis of flight _esults of the Cactus accelerometerfor accelerations below 10-_ g, Acta Astronautica, 6, 1387-1398, 1979.
Carignan, G. R., and W. H. Pinkus, Ogo-F04 experiment description, Tech. Note
08041-3-T , Univ. of Mich., Ann Arbor, 1968.
Carignan, G. R., B. P. Block, J. C. Maurer, A. E. Hedin,, C. A. Reber, and
N. W. Spencer, The neutral mass spectrometer on Dynamics Explorer, Space
Scl. Instrum., 5, 429-441, 1981.
Champion, K. S. W., and F. A. Marcos, The triaxial accelerometer system on
Atmosphere Explorer, Radio Sci., 8, 297-303, 1973.
Cook, G. E., Satellite drag coefficients, Planet. Space Sci., 12, 929-946,
1965.
Hays, P. B., T. L. Killeen, and B. C. Kennedy, The Fabry-Perot interferometer
on Dynamics Explorer, Space Sci. Instrum., 5, 395-416, 1981.
Hedin, A. E., A revised thermospheric model based on mass spectrometer and
incoherent scatter data: MSIS-83, J. Geophys. Res., 88, 10170-10188,
1983.
Hedin, A. E., MSIS-86 thermospheric model,' J. Geophys. Res., 92, 4649-4662,1987.
Hedin, A_.E., H. G. Mayr, C. A. Reber, N. W. Spencer, and G. R. Carignan,
Empirical model of global thermospheric temperature and composition based
on data from the Ogo 6 quadrupole mass spectrometer, J. Geophys. Res., 79,
2215-225, 1974.
Hedin, A. E., J. E. Salah, J. V. Evans, C. A. Reber, G. P. Newton, N. W.
Spencer, D. C. Kayser, D. Alcayde, P. Bauer, L, Cogger, and J. P. McClure,
A global thermospheric model based on mass spectrometer and incoherent
scatter data MSIS I., N2 density and temperature, J. Geophys. Res., 82,
2139-2147, 1977a.
Hedin, A. E., C. A. Reber, G. P. Newton, N. W. Spencer, H. C. Brinton, H. G.
Mayr, and W. E. Potter, A global thermospheric model based on mass
spectrometer and incoherent scatter data, MSIS 2, Composition, J. Geophys.
Res., 82, 2148-2156, 1977b.
31
Hedin, A. E., C. A. Reber, N. W. Spencer, H. C. Brinton, and D. C. Kayser,Global model of longitude/UT variations in thermospheric composition andtemperature based on _%ss spectrometer data, J. Geophys. Res, 84, I-9,
1979.
Hickman, D. R., B. K. Ching, C. J. Rice, L. R. Sharp, and J. M. Straus, A
survey of currently important thermospheric models, Aerospace Corporation
Report SAMSO-TR-79-57, Los Angeles, Calif., 1979.
Jacchia, L. G., Static diffusion models of the upper atmosphere with empirical
temperature profiles, Smithsonian Contr. Astrophys., 8, 215, 1965.
Jacchia, L. G., New static models of the thermosphere and exosphere with
empirical temperature profiles, Spec. Rep. 313, Smithsonian Astrophys.
Observ., Cambridge, Mass., 1970.
Jacchia, L. G., Revised static models of the thermosphere and exosphere with
empirical temperature profiles, Spec. Rep. 3_2, Smithsonian Astrophys.
Observ., Cambridge, Mass., 1971.
Jacchia, L. G., Thermospheric temperature, density, and composition: new
models, Spec. Rep. 375, Smithson. Astrophys. Observ., Cambridge, MA, 1977.
Jacchia, L G., and'B. Slowey, Densities and temperatures from the atmospheric
drag on six artificial satellites, Spec. Rep., 171, Smithson. Astrophys.
Observ , Cambridge, Mass., 1965.
Jacchia, L G., and B. Slowey, A catalog of atmospheric densities from the
drag on five artificial satellites, S_pec. Rep., 326, Smithson. Astrophys.
Observ , Cambridge, Mass., 1970.
Jacchia, L G., and B. Slowey, A supplemental catalog of atmospheric densities
from satellite-drag analysis, Spec. Rep., 348, Smithson. Astrophys.
Observ , Cambridge, Mass., 1972.
Jacchia, L. G., and B. Slowey, A catalog of atmospheric densities from the
drag on five balloon satellites, Spec. Rep., 368, Smithson. Astrophys.
Observ., Cambridge, Mass., 1975.
Marcos, F. A., Accuracy of satellite drag models, AAS 87-552, AAS Publications
Office, San Diego, Calif., 1987.
Marcos, F. A., K. S. W. Champion, W. E. Potter, and D. C. Kayser, Density and
composition of the neutral atmosphere at 140 km from Atmosphere Explorer C
satellite data, Space Res., 17, 321-327, 1977.
Marcos, F. A., R. E. McInerney, and R. W. Fioretti, Variability of the lower
thermosphere determined from satellite accelerometer data, TR-78-O134, Air
Force Geophysics Laboratory, Hanscom AFB, Mass., 1978.
Newton, G. P., W. T. Kasprzak, and D. T. Pelz, Equatorial composition in the
137- to 225-km region from San Marco 3 mass spectrometer, J. Geophys.
Res., 79, 1929-1941, 1974.
32
Nier, A. 0., W. E. Potter, D. R. Hickman, and K. Maue_sberger, The open-sourceneutral mass spectrometer on Atmosphere Explorer-C, -D, -E, Radio Sci., 8,
271-276, 1973.
Pelz, D. T., C. A. Reber, A. E. Hedin, and G. R. Carignan, A neutral
atmosphere composition experiment for the Atmosphere Explorer-C, -D, -E,
Radio Sci., 8, 277-285, 1973.
Prag, A. B., A comparison of the MSIS and Jacchia-70 models with measured
atmospheric density data in the 120 to 200 km altitude range, Aerospace
Corporation report SD-TR-83-25, Los Angeles, Calif., 1983.
Spencer, N. W., H. B. Niemann, and G. R. Carignan, The neutral-atmosphere
temperature instrument, Radio Sci., 8, 284-296, 1973.
Spencer, N. W., D. T. Pelz, H. B. Niemann, G. R. Carignan, and J. R. Caldwell,
The neutral atmosphere temperature experiment, J. Geophys., 40, 613, 1974.
Spencer, N. W., L. E. Wharton, H. B. Niemann, A. E. Hedin, G. R. Carignan, and
J_ C. Maurer, The Dynamics Explorer wind and temperature spectrometer,
Space Sci. Instrum., 5, 417-428, 1981.
Trlnks, H., and U. yon Zahn, The Esro 4 gas analyzer, Rev. Sci. Instrum., 46,
213-217, 1975.
Villain, J. P., Traitement des donnees brutes de laccelerometre Cactus. Etude
des perturbations de moyenne echelle de la densite thermospherique, Ann.
Geophys., 36, 41-49, 1980.
33
Appendix A. Description of coverage files.
The following files contain the results of binning oxygen data used in
generating the MSIS-86 model. There are five files:
COVALLSA.DAT - binning for all mean FI0.7 values.
COVLOSA.DAT - binning for mean FI0.7 less than 100.
COVMLSA.DAT - binning for mean FI0.7 between 100 and 140.
COVMHSA.DAT - binning for mean FI0.7 between 140 and 180.
COVHISA.DAT - binning for mean FI0.7 greater than 180.
Each 80 byte record contains a 6 digit integer describing the bins based on
altitude, FI0.7 difference from the mean, day of year, local time, magnetic
activity (Ap), and universal time (UT), and 12 integers which indicate by a 0or I whether data was available in a 15 degree wide latitude bin between -90
and 90 degrees (e.g., the first of the twelve integers refers to -90 to -75
degrees and the last integer to 75 to 90 degrees). The fortran format
statement was (18,1216). The 6 digit bin code is AFDLPU where:
A : I indicates altitude less than 200 km_
2 indicates altitude between 200 and 400 km.
3 indicates altitude over 400 km.
F = I indicates daily minus mean FI0.7 less than -20.
2 indicates daily minus mean FI0.7 between -20 and 20.
3 indicates daily minus mean FI0.7 greater than 20.
D = I indicates days from I to 91.
2 indicates days from 91 to 182.
3 indicates days from 182 to 273.
4 indicates days greater than 273.L = I indicates local time between 0 and 4.
2 indicates local time between 4 and 8.
3 indicates local time between 8 and 12.
4 indicates local time between 12 and 16.
5 indicates local time between 16 and 20.6 indicates local time between 20 and 24.
P = I indicates daily Ap less than 10.
2 indicates daily Ap between 10 and 60.
3 indicates daily Ap between 60 and 110.
4 indicates daily Ap greater than 110.U = I indicates UT between 0 and 14400 seconds.
2 indicates UT between 14400 and 28800 seconds.
3 indicates UT between 28800 and 43200 seconds.2 indicates UT between 43200 and 57600 seconds.
2 indicates UT between 57600 and 72000 seconds.
2 indicates UT between 72000 and 86400 seconds.
There are 62208 possible bins and the overall coverage in each file was:
COVALLSA.DAT - 13187 or 21.2% for all solar activities.
COVLOSA.DAT - 4838 or 7.8% for low solar activity.
COVMLSA.DAT - 2235 or 3.6% for medium-low solar activity.
COVMHSA.DAT - 6384 or 10.3% for medium-high solar activity.
COVHISA.DAT - 3356 or 5.4% for high solar activity.
PRECEDING PAGE BLANK NOT FILMED
35
_ ? _¢_ _ _ Ir_Z-
Appendix B. Description of histogram files.
Corresponding to each Figure CI and C2 (see Appendix C) an ASCII file
was produced containing informations on comparisons of a single data subset
with the five models. The file names start with HISTRHO followed by anabbreviation of the data set and the lowest altitude. Each file contains the
following information:
Data set name and model name;
Day interval;
Ranges for Ap, altitude, or other factors;
Number of points and overall standard deviation;
Root mean square error;
Title (unscaled histogram);
Two column list giving interval start and fraction of points in
interval. -99999 and 99999 used for data beyond -I to I;
Data set name and model name;etc.
The following is a partial listing of file HISTRHOAECMESA120:
AEC MESA
DAYS 73358
MAGNETIC INDEX (Ap)
ALTITUDE(km)NUMBER POINTS =
RMS ERROR = 0.1919779
UNSCALED HISTOGRAM
RHO - MSIS-86
74334
-99999.0000 0.0000-1.0000 0.0174
-0.9800 0.O000-0.9600 0.0174
-0.9400 0.0174_0.9200 0.0000
-0.9000 0.0174-0.8800 O.0174
-0.8600 0.0174
-0.8400 0.O000
-0.8200 0.0348
-0.8000 0.0174
-0.7800 0.0174
-0.7600 0.0348
-0.7400 0.0000
-0.7200 0.0000
-0.7000 0.0348
-0.6800 0.0000
-0.6600 0.0000
-0.6400 0.0000-0.6200 0.0174
-0.6000 0.0000
-0.5800 0.0000
-0.5600 0.0522
-0.5400 0.0174
5746
0.0000 10.0000 INCLUDED
120.0000 200.0000 INCLUDED
STD DEV = O. 592570 AVERAGE = 0.1072O4O
37
PRECEDING PAGE BLANK NOT FILMED
38
-0.5200-0.5000-0.4800
-0.4600
-0.4400
-0.4200
-O.4OOO
-0.3800
-0.3600
-0.3400
-0.3200
-O.3000-0.2800
-0.2600
-0.2400
-0.2200
-0.2000
-0.1800
-0.1600
-0.1400
-0.1200
-0.1000
-O.O8OO
-O.0600
-0.0400
-0.0200
0.0000
0.0200
0.0400O.060O0.08000.10000.12000.1400
0.1600
o.18oo0.2000
0_2200
0.2400
0.2600
0.28000.30oo
0.3200
0.34000.3600O.38O0O.4000
0.4200
0.4400
0.4600
o.48QO0.50000.5200
O.54OO
0.0870
0.0348
0.0174
0.0174
0.0000
0.0174
0.1392
0.0000
0.0348
0.1566
0.0522
0.1044
0.0696
0.1740
o.31330.2785
0.4525
0.5395
0.7832
0.9572
1.3401
2.0536
2.2973
3.1326
3.6199
4.1072
5.4647
5.6387
6.0564
5.89985.8127
5.9172
5.30805.3777
5.0818
3.9680
3.7591
3.3415
3.1500
2.7323
2.4365
1.8448
1.5837
1.0964
0.9050
0.6961
0.5221
0.4351
0_4699
0.2436
0.1392
0.1218
0.0696
O.O87O
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
99999
.5600
.5800
.6000
.6200
.6400
.6600
.6800
.7000
.7200
.7400
.7600
.7800
.8000
.8200
.8400
.8600
.880O
.9000
.9200
.9400
.9600
.98OO
.0000
O.
O.
O.
O.
O.
O.
O.
O.
O.
O.
O.
O.
O.
O.
O.
O.
O.
O.
O.
O.
O.
O.
O.
0348
0522
0348
0348
0348
0000
0174
0522
0174
0000
0348
0174
03480522
0696
0174
O522
0000
0000
0696
0348
0696
0000
39
Appendix C. Description of data comparison and coverage plots.
An example of the of the data coverage and model comparison plots
which were prepared for this project is shown in Figures CI to C34 for the
case of the Jacchia drag data 200 to 400 km subset compared to the MSIS-86
model. Some plots for other cases do not have all the information described
here because of changes made as the project progressed. These plots are from
a general plot program and some of the information is not of importance for
this project.
In the following descriptions the term data residuals indicates the
logarithm (base e) of the ratio of measured to model densities (or measured to
model difference in the case of temperature). The residuals are generally
averaged into appropriate bins and plotted as an average with an error bar.Error bars show the standard deviation of the data within each bin and the
plot number indicates the logarithm (base 2) of the number of points in eachbin.
On each plot the data set acronym, measured quantity (RHO means total
density), and model are given at the top center of the plot. At the top right
are the plot date and data file name. There are generally three lines of
compressed information at the bottom of each plot with the following meaning
(* indicates particular relevance for this project):
GRP - local data set identification number
* ALT LIM- altitude limits for this plot
ISK - number points skipped for extreme deviation
* SEL - if first integer is none zero data were limited
by the following two numbers. 7 indicates
a test on magnetic activity and thus Fig. AI
is for Ap from 0 to 10 and A2 covers Ap from11 to 400. 2 indicates a test on latitude.
* PTS PLT - indicates actual number of data points
involved in this plot.* AVG - overall average of logarithms of data to model
ratio for this plot.* SD - overall standard deviations of logarithms of
data to model ratio for this plot
A - binning intervals for abscissa and ordinate
if applicable.SW - MSIS switch settings
*DI,D2 - date limits for plotted data
ST,SX,SY - points skipped for SEL test, or beyondabscissa or ordinate limits.
SM - smoothing factor in contour plots (1.00 is none).
* DATA-MODEL - residual plot using given model
identification number or zero for no model.
* SD:DATA-MODEL - contour plot of standard deviationsof data residuals within each data bin.
* AVG,SD - for binned data, average and standarddeviation of binned data averages.
AVXL,AVXH - points beyond low and high abscissa on
histogram plots.
PR]_CEIr)[NG PAGE BLANK NOT FI_LMED
41
* RMS- root mean square error for all points (fromAVG and SD values).
* DSD,MSD,R - for binned data, standard deviation of
bin averages of raw data, standard deviation
of bin averages of model values, and correlationcoefficient between data and model based on
bin averages.
42
;0
_0I
O_
I
0"l-
(J
"5
°o
0.--
,,.-,,
'_I
r,.
I ° I ' l
00
0
O
00
l I
0,I
0
HONHH_N300
O
0!
00i
oT
0
n-
F-
Z
_3
U
0m
.PO0
..J
°
00
_X
,..1
_ •
oo o
IN •
._,._
.,,2,.
°CO--PO
o
0
tO
"00
°,..I
o
tJ_- .
O''--_J v
0
.,_0 ._J0 0
0 _0
0
4J ._J
bO_"
o
o
•,.._ _'_
,,,..4
• -- 0
• _
43
.III
O
ICarl,.
I
O
QE
_EQ_O
o,
Zo°
_°°
Ld_h'--hle_
hl Ih-iQ
44
_0
I
o
o.it)
o
1 l 1 t
HONHH_330 %
.._
0
,--4
:_ .
___
0
(I_"_'_
OO
_.) _ '
b0_O
3:O e"e" _0
e"
_ 0
0
• _
@@@@
@@
O
@@I
D.w
w.-
I
0"l-n,
0
(.)
..}
m
.o
m
/y.o
W
co:) 0Wl Q
O.
I II _ Il I
I _ II I
I
- !
! _ |I !
! !
I I
I, I
I - I
!
! !
1 " I
I !
! !
I I I0 0U'_ 0
• i
0 0
OI±Y_
I I
@11 1SN3CI B 0 1
0LO
0
0p-
o
I--
0
I1)
,-I
,--4
•.-4 t_
0
4-) ,.-4
o,.--t
°_.,_
o e,,
_u
,,,,.4
_('xl
_d_,,--
•,-_ u'_
e-
¢_3 .,.-i
,,.-4
r.. ,_
45
'CD,CD
O
ClO
O
,.<
O
I(/$
I
O
n,'
o,
Z°.
IF*..
W
Wl
46
i I J !
!
I --r I
I I
! - I....I - I
I
!
I I
i I
!
! I
!! _. |! !
!
I " I
I - 1! !
! , !
I
i - i!
I
!,- I! !
!
,,, z 1 l !0 0 O
0 0
OIIVH AI ISNH(] S 0 l
I0LO
0I
! 0_
0
UFJmrI
0
0 t
u
m
!
F_
.JO.
Q.
rn
['--
_w
i--
,im
?0
I--
F_
pW
_1__
...J ,-
I--
X
?
0
_D
*,-I
0
.._
• bO
,Lf%bO
-,._.,-4
0
_9.t,')
_D
nO
(0nO
I
(/')
I
O.-rn_
o<
0
0<
_.°,-,e-)
fyo.
W
Gexo
Wl
"(Ii-e,,-
O.
0
0
m
, i
0
0
OIIVH
,,,,.,,.,
l , I ,.._ 11I • ' I
1 " I
I
!
I I
I I
! I
I ! .
• I
i i
I!
1
' " I!
I - l
! - 1
: - I
I
!
I I I , '1,,oQ
O
AIISNH_
I i
,.,J I
0UO
0
I
a
_0l
0
0
m
I
_W
O0
<),-
N_
e. m_o
I --o_
0 "- "
• .
I-- •
-..4
>.,,=4
O.,,..4
e_
,--t,--4
0
_d
O e')
m E
,-4 ,--t
°,-,4 ,--4
(1_ e.
e_ e-
• L,
47
cO00
_)
I,')
Cl
O
m
a)I
I
0-r
0
n-O
0,<
i
Z..
n,-..wOO
0oo OWl O
'<In-.,.,<,.-G.
48
' I ' I i I
I IO
O
OILVN
I _ II I
• I
I _,. !!
I
I "" I
I
I I
I m I
I I
I • I, 1 1 I
0 0
I
@AII SNHO _ o I
w
n
u
w
Ow N
91111
m
C)
Oh"O
m
!
_,
O
_ l,.hl
O ""CO
I--
©_"_"" ,,,: cNO ,r,o
0_.0
- ._ iM <>.
..1 _x
TI" if) - °
0 _-- •
Oe,_
o0 °
0
_ <
• D,-
¢N " _
'=,.:
_,.:T
.,-',1e-
e-
,-i
e.
r..
t-
0)E
,-t
e-
r._ r-.
o
@0a0
00
r.¢3
ooo
_oI
u1
ul:E
I
o-rr_-
C_
n-C_
...j
f1_°°
bJO0I--,--bJ
U_D
b.Ld
"CI
G.
00
i
I_ II
| ,P, I
I _ I
I I
|, I
, I
'" J
I - I
I _ I
I I
!
Iob_
o
OIIVH
, Ioo
o
AIISNHO 80!
i I
! !0
oI
0
0
o
0
I
0
CO
0 ,0
m
I
_w
.°_
_QD
* 9'_ CNI
o
>.
_. .r3=l ,_ x
_. _N_.-
U.,
&
• _,)
i-...i,_
I--.J ,=. _
°_,-I
r,.
t_"0
r_
0.-4
t- e-
"I_ .,-_
m _
,,,,4
r,.
49
0_
o
0oI
N
tm
I
o-e-n-
n-
-=_
,°
Zo,
w
Wl
G.
SO
0
0
N
I ° I ' I '
I0ttJ
0
01±V8
I .
F---4
I h,_ !
I I
t--t
n ,.,. _
I I
I
I
I "^ 1
I
I -
I J,_ II I
I 1 iOQ
O
AIiSN30 a_0l
] I
0U_
0
I
0
0
m
' I
oo
0
bOI
.XL_D
0 ZO
L_
_JZ
0
0
o._,,j
o
CDL_J
)-
°
CDU')
0. _ °
)-
_. ._x
,=.1
n t_
_,_ ,
IN_=
x
i"l
_ N
i'-
e-
°_..I
O
..,,4
._=1
v
.t_O.,-_
"00
•,.-I 0
0bO._
_._
•.-4 _4
m _
o
• L
e.,
0,..i Q
O
00aO
P,%
O
O
aOiCn
(n
I
o
n-
O
O
O
_°°
n.,°°
We%
Ual
_L
OO
m
! i I
_H
OUO
(D
|I
!!
!!
i
AIISN3_
#!
_...J
I
I
!!
OOe4
i
O
_m
O
Om
UO
u
<
X
"7
O
F--
Z
.<
_W
(n
?
a0
mG.
n
_'%¢N
_6
)-(n
X
---8_,-,
N
_,-:.
O_
l'q •
O
W
_1 •
.-4
e_
e-
ra
O
"O
_.,-4m_u
>
e_ U_: h0v _J
"OM
"O_D
•,-4 O
o.-40
C_0(D
_ N
,--t
•,-I O
5]
e4
°°_M.
o
I,(n-
t)
I
0"l-
0,<
0
0
,._°°
_°.
w
Wl
t_
52
00
0LfJ
0
011VH
oo
Al ISNHG a 0 I
OU_
O!
U9
,q,n
Coco
e4m
(D
OU)O
m
!
n_.<UJ>-
w_
_Lu
(n
(n
I")
>-(n
an(n
.Q
_DCnw_
,n--o4
'w"
w
X
_,-.j
°_
o.,_m
_ °
....4
f¢3
o
v
C__D
0
>
L
m(D
W>
-,=4W
6_D
.-4
r,.
p_
0
mm0
I
n-
I
0"!"
0<
,<
.0
w_._°°
W
Wl
<I
0.
oo
ou'Je
o
OIIYN
oo
o
J,_I_ISIq3CI _ 0 I
I0u2i
oI
I
u
:w
co In.l>-
(Di
UO
0 (o0
m
!
I,'I
_ud
o°
(n
0-,-
T_
)-
'q'X
t'NI--
.°_
=r'l__ •
,im ,
0o,..J ,_w
IN
0
(/I
• C.,-4
r_
>-,
C_
0
m
r.m
c_J
,-4
bo
>.
K,
o,):>
,-.4
.,..4_n
K,
d.3m
4_r..
53
ID
p_
OC:IP,4
+.<
%
I.<
i,,,.,i
I
O""r
O,(
0
U,<
._.°°
t,Y.°
w
Wl
<In-..-
0,.
54
I ' I ' I
OIIYN Xl ISMHG B 0 1
p,,,,
.<LI.i>-.
(D-J_ I.i,I
I,-,
_<
0
pw.@,-
0
_X
I,,-.. (j')
.JeLI¢_
(IDf,n _")
0,.1_
/40 .
_ •
°° ,
w
cM •
-1-
..__.<.
_._
• .. v-
i'- °
.,,.'4
"0
t_
,_
o
Q_
L_(.-.j,,.00
.,..,i
b_0"J
>,
,--I
Q_
C_
r._
,,,',=4
bO
(D
.,-4
| °P._
"_ 0
•_-I 0
0
('000'1
0 0
e" "l0-_
L0
0
e_
• L
55
==
o
oo
I
o::z:o,.
o,<o,.
o,,<,-j
L,).,I@,.
..@,,')
tY°°
w
£rxo
Wl
<I
,,,(--0,.
00_,
iiii i
i I ' I ' i '\0 )
o
l I , I z I ItO 0 bO
"_ 'qr
i
HOnlIiV7
56
Z _e_
.
3-_ •
N
°
• _
._'-
/
'
0
c-O
I
•,-.i e-,
0_'o
0 ""_bO
t. 0
04J Q;
00 bO
_'00
4J t','5
•_ p-)
"0
"OJo
j..)
• bO
,--'0
• m--
.,.__._E:_ -,'_
ii
COv-
P'i
(3
Q.=II
IO00Iin
In
I
0
(.9
_,,_-_
(J
_0,
_o0
W
G.
0
' I ' I ' I '
J_.", _. .,k=o f
. 00_ _ I o_./
, ItO
i I0
o._0_,
i I i
,q=
I
HO_iIiVq
0
(D
!
Oi
CO
4m.
._i
_W
I,-,• ..iC _
CD,,-
00
QIO_
Nff_me_
_.
0
-F-°_
tJ
..I il.-(#i
el
,q.
• 1,3
_,-_ol
< ,_.
57
o
o
I
m
!
o
o-
o.,(n_o
o
..
b_0
O
O-
58
, I , I[.ca o,q-
0
HGnlIIV7
0O_
!
qT
OJ
(Dm
O_
&.
.J
.<(D0-J
o
d
..,'__
Qp W)
>.
...d
Q.. P._
_ •_ °
";,.:
_ •
IN oI_'°._..
..3_
4-)
0
o_.-¢
o0
I
.,-¢
cO e-.
e- N
O o._ O
@
C OO e.,
,,,,.,4
I1)
,,.- ._
r.,,%
QDOOI
(,9
In
I
07r.n-
n-t%
0.<
bJ_9
--e9X.C4tY°0
bJ_F-,--W
bJl
Wb.I
<1
o,.
0
t
l I l I l I I
I
30_111V7
,,---I
0
¢0
o. •
u')
q5")
CD
&f) - .
I-.,.J
o >- ®_.
.o
..J_
hi
.(4
• ,(
m "J
i"
3,2,.
I _" ..
8
"0
0
r_
L
0
t_
,--irk]
OC'_I
_ ,,.--t '
59
0
II/"lMIn'nl
I
0n-
o
°°
w
Wl
Q.
6O
0
)
l I i
!
I
{o
#e
I I , I z I ,_r ,q'
I
3onlllVq
OO
trJ(0tD
D-.
D-.04
O
tnO4
(D
n
OO
O(D °!
n-<Ed>-
b.O
>-<r_
C_P.-
.._*
_QP
°m.-O
.. o
,IDX_-
,.J
o. o
..,.: __._.
• X
.,,2_
F- •.J ,-- _-
®,.:?
_ ..
o
t_
(P
0
I
,,-i
..ira
@Oq0('_1
•,-I _
•i_ Q)
_ -,-I
e_
._,)
Q
l
!
o
e_r_
o
,°
Wl
G.
I ' l ' i J
0
I , I , I ,LD 0 LO
I
hGfliIiV1
0
!
00
0
m
00
0
0
_w
.._
"-,2_O _
_O
if)
m
"_ CmN
_m
X R_ _
rn _N_
/4
.d
O_
®,.._
2._,,4
e"
O._,-4
e"NI
e"
_00 _-
o ¢)
O_0 .,-I
bO
t__OLI_t_,--
._._
rau _
61
o
II¢1
¢/1If
!
0
el-
",C
0
,o
_.°
wl
¢1
62
m
m"
0O_
' I _ I ° I
_5/_ \__.
00
, I I z 1 zm Q U'J
,q.!
H0nll±vq
0
CO
!
0
0
OJ
LO
O0
0
0
d
ot.
•. _'_O)
00
v- •
OQ
Om
_o)-
Q. ff,l ¢'._
Zd_ _-
e4
•
_-._
I.-
0
0
I
0
.-1
_de-,
e_e.-
o_,I
e...
o,1:1
-_ i1)
hi) 4._
r,.-_
o
o
0DI
N_
In
I
0
0
n_t'-_
Uq["5
,o
_°°
_i,,°° '
w
¢=0 OWl 03
<Irr,-
_k
' I ' I ' I
I ! l
/
I I0
O_ f
1
I
HG_II&VJ
O
I
00
(D
00C:O
(D
U
!,-
O
O
t.D
e4
o
mt_
;4®din
_o
(.4
>-In
_x_(nn-
..I
nt..
I
CDoo
ii#l
"l-
ll._ _m, ..J
I I_' i,D
,l',,,i
33
._,_
O
(n
61
e-°,,,l
'13
O
-,-4
bD
!
(D_D"I
.,..4
,-4
o_e-
e-, m0
,-4
•,-i I._r...-
63
O0
0
0
I
f.h
I
0"r
0
0,<
.o
.,_oo
iYoo
W
<In,'...
0,..
64
I ' I ' I '
i
0o")
\
ii
CO
/
}
= • \
f
(,0
b
IL_
Il I .... I
0
30171iIV7
• 0 _-
I ILO
!
0
I
00
CO
00cO
(D
LJ
00 ""
F,=
O(3(D
0
o
,C",4
,o,ID
Ill •lJ _"
"•"--J _ ("'4
• e_")
e,,
0
I
,-q
I=0
r.,O_.,
0
0
_'_.
,_,,I
@
_=o
I
o=r,,
==
<1
n
O O O O _
O O O O
O _ O
I I
_OnlIlVq
,
I,'3
=4"
_1.u)
e.:
e_.:
o°,m
_ •
,<
O
_3,-4O
E
ID
O,-I
t.
._,-(
"O
3=
O
_0
¢.
>Oo
(D
rn
r.,._
65
0 -
eo,
00I4(
N
I
0
e,-
o
nFo
(J,<
Z • •
w
Q_Wl
,<I
,(,..Q.
66
0
0
O_
' I
I
.<_J>-
OR_a,,,L PAGE IS
OF POOR QUALITY
¢=
,<
t,o °
S®ta3
ol
>-
_ o
¢',i o
I---
._
(N
:r,
.o
• W
Oq
°.,r-
v
_N
,..J .-
m._,<
,4_,,) ,p
0
r_
0
r_
0,--i
.,'-'1
0
_0
>,0o
•_ _•_ ._._
-,-(_-I(I) -
.1 c_
=I" f"
o_
r,.,_
ORI.I_N_L P_GE IS
OF POOR QUALITY
0
0 ,4
l.w,.oO
OOO,i
0
(DcOI
I
0"rn-
0
r_0
°,
Z°o
Wr_
W!
Q.
I ' I I
0 0 0(D 0 0
0 In 0(D "q"
30nlIlVq
l I
oo
L0
I '
u
w
0
u
(7}
90
I
n_.<Ld
>-
_, •
0)
0
)-
'_ I/'1
_O
3,-w
I'q
v
6
0
0,--I
0
(I)bO
(D
00
I.
(D
{D
>d
.Io ,---I
t'N _
° e..._
67
-+
<,
o
I
<_
hl
t.+'X,O
Wl
hl.l
+'<Ip_',+..,,<,,--
68
(saq) HNI± 7V007
0
,-..40
0,,....I
e-.
e.
-_.-I
e-
._.,_
0,--4
K.(D>00
(D
>
md
,--t ,--t
0._
.._
._._
r..,_
_o.
Q.
!
0
69
•.,_,=_l_. PAGE IS
OF ' POOR QUALITY
,,
o,..
ua_
t_
t., I
¢1.
7O
00aO
00
r.
O
..=<
O
U)anI
V)
U'):E
I
O-riX:
O
n-r_
--)
.o
._P..
:X.e4 *re.,..UmO
WC*
U_D
LUl
LU_J
,.,,..-,_..-O.
' I ' ie,i
' I '
e4 _ ._. (7, _ _- (_. u') (%1.e
,ql- _ _ .el- _- l'_ _') .- ,.-
i.-) e.i i,_ ,.1. .e
e4 u") ,,e eo (_ P,,,.
(__
01,eOO1.
1O0'2
I,,')
HOnlllVq
00 _ O_
w) P- O_ w_
U'% ',e 00 00',11"" e_l e,i e4
it) _ e4 e,,I
_D e,_ ,,-- e4
_" ,,-- e4
r_ _ _
LO
!
OCO
Om
(D
u..Ic_
[--.,-,-,',..'9ZO-.I
I
(D
O--
I
._,')
.:Z
._,,.I
•,-.I e-
m.lID
>- O >_(n m.m
_X _ _
w) (D
6__ - _
(D
N >m_ O'-"
O
,...J ,_ r"U,I 0 m
.__ e"_ °,-I
e_ . > 0
•e --I O
o,- (_4 L.• w) • (I)
• _
,J,-
q[ ,--.W5
7]
r.I'%
O
.._
O
Iu_
U_
I
On-n-
O
-/w°°
I--,-Wr_
W!
'_1
_,-Q.
m
0
' 1 ' I ' I !
0N
_m
72
I , I
I
, Iin
(Dm
in
U_
m I.-..I
<
O
OO
I
I¢)
.Z
_O
)-(n
_(n/w
-.I
;4• °
.o •
o )(
"e
I-- •
,( w)
.,._.-4
.,._ I=
.,-I
(D O
(D ,--_,O
_'edO
" _D
.,-_
.,-4
O,-4
(D
IM) _ •
>
O1/%¢_,--
_XZE O
,-4
OO
m o
o
o.m0,%
(._
0
co
I
/J)
!
0"rn,-
(._
oo
,7..
wIZl
Wl
'<1
_..
m
_ Ie) _ -_- _. _ it)
_'_ _1" _1" (_1 _ C_l ' _1
o
W
T , l , IL_ 0 LO
I
HQ_lllVq
m
,ee4
04
V)
04e4
I
.,=,
_ __ _ .
• _ _'-
_:- _
_,_. _
"8.
i'- m
0 o
!
73
I
(n
I
0n-e,,
0
e_0
0
.ij
'7.°
W
Wl
<1r,,_-
74
,r--
0
O_
' t ' I ' 1 '
eo
t,_ (',,4 ,- ,p
f,O _ _ _ f.D It)
, 1 , I
HGn±I1Vq
•e _'% -- ,-
l ] lU_
!
0
m
I
00
00
o
.Z
.1"3_r_
on
m_0
(oaO
!(/11
, ,p,q(J%=E
I
0"r
O
(E
O,(
Zo°
rr'°°
hir_
n
' I ' I
, I l I0 _0 0
3QQIIIV7
I ILO
I
r_
I0
I
0O
_D
(30
gO
(J
o_C5 "-"
I---
00
0
"Z
1_ ¸ .,,.4 o,.,.I
°_ _
._,. g8
I0 - - J:_
(/1 i_1 _-_ j= .,_
_ " 0 •
o._o
°o
(/I _ _ -,-_
X _ --_"'"> :3 0
--; g_
-_IID
• F_ mo
8
175
Appendix D. Data set comparison summaries.
The following pages have a stum_ary of the data subset comparisons with
the models emphasizing observed trends for the different models.
PRI_CEDING PAGE BLANK NOT FILMED
"/7
Altitude: 200-400FI0.7:70-190
Parameter: Total Density Dates: 61001-70365
Source: Jacchia sat. trk. Data File: DRAG
Best quiet mean: -.03 (J77) Worst quiet mean: -.21 (MSIS-77)
Best quiet SD: .15 (MSIS-86,J77) Worst quiet SD: .18 (MSIS-77)
Best active mean: .02 (J77) Worst active mean:-.16 (MSIS-77)
Best active SD: .18 (MSIS-86,J70) Worst active SD: .21 (J77)
Altitude: All equivalent.Residuals flat.
Latitude: MSIS-86
Residuals flat.
J70 least variance.
Local Time: MSIS-86 -77 least variance.
Residuals have terdiurnal variation for J77.
Mean FI0.7: MSIS-83 least variance.
Residuals flat.
Delta FI0.7: MSIS-77 least variance.
Residuals more positive with increasing activity for MSIS models and
more negative with increasing activity for J77 and J70.
Magnetic: MSIS-86 least variance.
Residuals near equator and poles flat except for MSIS-77 and J77.
Coverage to Ap 150 near equator and Ap 100 near pole.
Day/year: J70 least variance.
Residuals have annual, semiannual and longer trends.
Longitude/UT: Variances similar for all models.
Comments: Little overall difference between models except worse
for MSIS-77.
MSIS-86 trends: delta FI0.7 and annual.
78
Altitude: 400-800FI0.7:70-190
Parameter: Total Density Dates: 61134-70364Source: Jacchia sat. trk. Data File: DRAG
Best quiet mean: -.05 (MSIS-86,J77) Worst quiet mean: -.23 (MSIS-77)
Best quiet SD: .25 (J77) Worst quiet SD: .29 (MSIS-77)
Best active mean:-.04 (MSIS-86,-83) Worst active mean:-.16 (MSIS-77,J77)
Best active SD: .28 (J70) Worst active SD: .31 (MSIS-77)
Altitude: J77 least variance.
Residuals flat but irregular for all models.
Latitude: J77 J70 least variance.
Residuals more negative at midlatitudes for MSIS-86 -83.
Local Time: J70 least variance.
Residuals have diurnal and semidiurnal for all models.
Mean FI0.7: J77 least variance.
Residuals irregularly more positive with increasing activity for MSIS
models.
Delta FI0.7: J77 least variance.
Residuals more positive with increasing activity for MSIS models and
more negative with increasing activity for J70.
Magnetic: MSIS-86 least variance.
Residuals flat but irregular. Coverage to Ap 150.
Day/year: J77 least variance.
Residuals have annual, semiannual and longer trends.
Longitude/UT: MSIS-86, J77, J70 least variance.Residuals may have second harmonic at lower latitudes for all models.
Comments: Little overall difference between models except worsefor MSIS-77.
MSIS-86 trends: latitude, local time, mean and delta FI0..7 and annual.
79
Altitude: 800-1200FI0.7:70-190
Parameter: Total Density Dates: 61134-70364Source: Jacchia sat. trk. Data File: DRAG
Best quiet mean: -.03 (J77)
Best quiet SD: .22 (J70)Best active mean: .02 (MSIS-77)
Best active SD: .23 (J70)
Worst quiet mean: -.05 (J70)
Worst quiet SD: .30 (MSIS-77)Worst active mean:-.12 (J70)
Worst active SD: .33 (MSIS-77)
Altitude: MSIS-86 least variance.
Residuals irregular but flat for all models except more negative
at higher altitudes for J70.
Latitude: JTO has least variance.
Residuals more negative at midlatitudes for MSIS-86
Local Time: JTO least variance.
Residuals have diurnal and semidiurnal for all models.
-83.
Mean FI0.7: J70 least variance.
Residuals flat except more positive with increasing activity for MSIS-77.
Delta FI0.7: MSIS-86 , -831east variance.
Residuals more negative at low and high activity for MSIS models and
more negative with increasing activity for J77 and JTO.
Magnetic: J70 least variance.Residuals near equator and pole irregularly more negative with increasing
activity for all but J70. Coverage to Ap 150.
Day/year: J70 least variance.Residuals have annual, semiannual and longer trends.
Longitude/UT: J70 has least variance.
Comments: J70 best at this altitude.
MSIS-86 trends: latitude, local time, delta FI0.7 and annual.
8O
Altitude: 120-200FI0.7:90-190
Parameter: Total Density Dates_ 66362-73057Source: Barlier sat. trk. Data File: DRAG
Best quiet mean: .02 (MSIS-86) Worst quiet mean: -.07 (J77)Best quiet SD: .22 (MSIS-86,J77) Worst quiet SD: .23 (MSIS-83,-77)Best active mean: .00 (MSIS-86) Worst active mean: .13 (J77)Best active SD: .21 (J70) Worst active SD: .23 (J77)
Altitude: J77 J70 least variance.Residuals more positive at lowest altitudes for MSIS models.
Latitude: MSIS-77 has least variance.
Residuals flat.
Local Time: MSIS-86 -77 least variance.
Residuals have diurnal and semidiurnal variation for MSIS-83, J77,and J70.
Mean FI0.7: J77 J70 least variance.
Residuals flat.
Delta FI0.7: Variances similar for all models.
Residuals flat for all models.
Magnetic: Variances similar for all models.
Residuals near equator more negative with increasing activity for all
models and no clear trend near poles. Coverage to Ap 200 near
equator and Ap 50 near poles.
Day/year: J70 least variance.
Residuals have annual, semiannual and longer trends.
Longitude/UT: Variances similar for all models.
Comments: Little overall difference between models but local time
and latitude variations somewhat worse for Jacchia models.
MSIS-86 trends: altitude, annual, and magnetic activity.
8]
Altitude: 200-400FI0.7:70-190
Parameter: Total Density Dates: 64033-73058
Source: Barlier sat. trk. Data File: DRAG
Best quiet mean: .02 (MSIS-83) Worst quiet mean: -.06 (MSIS-77)
Best quiet SD: .20 (MSIS-86,-83) Worst quiet SD: .21 (MSIS-77,J70,J77)
Best active mean: .00 (J77) Worst active mean:-.09 (MSIS-77)
Best active SD: .20 (MSIS-86) Worst active SD: .23 (J77)
Altitude: MSIS-77 least variance.
Residuals more positive with increasing altitude except for MSIS-77.
Latitude: Variances similar for all models.
Residuals flat.
Local Time: MSIS-86 -77 least variance.
Residuals have terdlurnal variation for J77.
Mean FI0.7: MSIS-83
Residuals flat.
J77 leastvariance.
Delta FI0.7: MSIS-86 least variance.
Residuals more negative with increasing activity for J77 and J70.
Magnetic: MSIS-86 least variance.
Residuals near equator and poles more negative decreasing with
increasing activity for all models. Coverage to Ap 200 near
equator and Ap 150 near poles.
Day/year: MSIS-86 least variance.No obvious trends.
Longitude/UT: Variances similar for all models.
Comments: Little overall difference between models.
MSIS-86 trends: altitude and magnetic activity.
82
Altitude: 400-800FI0.7:70-190
Parameter: Total Density Dates: 66011-73058Source: Barlier sat. trk. Data File: DRAG
Best quiet mean: -.04 (J77) Worst quiet mean:
Best quiet SD: .31 (MSIS-86,J70) Worst quiet SD:Best active mean: .00 (J70) Worst active mean:
Best active SD: .29 (MSIS-83) Worst active SD:
.15 (J70)
•33 (MSIS-77)
.11 (J77)
.32 (MSIS-77)
Altitude: J77 J70 least variance.
Residuals flat but irregular for ali models.
Latitude: Variances similar for all models.
Residuals flat for all models.
Local Time: MSIS-86, -83, J70 least variance.
Residuals have diurnal and semidiurnal for MSIS-77 and J77.
Mean FI0.7: Variances similar for all models except MSIS-77 worse.
Residuals flat for all models except MSIS-77.
Delta FI0.7: MSIS-77 least variance.
Residuals more negative with increasing activity for all models.
Magnetic: MSIS models least variance.
Residuals near equator and pole more negative with increasing activity
for all models. Coverage to Ap 150 near equator and Ap 100 near
pole.
Day/year: Variances similar for all models except MSIS-77 worse.
Residuals have annual, semiannual and longer trends.
Longitude/UT: Variances similar for all models.
Residuals may have second harmonic at lower latitudes for all modeIs.
Comments: Little overall difference between models.
MSIS-86 trends: delta FI0.7, annual, and magnetic activity.
83
Altitude: 200-400FI0.7:100-170
Parameter: Total DensitySource: OGO-6 MS
Dates: 69187-71177Data File: OGORHOFIFTH
Best quiet mean: -.04 (MSIS-77) Worst quiet mean: .15 (J70)
Best quiet SD: .14 (MSIS) Worst quiet SD: .17 (J70)
Best active mean: .00 (MSIS-86) Worst active mean: .17 (MSIS-77)
Best active SD: .19 (MSIS-86,-83) Worst active SD: .21 (J77,J70)
Altitude: Insufficient coverage.
Latitude: MSIS-86 least variance.
Residuals more positive toward north pole for all models.
Local Time: Variances similar for all models.
Residuals flat but irregular.
Mean FI0.7: MSIS-86 least variance.
Residuals flat for all models.
Delta FI0.7: MSIS least variance.
Residuals more negative with increasing activity except flat for
MSIS-77.
Magnetic: MSIS-86 least variance.
Residuals more negative with increasing activity for all but MSIS-86.
Coverage to 160 Ap.
Day/year: Variances similar for all models.
Residuals more negative at beginning of 1971 for all models, but
less pronounced for Jacchia models.
Longitude/UT: Variances similar for all models.No obvious trends.
Comments: MSIS models best for standard deviations and overall.
MSIS-86 trends: delta FI0.7 and annual.
84
Altitude: 400-800 Parameter: Total Density Dates: 69179-71177FI0.7:100-170 Source: OGO-6 MS Data File: OGORHOFIFTH
Best quiet mean: .04 (MSIS-77) Worst quiet mean: .25 (J70)
Best quiet SD: .17 (MSIS-86) Worst quiet SD: .21 (J77)
Best active mean: .02 (MSIS-77) Worst active mean: .23 (J77)
Best active SD: .20 (MSIS-86,-77) Worst active SD: .22 (J77)
Altitude: Variances similar for all models.
Residuals more positive at higher altitudes for all models.
Latitude: MSIS least variance.
Residuals more positive at midlatitudes for J77 and J70.
Local Time: MSIS least variance.
Residuals have semidiurnal or terdiurnal variation for J77 and J70.
Mean FI0.7: MSIS-77 least variance.
Residuals flat but irregular.
Delta FI0.7: MSIS-86 least variance.
Residuals more negative at higher activity for J77 and J70.
Magnetic: MSIS-86 least variance.
Residuals more negative with increasing activity near equator for all
but MSIS-86 and more positive with increasing activity near poles
for MSIS-77, J77, J70. Coverage to Ap 160.
Day/year: MSIS-77 least variance.Residuals have no clear trend.
Longitude/UT: MSIS least variance.
Residuals more positive near magnetic poles for J70.
Comments: Little overall difference between models except Jacchia
models slightly worse in latitude variations.
MSIS-86 trends: altitude.
85r
Altitude: 130-200FIO.?: 70-110
Parameter: Total DensitySource: AE-C MESA Accel.
Dates: 73353-75100
Data File: MESA-C
Best quiet mean: .00 (J70)
Best quiet SD: .15 (MSIS-77)Best active mean: .07 (JTO)
Best active SD: .15 (MSIS-77)
Worst quiet mean: .11 (MSIS-86,-83
Worst quiet SD: .18 (J77,J70)Worst active mean: .14 (MSIS-86)
Worst active SD: .21 (J77)
Altitude: MSIS-83 J70 least variance.
Residuals more positive with decreasing altitude for MSIS models.
Latitude: MSIS least variance.
Residuals more positive toward north for MSIS models and more positive
toward both poles for for J77 and J70.
Local Time: MSIS least variance.
Residuals have diurnal variation for J77 and J70.
Mean FI0.7: MSIS least variance.
Residuals more positive with increasing activity for J70.
Delta FI0.7: MSIS-83 MSIS-77 least variance.
Residuals more positive with increasing activity for all models.
Coverage small.
Magnetic: MSIS least variance.Residuals near equator are irregular and near poles more positive with
higher activity for all models. Coverage to Ap 100 near equator
and Ap 130 near pole.
Day/year: MSIS-?7 least variance.
Longitude/UT: Variances similar for all models except worse for J77.
Comments: Jacchia models better than MSIS in absolute values but slightlyworse in standard deviations. Jacchia models worse for latitude
and daily variations.
MSIS-86 trends: latitude, delta FI0.7.
86
Altitude: 200-250 Parameter: Total DensityFI0.7:70-110 Source: AE-CMESAAccel.
Best quiet mean: .02 (MSIS-86)Best quiet SD: .21 (all)Best active mean: .02 (MSIS-86)Best active SD: .20 (MSIS-77)
Altitude: All models have low variance.Residuals flat for all models.
Latitude: MSIS-86 least variance.Residuals flat but irregular for all models.
Dates_ 73353-76110Data File: MESA-C
Worst quiet mean: -.08 (MSIS-77)
Worst quiet SD: .21Worst active mean: .07 (J77)
Worst active SD: .23 (J77)
Local Time: MSIS-77 least variance.
Residuals have semidiurnal variation for all but MSIS-77.
Mean FIO.7: J77 least variance.
No obvious trends.
Delta FI0.7: MSIS-77 J77 least variance.
Residuals more positive with increasing activity for all models.
Magnetic: MSIS-77 least variance.
Residuals irregular. Coverage to Ap 130.
Day/year: J70 least variance.No obvious trends.
Longitude/UT: All models have low variance.
Comments: Little overall difference between models but Jacchia models
worse for daily variation and J77 worse for latitude variation.
MSIS-86 trends: delta FIO.7.
87
Altitude: 130-200 Parameter: Total Density Dates: 74001-75100FI0.7:70-110 Source: AE-C OSS MS Data File: OSSC48
Best quiet mean: -.O1 (MSIS-86,-83) Worst quiet mean: -.07 (J70)
Best quiet SD: .10 (MSIS-77) Worst quiet SD: .13 (J77,J70)Best active mean: .00 (MSIS-77) Worst active mean:-.02 (J70)
Best active SD: .11 (MSIS-77) Worst active SD: .16 (J77)
Altitude: MSIS-77ieast variance.
Residuals more positive with increasing altitude for J77.
Latitude: MSIS least variance.
Residuals more positive toward north for MSIS models and more positive
toward both poles for for J77 and J70.
Local Time: MSIS least variance.
Residuals have diurnal variation for J77 and J70.
Mean FI0.7: MSIS-77 least variance.
No obvious trend.
Delta FI0.7: MSIS-77 J77 least variance.
Residuals more positive with increasing activity for all models.
Coverage small.
Magnetic: MSIS-77 least variance.
Residuals near equator are more negative with higher activity for
J77 and J70. Coverage to Ap 130.
Day/year: MSIS-77 least variance.
Longitude/UT: MSIS least variance.
Co,_nents: Little overall difference between models but Jacchia models
worse for latitude and daily variations.
MSIS-86 trends: latitude, delta FI0.7.
88
Altitude: 200-390FI0.7:70-110
Parameter: Total DensitySource: AE-C OSS MS
Best quiet mean: .01 (J70)
Best quiet SD: .14 (MSIS)Best active mean: .00 (J70)
Best active SD: .15 (MSIS-86)
Dates: 74001-76161
Data File: OSSC2OOS20
Worst quiet mean: .09 (J77)
Worst quiet SD: .16 (J70)
Worst active mean: .10 (MSIS-83)
Worst active SD: .19 (J77)
Altitude: MSIS-86, -83 least variance.
Residuals more negative at higher altitudes for all models.
Latitude: MSIS-83 least variance.
Residuals more positive near the equator for all models.
Local Time: MSIS least variance.
Residuals have terdiurnal variation for J77.
Mean FI0.7: MSIS-86 least variance.
No obvious trends.
Delta FI0.7: MSIS-77 J70 least variance.
Residuals more positive with increasing activity for MSIS-86, -83,and J70.
Magnetic: MSIS-83 least variance.
Residuals irregular. Coverage to Ap 130.
Day/year: MSIS-83 least variance.No obvious trends.
Longitude/UT: All models have low variance.
Comments: Little overall difference between models but Jacchia models
worse for daily variation J77 worse for latitude variation.
MSIS-86 trends: altitude, latitude, and delta FI0.7.
89
Altitude: 130-200
FI0.7:70-80
Parameter: Total DensitySource: AE-D MESA Accel.
Dates: 75284-76029
Data File: MESA-D
Best quiet mean: .00 (MSIS-83,-77) Worst quiet mean: -.03 (MSIS-86)
Best quiet SD: .14 (MSIS-77,J70) Worst quiet SD: .15 (MSIS-86,-83,J77)Best active mean: .00 (MSIS-83) Worst active mean: .04 (MSIS-77)
Best active SD: .15 (all but J77) Worst active SD: .17 (J77)
Altitude: Jacchia least variance.
Residuals more negative with decreasing altitude for MSIS models.
Latitude: J70 least variance.
Residuals flat except more positive toward south for J77.
Local Time: MSIS-86 JTO least variance.
Residuals flat but irregular.
Mean FI0.7: InsuffiCient coverage.
Delta FI0.7: J77 J70 least variance.
Residuals more positive with increasing activity for all models.
Coverage small.
Magnetic: MSIS least variance.
Residuals flat near equator and more positivewith higher activity for
all models. Coverage to Ap 50 near equator and Ap 70 near pole.
Day/year: Insufficient coverage.
Longitude/UT: Variances similar for all models except worse for J77.
Residuals negative near south magnetic pole.
Comments: Little overall difference between models.
MSIS-86 trends: altitude, and magnetic activity.
9O
Altitude: 200-250 Parameter: Total Density Dates: 75284-76029FIO.7:70-80 Source: AE-DMESAAccel. Data File: MESA-E
Best quiet mean: .01 (J77) Worst quiet mean: -.11 (MSIS-77)Best quiet SD: .18 (not MSIS-83) Worst quiet SD: .19 (MSIS-83)Best active mean: .03 (MSIS-83) Worst active mean:-.O8 (MSIS-86)Best active SD: .19 (not J77) Worst active SD: .20 (J77)
Altitude: All models have low variance.Residuals flat for all models.
Latitude: Variances similar for all models.Residuals flat but irregular for all models.
Local Time: Variances similar for all models.Residuals flat but irregular for all models.
MeanFI0.7: Insufficient coverage.
Delta FI0.7: J77 J70 least variance.Residuals more positive with increasing activity for all models.
Magnetic: Variances similar for all models.Residuals irregular. Coverage to Ap 50 near equator and Ap 70 nearpoles.
Day/year: Insufficient coverage.
Longitude/UT: All models have low variance.
Comments: Little overall difference between models.
MSIS-86 trends: delta FI0.7.
91
Altitude: 130-200FI0.7:70-80
Best quiet mean: -.01 (MSIS,J70)
Best quiet SD: .11 (MSIS,J70)Best active mean: .00 (J70)
Best active SD: .13 (MSIS)
Altitude: Variances similar for all models.
Residuals flat for all models.
Parameter: Total DensitySource: AE-D OSS MS
Dates: 75291-76029
Data File: OSSD48
Worst quiet mean: .04 (J77)
Worst quiet SD: .13 (J77)
Worst active mean: .06 (MSIS-77,J77)
Worst active SD: .15 (J77)
Latitude: MSIS-77 least variance.
Residuals flat except more positive toward south for J77.
Local Time: Variances similar for all models.
Residuals flat but irregular.
Mean FI0.7: Insufficient coverage.
Delta FI0.7: All models have low variance.
Residuals flat for all models. Coverage small.
Magnetic: Variances similar for all models.
Residuals flat near equator and more positive with higher activity for
all models. Coverage to Ap 50 near equator and Ap 70 near pole.
Day/year: Insufficient coverage.
Longitude/UT: Variances similar for all models except worse for J77.Residuals more negative near magnetic poles for all models.
Comments: Little overall difference between models.
MSIS-86 trends: magnetic activity and longitude.
92
Altitude: 200-390 Parameter: Total Density Dates: 75291-76029
FI0.7:70-80 Source: AE-D OSS MS Data File: OSSD48
Best quiet mean: -.02 (MSIS-83) Worst quiet mean: -.20 (MSIS-77)
Best quiet SD: .17 (MSIS-86,Jac) Worst quiet SD: .18 (MSIS-83,-77
Best active mean: .00 (MSIS-83) Worst active mean:-.12 (MSIS-77)
Best active SD: .21 (MSIS-86,J70) Worst active SD: .23 (J77)
Altitude: MSIS-83 J70 least variance.
Residuals flat for all models.
Latitude: MSIS-77 least variance.
Residuals more positive near equator for all models.
Local Time: MSIS-77 least variance.
Residuals flat but irregular for all models.
Mean FI0.7: Insufficient coverage.
Delta FI0.7: MSIS-77, J77 J70 least variance.
Residuals more positive with increasing activity for all models.
Magnetic: MSIS-83, -77 least variance.
Residuals irregular. Coverage to Ap 50 near equator and Ap 70 near
poles.
Day/year: Insufficient coverage.
Longitude/UT: MSIS-77 least variance.
Residuals more negativ e near magnetic poles.
Comments: Little overall difference between models except MSIS-77
worse.
MSIS-86 trends: delta FI0.7 and longitude.
93
Altitude: 200-400FI0.7:70-200
Parameter: Total Density Dates: 75178-79021Source: Cactus Accel. Data File: CACTUSTENTH
Best quiet mean: -.03 (J77)
Best quiet SD: .14 (MSIS-86)Best active mean:-.02 (MSIS-86)
Best active SD: .17 (MSIS-86)
Worst quiet mean: -.26 (MSIS-77)
Worst quiet SD: .18 (MSIS-77)
Worst active mean:-.15 (MSIS-77)
Worst active SD: .21 (J77)
Altitude: J77 J70 least variance.
Residuals sharply negative below 270 km for MSIS-86 -83, more
negative with increasing altitude for MSIS-77, and rather flat for J70 andJ77.
Latitude: MSIS-86, -77, J70 least variance.
Residuals more positive near equator for MSIS-83 and more negative
near equator for J77. Coverage 30S to 3ON.
Local Time: MSIS-86 least variance.
Residuals have diurnal variation for all models.
Mean FI0.7: J77 J70 least variance.
Residuals irregular.
Delta FI0.7: J77 J70 least variance.
Residuals more positive with increasing activity for MSIS models.
Magnetic: MSIS-83 least variance.
Residuals irregularly more positive with increasing activity for all models.
Day/year: MSIS-86 -83 least variance.Residuals have semiannual variation with equinox minimum for all models.
Longitude/UT: All models have low varlance.
Comments: Little overall difference between models except MSIS-77
(significantly worse) and Jacchia models significantly worse forlocal timevariations.
MSIS-86 trends: altitude, mean and delta FI0.7, and magnetic activity.
94
Altitude: 400-600FI0.7:70-200
Parameter: Total DensitySource: Cactus Accel.
Dates_ 75178-79021
Data File: CACTUSTENTH
Best quiet mean: -.01 (MSIS-83,J77) Worst quiet mean: -.36 (MSIS-77)
Best quiet SD: .24 (MSIS-86) Worst quiet SD: .28 (MSIS-77)
Best active mean: .00 (MSIS-86) Worst active mean:-.23 (MSIS-77)
Best active SD: .28 (MSIS-86) Worst active SD: .31 (J77)
Altitude: J70 least variance.
Residuals more positive with increasing altitude for all models.
Latitude: MSIS-77, J70 least variance.
Residuals more positive near equator for MSIS-86, -83, flat for
MSIS-77 J77, S shaped for J70. Coverage 30S to 3ON.
Local Time: MSIS-86 least variance.Residuals have semidiurnal variation for all models.
Mean FI0.7: J77 least variance.
Residuals irregular.
Delta FI0.7: MSIS-83 least variance.
Residuals irregularly more positive with increasing activity forMSIS models.
Magnetic: MSIS-83 least variance.Residuals irregularly more positive with increasing activity for all models.
Day/year: MSIS-86 -83 least variance.
Residuals have semiannual variation with equinox minimum for all models.
Longitude/UT: All models have low variance.
Comments: Little overall difference between models except MSIS-77
(significantly worse) and Jacchia models significantly worse forlocal time variations.
MSIS-86 trends altitude, delta FI0.7, and magnetic activity.
95
Altitude: 130-200
FI0.7:70-80
Parameter: Total Density Dates: 75335-76320Source: AE-E MESA Accel. Data File: MESA-E
Best quiet mean: -.O1 (MSIS-77,J77) Worst quiet mean: -.03 (J70)
Best quiet SD: .12 (MSIS) Worst quiet SD: .14 (J77,J70)Best active mean:-.02 (J70) Worst active mean: .07 (MSIS-83)
Best active SD: .14 (MSIS-77) Worst active SD: .16 (J77)
Altitude: J77 least variance.
Residuals more positive with decreasing altitude for all models.
Latitude: All models have low variance.
Coverage 20S to 2ON.
Local Time: MSIS-86 least variance.Residuals have semidiurnal variation for all but MSIS-86.
Mean FI0.7: Insufficient coverage.
Delta FI0.7: MSIS least variance.
Residuals more negative with increasing activity for J77 andJ70 models.
Magnetic: MSIS least variance.
Residuals more positive at highest activity for all models.
to Ap 110.
Coverage
Day/year: MSIS-86 -83 least variance.
Residuals more negative in middle of year for all models.
Longitude/UT: All models have low variance.
Comments: Little overall difference between models although J70
and J77 significantly worse for diurnal, delta FI0.7 and magnetic
activity variations.
MSIS-86 trends: altitude.
96
Altitude: 200-250FI0.7:70-110
Parameter: Total DensitySource: AE-E MESA Aeoel.
Best quiet mean: .00 (MSIS-83)
Best quiet SD: .18 (MSIS-86)Best active mean: .O1 (MSIS-77)
Best active SD: .17 (MSIS-86)
Altitude: MSIS-77 least variance.
Dates: 75335-77278
Data File: MESA-E
Worst quiet mean: -.06 (MSIS-77)
Worst quiet SD: .20 (J77,J70)Worst active mean: .05 (MSIS-83)
Worst active SD: .20 (J77)
Residuals more positive with increasing altitude for all models.
Latitude: All models have low variance.
Coverage 20S to 2ON.
Local Time: MSIS least variance.
Residuals have semidiurnal variation for all models.
Mean FI0.7: JTO least variance.
Residuals irregular for all models.
Delta FI0.7: MSIS least variance.
Residuals more negative with increasing activity for J77 and J70.
Magnetic: Variances similar for all models.
Residuals irregular. Coverage to Ap 110.
Day/year: MSIS-86 -83 least variance.Residuals have no clear trend.
Longitude/UT: All models have low variance.
Comments: Little overall difference between models although J70
and J77 significantly worse for diurnal, and delta FI0.7 variations.
MSIS-86 trends: altitude, local time.
97
Altitude: 130-200 Parameter: Total Density Dates: 75343-76319
FI0.7:70-80 Source: AE-E OSS MS Data File: OSS-E
Best quiet mean: .00 (MSIS-83) Worst quiet mean: -.07 (MSIS-77,JTO)
Best quiet SD: .13 (MSIS-77) Worst quiet SD: .17 (JTO)
Best active mean: .02 (MSIS-86,-83) Worst active mean:-.11 (J77)
Best active SD: .15 (MSIS-83,-77 Worst active SD: .19 (J77)
Altitude: J77 least variance.
Residuals more positive with decreasing altitude for all models.
Latitude: All models have low variance.
Coverage 20S to 2ON.
Local Time: MSIS-86 least variance.
Residuals have semidiurnal variation for MSIS-77, J77, and JTO.
Mean FI0.7: Insufficient coverage.
Delta FI0.7: MSIS least variance.
Residuals more negative with increasing activity for MSIS-77, J77 and
JTO models.
Magnetic: MSIS-77 J77 least variance.Residuals more positive with increasing activity for all models. Coverage
to Ap 110.
Day/year: MSIS-77 least variance.
Residuals have no clear trend.
Longitude/UT: All models have low variance.
Comments: MSIS models slightly better overall than the Jacchia models,
but Jacchia models significantly worse for local time variations.
MSIS-86 trends: altitude and magnetic activity.
98
Altitude : 200-400FI0.7:70-170
Parameter: Total DensitySource: AE-E OSS MS
Dates: 75343-78336
Data File: OSS-E
Best quiet mean: -.O1 (J77) Worst quiet mean: -.15 (MSIS-77)
Best quiet SD: .18 (MSIS-83) Worst quiet SD: .21 (MSIS-77,J70)
Best active mean:-.03 (MSIS-86,-83) Worst active mean:-.12 (MSIS-77)
Best active SD: .19 (MSIS-86) Worst active SD: .23 (J77)
Altitude: MSIS-83 least variance.
Residuals more negative with increasing altitude for MSIS-77
J70.
Latitude: All models have low variance.
Coverage 20S to 2ON.
Local Time: MSIS-86 least variance.
Residuals have semidiurnal variation for all models and diurnal
for J77 and J70.
Mean FI0.7: All models same variance.
Residuals flat for all models.
Delta FI0.7: MSIS least variance.
Residuals more negative with increasing activity for J77 and J70.
Magnetic: J77 MSIS-77 least variance.
Residuals more positive at highest activity for all models.
to Ap 110.
Coverage
Day/year: MSIS-83 J70 least variance.Residuals have no clear trend.
Longitude/UT: All models have low variance.
Comments: Little Overall difference between models except
MSIS-77 (worse), but Jacchia models significantly worse forlocal time and delta FI0.7 variations.
MSIS-86 trends: altitude, local time, and magnetic activity.
Altitude: 130-200FIO.7:70-80
Parameter: Total DensitySource: AE-E NACE MS
Dates: 75341-76307
Data File: NACE-E
Best quiet mean: -.O5 (all but J70) Worst quiet mean: -.07 (JTO)
Best quiet SD: .11 (MSIS-86) Worst quiet SD: .17 (J77)
Best active mean:-.04 (MSIS-86,-83) Worst active mean:-.15 (J77)
Best active SD: .13 (MSIS-86,-83 Worst active SD: .19 (J77)
Altitude: MSIS-86,-83 least variance_
Residuals more negative with decreasing altitude for all models.
Latitude: MSIS-86 least variance.
Residuals more positive in northern hemisphere for J70.
Coverage 2OS to 2ON.
Local Time: MSIS-86 least variance.
Residuals have semidiurnal variation for MSIS-77, J77, J70.
Mean FI0.7: Insufficient coverage.
Delta FIO.7: MSIS-86,-83 least variance.
Residuals more negative with increasing activity for MSIS-77 andJ77 models.
Magnetic: MSIS-86, -83 least variance.
Residuals flat but irregular for all models except more negative
with increasing activity for J77. coverage to Ap 150.
Day/year: MSIS-86 -83 least variance.Residuals have annual variation with June minimum for MSIS-77, J77,
and J70 models.
Longitude/UT: All models have low variance.
Comments: MSIS models better than Jacchia models overall. Jacchia models
significantly worse for local time variations and annual.
MSIS-86 trends: altitude.
100
Altitude: 200-400
FI0.7:70-230
Parameter: Total DensitySource: AE-E NACE MS
Dates; 75341-81155
Data File: NACE-E
Best quiet mean: -.07 (J70)
Best quiet SD: .19 (MSIS-83)Best active mean:-.07 (MSIS-83)
Best active SD: .22 (MSIS-83)
Worst quiet mean: -.22 (MSIS-77)
Worst quiet SD: .23 (MSIS-77)Worst active mean:-.18 (MSIS-77)
Worst active SD: .24 (MSIS-77)
Altitude: J70 least variance.
Residuals more negative with increasing altitude for MSIS.
Latitude: All models have low variance.
Coverage 20S to 2ON.
Local Time: MSIS-86 least variance.Residuals have semidiurnal variation for all models and diurnal
for J77 and J70.
Mean FI0.7: MSIS-86 least variance.
Residuals irregularly more negative with increasing activity for MSIS
models and flat for J70 and J77.
Delta FI0.7: J70 least variance.
Residuals flat but irregular for all models.
Magnetic: J77 MSIS-77 least variance.
Residuals irregularly more positive for all models. Coverage to Ap 150.
Day/year: J77 J70 least variance.Residuals have semiannual variation with equinox minimum for all models.
Longitude/UT: All models have low variance.
Comments: Little overall difference between models except
MSIS-77 (worse) but Jacchia models worse for local time variations.
MSIS-86 trends: altitude, local time, mean FIO.7, annual, and magnetic
activity.
101
Altitude: 400-480
FI0.7:160-230Parameter: Total DensitySource:AE-E NACE MS
Dates: 78336-80281
Data File: NACE-E
Best quiet mean: -.08 (J70) Worst quiet mean: -.32 (MSIS-77)
Best quiet SD: .18 (J77) Worst quiet SD: .22 (MSIS-77)
Best active mean:-.14 (J77,J70) Worst active mean:-.29 (MSIS-77)Best active SD: .23 (not MSIS-77) Worst active SD: .27 (MSIS-77)
Altitude: J70 least variance.
Residuals flat except morepositive with increasing altitude forMSIS-77 models.
Latitude: All models have low variance.
Coverage 20S to 2ON.
Local Time: MSIS-86,-83, J77 least variance.
Residuals have semidiurnal variation for all models except terdiurnal
for J77.
Mean FI0.7: MSIS-86,-83 least variance.
Residuals more negative with increasing activity except for MSIS-83.
Delta FI0.7: MSIS-86,-83 least variance.
Residuals flat but irregular for all models.
Magnetic: J77 least variance.
Residuals flat but irregular for all models. Coverage to Ap 150.
Day/year: All models the same except MSIS-77 worse.
Residuals irregular for all models.
LOngitude/UT: All models have low variance.
Comments: Little overall difference between models except MSIS-77 (worse).
MSIS-86 trends: local time mean FI0.7.
102
Altitude: 200-400
FI0.7:125-230
Parameter: Total DensitySource: DE-B NACS MS
Dates: 81220-83047
Data File: NACSBRHOFIFTH
Best quiet mean: -.15 (J77)Best quiet SD: .14 (MSIS-86)
Best active mean:-.15 (MSIS-83)
Best active SD: .16 (MSIS-86)
Worst quiet mean: -.33 (MSIS-77)
Worst quiet SD: .23 (MSIS-77)Worst active mean:-.35 (MSIS-77)
Worst active SD: .21 (MSIS-77)
Altitude: MSIS least variance.
Residuals more negative at lowest altitude for all models.
Latitude: MSIS-86 least variance.
Residuals more negative near poles for MSIS083 and J70, and more
positive for J77. MSIS-77 more negative at south pole and positive
at north pole.
Local Time: MSIS-86 -83 least variance.
Residuals have semidiurnal or terdiurnal variation for MSIS-77,
J77 and J70.
Mean FI0.7: MSIS-86 -83 least variance.
Residuals more negative at higher activity for MSIS-77.
Delta FI0.7: MSIS-86 -83 least variance.
Residuals more negative at lowest activity for all models.
Magnetic: MSIS-86 least variance.
Residuals more negative with increasing activity for MSIS-83 and
-77 near pole. Coverage to Ap 200.
Day/year: MSIS-86 -83 least variance.Residuals have semiannual variation for all models.
Longitude/UT: MSIS-86 -83 least variance.Residuals more negative near magnetic poles for MSIS-77, J77, and
J70.
Comments: MSIS-86 model best for standard deviations and J77 best
for absolute values. Instrument calibration may be inaccurate.
MSIS-86 trends: altitude, delta FI0.7 and annual.
103
Altitude: 400-800
FI0.7:150-230
Parameter: Total DensitySource: DE-B NACS MS
Best quiet mean: -.09 (J77)
Best quiet SD: .17 (MSIS-86)Best active mean:-.03 (J77)
Best active SD: .24 (MSIS-86)
Altitude: MSIS-86 least variance.
Dates: 81228-83014
Data File: NACSBRHOFIFTH
Worst quiet mean: -.38 (MSIS-77)
Worst quiet SD: .22 (MSIS-77)Worst active mean:-.31 (MSIS-77)
Worst active SD: .28 (MSIS-77)
Residuals more positive at higher altitudes for all models.
Latitude: MSIS'86 J70 least variance.
Residuals more positive at midlatitudes for J77.
Local Time: Variances similar for all models.
Residuals have semidlurnal or terdlurnal variation for all models.
Mean FI0.7: MSIS-86 least variance.
Residuals more positive at higher activity for all models.
Delta FI0.7: MSIS-86 least variance.
Residuals more negative at lowest activity for all models.
Magnetic: MSIS-86 least variance.
Residuals flat but irregular for all models. Coverage to Ap 200.
Day/year: MSIS-86 least variance.
Residuals have no clear trend. Coverage inadequate.
Longitude/UT: MSIS-86 J70 least variance.Residuals more negative near magnetic poles for all but MSIS-86.
Comments: MSIS and Jacchia models equivalent except MSIS-77 worse.
Instrument calibration may be inaccurate.
MSIS-86 trends: altitude, mean and delta FI0.7.
104
NAS NalO_aiAeronauTcsandSoaceAcJrn_sI_aT_n
Report Documentation Page
I. Report No.
NASA TM 100707
2, Government Accession No.
4. Title and Subtitle
High Altitude Atmospheric Modeling
7. Author(s)
Alan E. Hedin
9. Performing Organization Name and Address
Laboratoryfor Atmospheres
Goddard Space Flight Center
Greenbelt, MD 20771
12. Sponsoring Agency Name and Address
National Aeronautics and Space Administration
washington, DC 20546-1000
15. Supplementary Notes
3. Recipient's Catalog No.
5. Report Date
October 1988
6. Performing Organization Code
614
8. Performing Organization Report No.
88B0260
10. Work Unit No.
11. Contract or Grant No.
13. Type of Report and Period Covered
Technical Memorandum
14. Sponsoring Agency Code
Sponsored .under USAF MIPR FY 616870485
16. Abstract
Five empirical models were compared with 13 data sets, including both atmospheric
drag-based data and mass spectrometer data. The most recently published model,
MSIS-86, was found to be the best model overall with an accuracy around 15 percent.
The excellent overall agreement of the mass spectrometer-based MSIS models with
the drag data, including both the older data from orbital decay and the newer
accelerometer data, suggests that the absolute calibration of the (ensemble of)
mass spectrometers and the assumed drag coefficient in the atomic oxygen regime
are consistent:tof5 pereent.
This study illustrates a number of the reasons for the current accuracy limit such
as calibration accuracy and unmodeled trends. Nevertheless, the largest variations
in total density in the thermosphere are accounted for, to a very high degree, by
existing models. The greatest potential for improvements is in areas where we
still have insufficient data (lik4 the lower thermosphere or exosphere), where
there are disagreements in technique (suc h as theexosphere) which can be resolved,
or wherever generally more accurate measurements become available.
17. Key Words(SuggestedbyAuthor(s))
Upper Atmosphere Structure
Thermospheric Density
18. Distribution Statement
Unclassified-Unlimited
Subject Category 46
19. Secur_ Classif. (of this report)
Unclassified
20. Securiw Classif.(ofthispage)
Unclassified
21. No. of pages 22. Price
NASA FORM 1_6 OCT
