- 39 -

5: F-15E AND F-16C BOMBING SKILLS

LOW-ALTITUDE BOMBING SKILLS

Because fighter crews participating in them often go for months at

a time without practicing some of their important combat skills, peace
operations deployments offer an opportunity to conduct a quantitative
analysis of the impact of prolonged lack of practice on the low-altitude
air-to-ground combat skills of USAF aircrew. While the data available
on F-15E and F-16C crews does not allow an analysis of all important
aspects of low-altitude combat skills such as mission planning, low-
level navigation, threat reactions, tactical formation, mutual support,
etc., it does allow for an analysis of the relationship between bombing
accuracy and practice and experience variables. Bombing accuracy is not
the only important aspect of low altitude air-to-ground combat, but it
is the best documented and most accurately measured.1 In addition, to
the extent that it is positively correlated with the other aspects of
the low-altitude combat art, the relationships revealed from an analysis
of bombing accuracy and practice should shed some light on the
relationship of these other skills and practice as well. Or, in other
words, bombing accuracy will be used as an easily quantifiable proxy
measure for the less easily measured tactical skills listed above.

Types of Low Altitude Bomb Deliveries

The two broad categories of low altitude bomb deliveries analyzed

in the following sections are Low Angle Low Drag and High Drag (LALD/HD)
and Loft deliveries. LALD/HD deliveries are performed by pilots of both
F-15E and F-16C aircraft.2 When performing a LALD/HD delivery pilots
acquire and track a target visually. They align the target with the

____________
1 See Appendix A for a discussion of air-to-ground and air-to-air
combat skill measures and ways they could be improved.
2 These aircraft types were analyzed because the F-15E force
provides the bulk of the USAFs deep strike/interdiction capability while
F-16Cs are the most numerous aircraft in the active fighter force.
 - 40 -

illuminated “pipper” in their Heads-Up Display (HUD). They then attempt

to stabilize the aircraft in wings level one-G flight and release the
bomb(s). All of this takes place while the aircraft is diving toward
the target at an angle of not more than 30 degrees. The aircraft then
over-flies the target and leaves the area. After leaving the aircraft
high drag bombs deploy large hinged fins that rapidly slow the bomb.
This creates greater separation between the aircraft and the bomb, and
when compared to low drag bombs (which do not have the hinged fins)
allows the aircraft to attack from lower altitudes and still escape the
fragmentation pattern of the bomb. The delivery techniques for low drag
and high drag bombs are so similar they form a single category in USAF
fighter squadron bombing competitions. They will be treated as a single
category in this analysis. Figure 5.1 illustrates a typical LALD/HD
delivery profile.

LALD/HD Deliveries
1

<30 Degrees

Target

Figure 5.1 -- LALD/HD Delivery Profile

The second type of low altitude bomb delivery analyzed in the

following sections is the loft delivery. All loft deliveries analyzed
in this chapter were performed by F-15E Weapons System Officers (WSOs).
A WSO is the second crew member in two seat fighters such as the F-15E
 - 41 -

responsible for operating navigation, electronic countermeasures, and

other sophisticated systems. During a loft delivery WSOs use the
aircraft’s radar and computer systems to aim bombs.
The loft is a low altitude climbing delivery that maximizes weapon
range, and minimizes aircraft exposure to target area defenses. Loft
deliveries are generally less accurate than LALD/HD deliveries. To
perform a loft delivery the WSO identifies and tracks a target using the
aircraft’s ground mapping radar. At a pre-determined range from the
target the pilot pulls the aircraft into a climb, usually less than 45
degrees nose high, and the aircraft computer targeting system release
the bomb(s). Following bomb release, the pilot begins a tight
descending turn designed to minimize aircraft exposure to target
defenses. Unlike LALD/HD deliveries the aircraft does not over fly the
target.

Loft Deliveries

1
<45 Degrees

2
Target

Figure 5.2 -- LOFT Delivery Profile

 - 42 -

F-15E LOW LEVEL BOMBING SKILLS

The Variables

In order to conduct a quantitative analysis of the importance of

practice, experience, and other relevant variables on aircrew combat
skills I collected data on 55 F-15E aircrew - 30 pilots and 25 WSOs -
assigned to the 4th Fighter Wing, 335th Fighter Squadron at Seymoure-
Johnson Air Force Base, North Carolina. I collected the following seven
“demographic” variables on each crewmember:

• Aeronautical Rating - Dummy variable for aeronautical rating.

The sample contains 30 pilots and 25 WSOs
• Experienced Dummy variable for whether or not a crewmember
is considered experienced by USAF standards. 73% of pilots and
76% of WSOs in the sample were experienced.

• Total USAF Flight Hours - Total flight hours in USAF aircraft.

• Minimum: 125.4
• Maximum: 3333.2
• Mean: 1385.2
• Total Fighter Hours - Total flight hours in fighter aircraft.
• Minimum: 125.4
• Maximum: 2932.7
• Mean: 1210.3

• F-15E Flight Hours - Total hours in F-15E aircraft.

• Minimum: 125.4
• Maximum: 2374.4
• Mean: 537.4
• Deploy - Dummy variable for whether or not a particular
crewmember deployed to Saudi Arabia in support of Operation
Southern Watch from February through June 1997. ~47 % of both
pilots and WSOs deployed.
• Instructor - Dummy variable for whether or not a particular
crewmember was a qualified instructor in the F-15E. 30% of
pilots and 39% of WSOs were instructors.
 - 43 -

In addition, I collected the following six variables related to

low-altitude bomb delivery practice and performance:

• Visual Low Drag/High Drag Bomb Deliveries - Number of visual low drag
and high drag deliveries a particular crewmember performed by month
for January through November 1997
• Visual Low Drag/High Drag Bomb Hits - Number of visual low drag and
high drag bombs dropped by a particular crewmember judged hits by
USAF hit criteria by month for January through November 1997
• Visual Low Drag/High Drag Bomb CEA - Average visual low drag and high
drag delivery miss distance by crewmember by month for January
through November 1997
• RADAR Loft Bomb Deliveries - Number radar loft deliveries a
particular crewmember performed by month for January through November
1997
• RADAR Loft Bomb Hits - Number of radar loft bombs dropped by a
particular crewmember judged hits by USAF hit criteria by month for
January through November 1997
• RADAR Loft Bomb CEA - Average visual low drag and high drag delivery
miss distance by crewmember by month for January through November
1997

Data Overview

Figure 5.3 below shows the average number of radar and visual bomb
deliveries performed by all 55 aircrew in the sample by month (1
corresponds to January, 2 to February, etc.). In addition, it shows the
average number of radar and visual bomb hits for the sample by month.
The five months that stand out are February and March with almost no
bomb deliveries and hits, and August through October with unusually high
numbers of deliveries and hits - especially visual deliveries and hits.
As explained above, the squadron providing this data deployed to Saudi
Arabia from late February through late June 1997. When one takes into
account the disruption of normal training due to preparations for the
deployment in early February (allowing deploying aircrew and support
personnel to take leave to tend to family matters prior to a long
 - 44 -

separation, the need to configure and otherwise prepare the aircraft for
the long deployment flight, etc.) and post deployment “spin-up” in
Saudi Arabia (reconfigure/repair aircraft, familiarize aircrew with
local procedures, etc.) the decrease in air-to-ground training in
February, and to some extent March, is understandable. Overall the
squadron seems to have accomplished somewhat reduced bombing practice
during its deployment.

Average Radar and Visual Bomb

Deliveries and Hits (total)

4.50
4.00
3.50
3.00
2.50
2.00
1.50
1.00
0.50
0.00
1

10

11
Month
Total Average of Loft Total Average of Lofthit
Total Average of LD/HD Total Average of LD/HD Hit

Figure 5.3 -- F-15E bomb deliveries and hits: entire squadron

However, this overall picture is deceiving. Again, recall from

Table 5.1 that approximately half the aircrew assigned to the squadron
did not deploy between February and June 1997. Figures 5.4 and 5.5
below make it clear that it was these non-deployed aircrew who recorded
ALL of the practice low-level bomb deliveries for the squadron for the
four months between March and July. In other words, none of the
deployed aircrew accomplished a low altitude practice bomb delivery
during the deployment.
 - 45 -

Average Radar and Visual Bomb

Deliveries and Hits Non-Deployed Crews

4.50
4.00
3.50
3.00
2.50
2.00
1.50
1.00
0.50
0.00
1

10

11
Month
Average of Loft Average of Lofthit
Average of LD/HD Average of LD/HD Hit

Figure 5.4 -- Average Radar and Visual Bomb Deliveries and Hits: Non-
Deployed Crews

Average Radar and Visual Bomb

Deliveries and Hits Deployed Crews

4.50
4.00
3.50
3.00
2.50
2.00
1.50
1.00
0.50
0.00
1

10

11

Month
Average of Loft Average of Lofthit
Average of LD/HD Average of LD/HD Hit

Figure 5.5: Average Radar and Visual Bomb Deliveries and Hits: Deployed
Crews
 - 46 -

Statistical Analysis

In order to determine the best predictors for F-15E aircrew low-

level bomb delivery performance extensive regression analysis was
performed. A large number of possible model specifications and
functional forms were investigated and compared to determine which of
the available independent variables, or combinations and/or
transformations of them, were the best predictors of visual and radar
bomb miss distance.
The first regression model specifications tried were those
suggested by the literature on learning theory discussed in Chapter Two.
These specifications involved predicting bomb miss distance as a
function of the log or a power transformation of an experience
component, such as total or fighter hours, and the log or power
transformation of a recent experience component such as the number of
bombs dropped in the past 30 days.
These theory based model specifications produced coefficients with
the expected signs. They predicted that as experience (measured by
fighter hours) increased bomb miss distance (in terms of Circular Error
Achieved or CEA) decreased for a given number of practice bombs dropped
in a month. They also indicated that holding experience (fighter hours)
constant, increasing the number of bombs dropped also decreased bomb
miss distance. However, these specifications were not particularly good
at predicting CEA even though they often had statistically significant
coefficients. The best of these models achieved R-square values less
than 0.10 for visual deliveries and less than 0.01 for radar deliveries.
For this particular data set a linear model controlling for
instructor status and the number of practice deliveries in the past 30
days is a much better fit with an R-square value of 0.18. for visual
deliveries performed by pilots about 0.03 for radar, or loft, bomb
deliveries performed by WSOs. 3. These R-square values indicate the

____________
3 The impact of number of bombs dropped 31 to 60 and 61 to 90 days
prior on CEA was also investigated. Coefficients for these variables,
either singly or in combination, were not significant in either the
linear or non-linear model specifications.
 - 47 -

linear specification fits the available data about twice as well as the
theory based models for visual bomb deliveries and about seven times
better for radar deliveries. This may result from the level of
experience of the individuals involved. Recall from Chapter Two that
learning theory postulates s-shaped learning curves as shown in figure
5.6 below. The figure illustrates how performance increases from some
arbitrary minimum level to a maximum as a function of the number of
times a person has accomplished the task in question, or trials.

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
0

Figure 5.6 – Typical S-Shaped Learning Curve

Theory states that novices learn slowly with each trial. However,
as their experience builds, they fairly quickly acquire enough
experience so that the proficiency they gain from each successive trial
increases, and then remains constant over a broad range of experience.
Eventually, experience builds up to the point that there is little room
for improvement due to physical or mental limitations on human
performance, and the performance gain from additional trials again
declines. The broad middle range of experience in the center part of
 - 48 -

the s-curve is essentially linear. So, if all, or even the vast

majority of the crews under study here are in this broad middle range of
experience, then a linear model will fit the data better than an s-
curve.
This is quite plausible since combat ready USAF squadrons do not
have any true novice crews. Even the least experienced second
lieutenants have considerable by the time they arrive at an operational
squadron. Much of the experience they have was gained in the
Replacement Training Unit (RTU) where they had their initial exposure to
all types of tactical flying in the same type of aircraft they
eventually fly when they report to an operational squadron – in this
case the F-15E. So, there are probably no true novices among the crews
under study here. For these crews it appears that the impact of practice
within the past 30 days on the performance of all crews – from the least
to the most experienced – is essentially the same.
This does not mean that experience does not matter, or that crews
perform the same regardless of their experience level – far from it.
Recall that the model under discussion here uses two components to
predict average bomb miss distance or CEA – recent practice (represented
by the number of practice bomb deliveries of a given type accomplished
in the past 30 days) and instructor status.

Visual Bomb Deliveries

Instructor status is a proxy measure for three important factors

that contribute to aircrew bombing performance. First, since
instructors tend to be among the more experienced members of a squadron,
instructor status is something of a proxy measure for cumulative
experience.4 Second, in order to become an instructor a fighter crew
member must compete against his or her peers for a limited number of
instructor upgrade course spots. As a result, the best performing – or
most talented – have a better chance of upgrading to instructor status.
Third, as the second point implies, instructors complete rigorous formal
upgrade training courses where they learn instructional techniques while
further honing their already considerable skills in the tasks they will

____________
4 For the F-15E crews under study here instructors had, on average,

almost 70 percent more fighter hours than non-instructors.

 - 49 -

be teaching to their squadron mates. So, in short, instructor status is

a powerful predictor of air-to-ground combat skill performance because
instructors tend to be more experienced, more talented, and more highly
trained than non-instructors. Table 5.3 below presents the regression
coefficients for a model estimating LDHD CEA as a function of instructor
status and LDHD deliveries in the current month.5

____________
5 Note: The t-statistics and p-values presented in table 5.3
reflect the use of Huber standard errors. One of the underlying
assumptions of ordinary least squares regression is that the error terms
(the difference between predicted and actual bomb miss distance (CEA) in
this case) are normally distributed and do not vary systematically as
the magnitude of the independent variables increase. A Goldfeld-Quandt
test was performed to test if this important assumption held. The
results of the test indicated that error terms were much larger for low
values of LDHD than for high values (p=0.025). In other words, the
model predicts bomb miss distance (CEA) better for pilots who have
dropped a relatively large number of bombs in a month than for those who
have dropped relatively few bombs. This is logical because pilots who
have dropped a relatively large number of bombs recently will probably
be more consistent than those who have dropped relatively few. Huber’s
formula produces robust standard errors and covariance matrices
regardless of error distribution. Therefore, it compensates for the
non-constant error terms and produces more accurate standard errors, t-
statistics and p-values for data with non-constant variance
(heteroscedasticity).
A final technical point on the data analyzed here. It consists of
cross sectional data on 55 individual pilots and WSOs collected over a
period of eleven months. This combination of cross sectional and time
series data is called pooled data. In pooled data it is possible that
cross sectional parameters may shift over time, or that larger
processes, related to the passage of time, may explain some of the
variation in performance. One way to correct for these possibilities is
to control for individual variation and differences across time periods
by including a series of dummy variables for the individuals and time
periods. This allows average performance to vary over both time and
between individuals.
To determine if it was necessary to include dummy variables in my
models I conducted a nested models test. The error sum of squares for
models containing the dummy variables were compared to the error sum of
squares of models without the dummy variables. Since the models without
the dummy variables have more restrictions (coefficients for all
individuals and time periods are assumed to be the same), if the dummy
variables have significant explanatory power, we would expect the models
with the dummy variables to have significantly lower error sums of
squares than the restricted models. (See Robert S. Pyndyck and Daniel L.
Rubinfeld, Econometric Models & Economic Forecasts, Third Edition,
Mcgraw-Hill, New York, 1991, pp. 223-226.) For the F-15E LALD/HD model
the F-test statistic with 101 and 66 degrees of freedom is 0.595
 - 50 -

Variable Coefficient Standard t-Statistic P-Value

Error
N=104
Constant 25.34121 1.792022 14.181 0.000

Instructor -7.564915 1.873251 -4.038 0.000

Practice -0.4662477 0.2119053 -2.200 0.030

(LDHD
Current
Month)

Table 5.3 – F-15E Visual Bomb Delivery Regression Coefficients

All of the coefficients meet the standard test of statistical

significance (p-values less than 0.05 for their individual t-
statistics), but what does this regression analysis say about the
practical impact of instructor status and recent practice on average
visual bomb miss distance? First, the constant can be interpreted as
the average bomb miss distance – in meters – for a non-instructor who
has not dropped any practice bombs in the past 30 days. In this case it
is approximately 25 meters. The instructor coefficient is an estimate
of how much better an instructor does, on average, than a non-
instructor. In this case, instructors’ bombs landed about 7.5 meters
closer to the target than a non-instructors’ if both have dropped the

(p=0.991) and for the Loft model the F-test statistic with 192 and 133
degrees of freedom is 0.41 (p=0.99). For the F-16 LALD/HD model
described later the F-statistic with 164 and 114 degrees of freedom is
1.15 (p=0.216). So, in all cases I failed to reject the null hypothesis
that the models without dummy variables were the correct specification
at the p=0.05 level. Therefore, the simpler models are used throughout
this analysis.
This has a big advantage. Sample size here is relatively small –
only 104 observations for visual bombs and 195 for the radar bomb
analysis presented below, and 166 for the F-16 visual bomb analysis
presented later. In the F-15 cases adding dummy variables for each
pilot/WSO and each time period adds 35 variables to the LALD/HD and 59
to the Loft regression equation (50 variables in the F-16 case), so the
models is probably overspecified for the available data So in this case
the best option is probably the Huber regression since it corrects for
known heteroscedacity while providing unbiased and consistent parameter
estimates.
 - 51 -

same number of practice bombs in the current month. Finally, the

practice coefficient indicates that for each additional practice bomb
dropped CEA decreases by about 0.46 meters, or about 18 inches, on
average for both instructors and non-instructors.6
With these estimates of average bomb miss distance, the instructor
effect and the impact of practice within the past 30 days on visual bomb
miss distance it is possible to estimate how far any given visual bomb
dropped will fall, on average, from the target if we know the pilot’s
instructor status and the number of practice low altitude bombs he has
dropped in the past month. We can then compare these predicted average
bomb miss distances to established USAF hit criteria for various bomb
deliveries to determine if pilot performance under a variety of
assumptions about instructor status and recent practice is likely to
meet minimum USAF standards for bomb accuracy.
At first glance it would appear that we can simply compare the
estimated miss distances to the USAF low angle low drag/high drag bomb
hit criteria. The following calculations assume any LALD or LAHD bomb
predicted to land within 27.5 meters of the intended target is a hit.
This is in line with USAF hit criteria for these types of bomb
deliveries from F-15E aircraft.7 Comparing predicted bomb miss
distances to the hit criteria indicates that, on average both
instructors and non-instructors will achieve a hit even if they have not
dropped any bombs in the past 30 days. Non-instructors are predicted to
have an average miss distance of about 25.3 meters, instructors about
17.8 meters under these conditions. So, with everyone achieving hits on
____________
6 An additional model incorporating an interaction term for
instructor and LDHD was estimated to investigate the possibility that
incremental improvement from recent practice bombs was not the same for
the two groups. This term was not significant. This indicates that an
additional practice bomb improves bomb miss distance by the same amount
for instructors and non-instructors.
7 Actual USAF hit criteria for LALD bombs for F-15E pilots is 30
meters. A hit for a LAHD bomb is 25 meters. Source: Multi-Command
Instruction 11-F-15E. Since these deliveries are almost identical from
a technical point of view, and the data consists of equal numbers of
each, I combined them to increase sample size and increase the
statistical power of the regression analysis discussed above. Averaging
the 25 and 30 meter hit criteria gives an new average hit criteria for
the combined data of 27.5 meters.
 - 52 -

average, it seems that lack of practice has little impact on an

individual pilots ability to accurately drop visually aimed bombs from
low level.
However, before we can compute the predicted probability of hit
(Ph) we must account for the expected variation around the estimated
average miss distances. The predicted 25.3 and 17.8 meter miss
distances are average miss distances. Some bombs will land closer to
the target and some much farther away. Very few will land at exactly
25.3 or 17.8 meters from the intended target. To calculate the
predicted probability that a particular pilot will drop a bomb within
the 27.5 meter hit criteria, we need to know not only the average
predicted miss distance, but also the magnitude of the variation around
that average miss distance. An example can help illustrate this
important point. Lets assume ten non-instructors who have not dropped
any visual practice bombs in the past 30 days fly out to the bombing
range and all drop one bomb and return to base. Based on the regression
analysis presented earlier, and assuming normally distributed errors, we
would expect five of the bombs to land between zero and 25.3 meters from
the target and the other five bombs to land more than 25.3 meters from
the target. Given the relatively small difference between the average
impact distance and the limit of the hit criteria, it is likely that a
sizable fraction of the bombs dropped – say four out of ten – will
actually land outside the hit criteria limit. It is possible virtually
all the bombs could land within the hit criteria if dispersion about the
average miss distance is small enough – on the order of about 1 meter.
I calculated the standard deviations of the bomb miss distance
predictions for all possible values of LDHD (from 0 to 18) for both
instructors and non-instructors. For non-instructors the standard
deviations of the predicted bomb miss distances ranged from 11 to about
18 meters. For instructors they ranged from about 15 to about 22
meters. Combining these standard deviations, and the regression
analysis parameters from table 5.3 it is possible to calculate the
probability of hit for all combinations of instructor status and number
of LALD/HD deliveries in the data set. These probabilities are
presented graphically in figure 5.7 below.
 - 53 -

F-15E Visual Bomb Ph

1
Non-instructor
Mean When Not Deployed
Instructor
0.9

Predicted Ph
0.8

0.7

0.6

0.5

0.4
0 2 4 6 8 10 12 14 16 18
Visual Practice Bombs Last 30 Days

Figure 5.7 -- Visual delivery probability of hit for F-15E instructor-

and non-instructor pilots as a function of bombs dropped in the past 30
days.

Figure 5.7 shows that instructors generally have a higher

probability of hitting a target with a LALD/HD delivery When the
squadron is not deployed to a peace operation, both instructors and non-
instructors drop, on average, about four practice LALD/HD bombs per
month. At this level of practice non-instructors achieve about the same
hit probability as instructors have had no practice in the past 30
days.. Non-instructor hit probability decreases from 0.62 for a normal
month at home with three practice deliveries to about 0.54 when a non-
instructor has not practiced visual bomb deliveries in the past 30 days.
This may seem to be a relatively trivial decrease in hit probability,
but if viewed from the perspective of the probability of miss things
look a bit more serious. For non-instructors, the probability of
missing an important target with the most common visual low-level
delivery increases almost 20 percent (from 0.38 to 0.46) if the pilot in
question has been deployed to support a peace operation, or other
 - 54 -

mission, where training opportunities for important MTW low altitude

combat skills are limited or non-existent. For instructors the
probability of missing increases about 25 percent from 0.25 with recent
practice to about 0.33 without.

Radar Bomb Deliveries

An analysis similar to the one conducted for visual bomb deliveries

was conducted for radar bomb delivery accuracy. The focus was on loft
deliveries – the primary night low level radar aimed bomb delivery
technique for F-15E crews. During a loft delivery the crew cooperates
to deliver a weapon. The pilot flies the aircraft through the loft
maneuver while the WSO uses the aircraft’s radar and computer systems to
aim the bomb and compute the proper release point. The loft delivery
profile allows the WSO to keep the target within the radar’s field of
view up to bomb release, and decreases exposure to terminal defenses.
However, these advantages come at the cost of increased bomb release
distances. This, of course, increases the unguided time of flight of
the bomb and makes these deliveries inherently less accurate than visual
deliveries. This difference is reflected in both the regression
analysis results presented below, and the expanded USAF hit criteria of
110 meters for loft deliveries.
Like the F-15E visual bomb delivery model discussed above the
constant and instructor coefficients are statistically significant.
Just as in the visual delivery case the regression results presented in
table 5.4 use Huber standard errors, and the model fits the available
data better than any other functional form examined. However, the
coefficient for recent practice – measured by the number of loft
deliveries in the current month – does not meet the test of statistical
significance.
 - 55 -

Variable Coefficient Standard t-Statistic P-Value

Error
N=195
Constant 85.58033 5.768699 14.835 0.000

Instructor -13.43073 5.597754 -2.399 0.017

Practice -1.599719 1.207627 -1.325 0.187

(LOFT
Current
Month)

Table 5.4 – F-15E Radar Bomb Delivery Regression Coefficients

I have elected to retain LOFT as an explanatory variable for

several reasons. First, there are sound theoretical reasons, outlined
in Chapter Four, to believe that recent practice is an important
determinant of performance. Second, these theoretical reasons for
including LOFT are reinforced by significant results for the recent
practice measure in both the F-15E visual bomb delivery model discussed
above and the F-16C visual bomb delivery model presented below. Third,
as described above, LOFT deliveries depend – to a much greater extent
than visual deliveries – on the performance and skill of both the pilot
and WSO. Unfortunately, the data available did not include information
on which pilots and WSOs were teamed for LOFT deliveries. So it is
possible that on his first delivery in a month a WSO might be teamed
with a instructor with lots of recent practice in LOFT deliveries, while
on subsequent delivery attempts that month the WSO is teamed with an
inexperienced pilot with less recent practice. Under these conditions
the impact of pilot performance on bomb accuracy will tend to weaken the
link between WSO practice and bomb accuracy. Statistically this
situation will result in larger standard errors for the LOFT coefficient
than we would expect if we could control for pilot practice. Larger
standard errors will, in turn, result in less significant t-statistics
and p-values. Obviously, this is not the best imaginable model for
predicting LOFT miss distance. However, it is the best possible model
given the limitations of existing data.
 - 56 -

Interpretation of the regression coefficients is essentially the

same as for the LALD/HD bomb delivery model described earlier. The
constant represents the average bomb miss distance for a non-instructor
WSO with zero loft deliveries in the preceding 30 days. In this case
this is approximately 85.6 meters. The instructor coefficient indicates
that, on average, instructor’s LOFT bombs land about 13.4 meters closer
to the target than non-instructors with the same amount of recent
practice. And finally, the LOFT coefficient is an estimate of the
incremental effect, controlling for instructor status, of each
additional practice LOFT delivery in the preceding 30 days on average
bomb miss distance measured in terms of CEA. In this case an additional
practice delivery reduces CEA by about 1.6 meters.
Estimates for CEA and standard deviations were estimated for both
instructors and non-instructors while previous practice varied from zero
to 12 LOFT deliveries in the previous month. The CEAs, USAF hit
criteria, and standard deviations were then used to estimate the
probabilities of hit shown in figure 5.8 below.
 - 57 -

F-15E Radar Bomb Probability of Hit

0.9
Non-instructor
Mean When Not Deployed
Instructor
0.8

Predicted Ph
0.7

0.6

0.5

0.4
0 1 2 3 4 5 6 7 8 9 10 11 12
Radar Bomb Deliveries Last 30 Days

Figure 5.8 -- Radar delivery probability of hit for F-15E instructor-and

non-instructor WSOs as a function of bombs dropped in the past 30 days.8

What do these predicted hit probabilities tell us? First of all,

again it seems instructors have an advantage when it comes to hit
probability. Here again, non-instructors with normal recent practice
achieve hit probabilities approximately equal to instructors with no
practice. When not deployed in support of a peace operation the average
WSO – whether instructor or not - in this data set dropped 3 practice
LOFT bombs per month. However, just as with visual bombs, the
probability of non-instructors dropping acceptable bombs declines for
deployed crews who can not practice LOFT deliveries. The probability a
non-instructor WSO with no practice in the previous month hits a target
with a LOFT delivery is only about 0.62 compared to approximately 0.71
for a non-instructor pilot dropping visual bombs with no recent
practice. Similar probabilities for instructors are 0.67 and 0.76. Here
again the probability crews with no recent practice will miss their
____________
8 Note: The radar bomb practice effect is not significant at the
p=0.05 level.
 - 58 -

intended targets increases by 20 percent for non-instructors and 25

percent for instructors.

Estimated Impact of Deployments on F-15E Squadron Low Level Bomb

Delivery Performance
The coefficients from the regression analyses of both visual and
radar bomb deliveries can be combined with data on the average number of
practice deliveries F-15E crews from the 335th Fighter Squadron
accomplished between January 1 and November 30 1997 to estimate the hit
probabilities for instructors and non-instructors. Figure 5.9 below
shows the results for pilots dropping visual bombs.

Predicted Ph Deployed Crews Visual

Deliveries

0.9

0.8

0.7

0.6
Instructors
0.5 Non-Instructors

0.4
1 2 3 4 5 6 7 8 9 10 11
Month

Figure 5.9 – Predicted Probability of Hit (Ph) for Visual Bombs for F-
15E Crews Who Deployed to Saudi Arabia February through June 1997

Figure 5.9 presents estimated probabilities of hit for instructors

and non-instructors assigned to the 335th Fighter Squadron based on the
average number of practice LALD/HD deliveries accomplished by each group
per month from January through November 1997. The squadron deployed to
 - 59 -

Saudi Arabia from February through June.9 During the deployment none of
the pilots in Saudi Arabia accomplished a single practice LALD/HD bomb
delivery due to the lack of training facilities and extreme demands no-
fly zone sortie commitments placed on the squadron’s available sorties.
As one might expect from the preceding discussion, both
instructors’ and non-instructors’ LALD/HD accuracy declines with lack of
practice lack of practice. Within 30 days – and probably less – of
arriving in Saudi Arabia, crews chances of hitting a target with a
LALD/HD bomb had decreased from about ten percent from their non-
deployed average where they remained for the rest of the deployment.
Bomb accuracy does not decline further as the months without practice go
by because, as described previously, an analysis of bombs dropped in the
past 03, 60 and 90 days indicated only bombing practice within the most
recent 30 days had a statistically significant effect on bombing
accuracy. In other words, the statistical analysis indicates that there
is no significant difference in bombing skills between crews who have
gone 30 days without practice and those who have gone 90 days without
practice. Skills may deteriorate beyond 90 days, but it is not possible
to determine if or how much they decline with the data available.
Figure 5.10 below presents similar predictions for radar aimed LOFT
deliveries.

____________
9 Numbers on the horizontal axis of figures 5.9 through 5.11
correspond to months of the year (e.g. 1 = January, 2 = February, etc.).
 - 60 -

Predicted Ph Deployed Crews Radar

Deliveries

0.9

0.8

0.7

0.6

0.5

0.4 Instructors
1 2 3 4 5 6 7 8 9 10 11 Non-Instructors
Month

Figure 5.10 -- Predicted Probability of Hit (Ph) for Radar Bombs for F-
15E Crews Who Deployed to Saudi Arabia February through June 1997

A final step toward an estimate of the overall impact of lack of

practice on the probability the squadron as a whole will hit – or miss –
targets it might be tasked to attack from low altitude is to compute a
weighted average of the instructor and non-instructor hit probabilities.
Approximately 35 percent of both pilots and WSOs in the 335th Fighter
Squadron are instructors. So, the overall probability the squadron will
hit its assigned visual and radar targets is closer to the non-
instructor probability of hit than the instructor probability of hit.
Figure 5.11 below presents the other side of the probability of hit
analysis shown so far by presenting estimates of the overall chances
pilots and WSOs of the 335th Fighter Squadron would have MISSED assigned
low level targets during 1997.
 - 61 -

0.45

0.4

0.35

0.3
Predicted Pmiss
0.25 deployed crews Visual
0.2 Predicted Pmiss
deployed crews Radar
0.15

0.1

0.05

0
1 2 3 4 5 6 7 8 9 10 11
Month

Figure 5.11 – Predicted Probability of Miss for F-15E Crews Who Deployed
to Saudi Arabia 1997

Figure 5.11 clearly illustrates how the overall probability a

squadron will miss assigned low level targets increases about 20 percent
when the squadron deploys in support of peace operations for both visual
and radar bomb deliveries. So, if crews deployed to a peace operation
were suddenly faced with the requirement to conduct low altitude strikes
against well defended targets, it is likely that many more targets would
require re-attack due to inaccurate bomb deliveries than one might
expect based on their performance just prior to deployment. This would
be inconvenient and somewhat costly, but since crews would rapidly
regain proficiency as operations progressed, the impact of decreased
accuracy would probably be both temporary and relatively minor when
compared to the overall air campaign objectives and timetable of a MTW.
Figures 5.12 and 5.13 below illustrate how bombing accuracy might
impact the number of targets hit in the opening days of a MTW air
campaign. The figures use data from Operation Desert Storm as a
 - 62 -

starting point.10 The top line in each figure shows the cumulative
total of targets attacked by F-15Es and F-16s. During this phase of
Desert Storm there were 48 F-15Es and 210 F-16s in theater. On average
these aircraft flew about one sortie per day and attacked one target per
sortie. The middle line on each chart uses the predicted squadron hit
probabilities for F-15E loft (presented above) and F-16 visual
deliveries (presented in the next section) assuming crews had an average
number of practice loft deliveries in the previous 30 days at the start
of the war, and that combat bomb deliveries improve accuracy to the same
extent as practice deliveries.11 The bottom line on each chart assumes
crews start the notional replay of Desert Storm while in the middle of a
peace operations deployment with no practice bomb deliveries in the past
30 days. F-15E WSOs hit about 12 percent fewer targets on the first
night without practice than they would with practice, while F-16 pilots
with no practice miss about 18 percent more targets on the first day.
After ten days in action, this analysis predicts F-15E WSOs who start
the war with no recent practice will have hit about 95 percent of the
targets they would have if they had begun the war with normal peacetime
training in the previous month. F-16 pilots without recent practice are
predicted to have hit about 88 percent of the targets they would have in
the first ten days with practice.

____________
10 See GWAPS Vol. V pp. 419-429 and 556-565 for data on F-15E and
F-16 aircraft in theater, and targets attacked during the first ten days
of Operation Desert Storm.
11 During the opening days of Desert Storm almost all F-16C ground
attack sorties were flown during daylight hours and used visual bomb
delivery techniques while about 90 percent of F-15E sorties were flown
at night and used radar bomb delivery techniques.
 - 63 -

F-15E Loft Targets

500

450

400
Targets

350
F-15E Targets Attacked
300
F-15E Perdicted Hits With
250
Cumulative

Practice
F-15E Predicted Hits No
200 Practice

150

100

50

0
1 2 3 4 5 6 7 8 9 10
Day

Figure 5-12 – F-15E Targets Attacked and Hit by Crews With and Without
Practice in a Notional Replay of the First Ten Days of Desert Storm

F-16 Visual Targets

2500

2000
Targets

1500
F-16 Targets Attacked

F-16 Practice Cumulative

Cumulative

F-16 No Practice Cumulative

1000

500

0
1 2 3 4 5 6 7 8 9 10

D a y

Figure 5-13 – F-16C Targets Attacked and Hit by Crews With and Without
Practice in a Notional Replay of the First Ten Days of Desert Storm
 - 64 -

However, these predictions represents an upper limit of how well

crews who transition directly from peace operations to combat missions
would do compared to crews fresh from normal peacetime training. To the
extent that decreased low level bombing accuracy is a proxy for the loss
of other important tactical skills crews need to successfully carry out
demanding missions during the opening days of a MTW, we can expect crews
ability to navigate at low altitude, employ electronic systems and
countermeasures, evade threats, maintain tactical formation, acquire and
engage a target(s) etc. to decrease as well. Decreased performance along
any or all of these dimensions of tactical air operations will almost
certainly lead to additional missed targets due to navigation errors,
and possibly to increased losses to enemy defenses and increased “ground
kills” resulting from crews with insufficient recent practice attempting
to execute tactical maneuvers and threat reactions in the unforgiving
low altitude environment.
In addition, since successful mission accomplishment requires crews
to competently perform all of these demanding tasks, the decline in
bombing accuracy discussed above almost certainly underestimates the
overall impact of lack of tactical training on the ability of USAF
fighter crews to successfully accomplish typical early MTW ground attack
mission profiles. For example, suppose fighter crews must accomplish
ten tasks such as those just enumerated in order to successfully
complete a combat mission. If lack of practice reduces their ability to
perform each individual task by ten percent (as it does with bomb hit
probabilities), then the overall probability of successfully
accomplishing the mission is not 90 percent. Because each task must be
accomplished successfully, the probability of successfully accomplishing
the mission is only 35 percent.12
In addition, the preceding analysis of bombing accuracy considers
only how lack of practice impacts the chances that an individual
aircraft will hit an individual target. It does not take into account
how lack of practice might impact the ability of a formation of two or
more aircraft to successfully coordinate their attacks against a target
____________
12 I am indebted to RAND colleague Alan Vick for this observation.
 - 65 -

or targets. Since most early MTW ground attack missions, especially

daytime visual bombing missions, will be conducted by multiple aircraft
flying and attacking in formation, simply aggregating the individual
impact of lack of practice probably overstates the probability a
formation of aircraft will successfully complete its mission.
Furthermore, it is important to note that the regression analysis
showed that practice improved instructor bombing accuracy just as much
as non-instructor accuracy. Even though instructors performance was
generally better than non-instructors, practice still helped them
perform better. So, even for seasoned, well trained and talented crews
recent practice is critical for MTW combat skill performance.
Finally, the discussion of hit probabilities in this section and
the next rest on the assumption that USAF hit criteria are adequate to
ensure the destruction of all types of targets USAF fighter crews might
be tasked to attack early in a MTW. While dropping 500 to 2000 pound
bombs within 25 to 30 meters (the USAF hit criteria for visual bombs) is
adequate to ensure destruction of some types of targets such as
petroleum storage facilities or personnel in the open, other important
target classes such as armored vehicles and artillery require much
higher accuracy. Therefore, the decline in hit probability discussed
above probably represents only the proverbial tip of the iceberg in
terms of the detrimental impact of peace operations deployments on USAF
fighter crews’ ability to successfully complete many important air-to-
ground combat missions during the opening phases of a MTW.

F-16C BOMBING SKILLS

In addition to looking at F-15 pilot and WSO low level bombing
skills, I also analyzed low level visual deliveries and medium altitude
laser guided bomb deliveries accomplished by pilots from the 388th
Fighter Wing, 4th Fighter Squadron.

F-16C Low Altitude Visual Bomb Deliveries

The data available from the 4th Fighter Squadron was almost
identical in terms of structure and variables to the data from the 335th
Fighter Squadron described in tables 5.1 and 5.2 above. The major
difference is that data were available over a 14 month period from
 - 66 -

January 1997 to February 1998 for 4th Fighter Squadron crews versus the
11 month period for the 335th Fighter Squadron crews. Another
difference is that the maximum number of LALD/HD deliveries used in this
analysis was 15 versus 18 for the F-15E data.13 An additional
difference is that the F-16C is a single seat aircraft, and radar aimed
Loft deliveries are not an important part of the squadrons combat
tactics. Accordingly, the only F-16 low level bomb delivery skills
analyzed here are for Low Angle Low Drag/High Drag visual bomb
deliveries.
Initially, just as with the F-15E LALD/HD and LOFT deliveries, a
variety of functional forms and variables were tested to determine which
model specification best fit the available data on CEA. The best model
specification turned out to be the same as that used for the F-15E data.
Here again, the model predicts average LALD/HD miss distance in terms of
CEA as a function of a pilot’s instructor status and the number of
LALD/HD deliveries he had performed in the past 30 days. Table 5.5
below summarizes the model results for 163 F-16C LALD/HD CEAs.

Variable Coefficient Standard t-Statistic P-Value

Error
N=166
Constant 30.30446 3.11003 11.442 0.000

Instructor -10.72795 3.41654 -3.140 0.002

Practice -0.5632309 0.23725 -2.374 0.019

(LDHD
Current
Month)

Table 5.5 – F-16C Visual Bomb Delivery Regression Coefficients

As with the two previous sets of regression results, interpretation

of these coefficients is relatively straightforward. The constant
____________
13 One very inexperienced pilot did accomplish 24 practice LALD/HD
deliveries in a single month with only mediocre accuracy. This
observation was dropped from the analysis because it was an extreme
outlier with high leverage, and was therefore driving the results of the
model.
 - 67 -

represents the average miss distance for non-instructor pilots who have
not dropped any practice LD/HD bombs in the preceding 30 days. The USAF
hit criteria for the mix of LALD and LAHD deliveries in this data again
about 27.5 meters.14 So, unlike non-experienced F-15E pilots who drop
bombs just inside their hit criteria on average, non-experienced F-16C
pilots in this data set tend drop their bombs outside the hit circle at
30.3 meters on average unless they have had some recent practice. The
instructor coefficient is simply the difference in average instructor
and non-instructor performance. In this case, F-16C instructors dropped
their bombs almost 11 meters closer to the target on average than did
non-instructors. Finally, the practice coefficient is the average
decrease in CEA for each additional practice bomb dropped. For F-16C
pilots this incremental increase in accuracy from each additional
practice bomb is estimated to be about 0.56 meters (approximately 22
inches) on average. Just as in the F-15E analysis, these coefficients
were combined with estimated standard errors for recent practice ranging
from zero to 15 LALD/HD deliveries for both instructors and non-
instructors to produce estimated probabilities of hit. These
probabilities are presented in figure 5.12 below.

____________
14 Multi-Command Instruction 11-F16, 1 July 1997.
 - 68 -

F-16 Visual Bomb Ph

1
Non-instructor
Mean When Not Deployed
0.9 Instructor

0.8

0.7

0.6

0.5

0.4

Visual Practice Bombs Last 30 Days

Figure 5.12 -- Visual delivery probability of hit for F-16C instructor

and non-instructor pilots as a function of bombs dropped in the past 30
days.

The figure shows the same general pattern as that produced by the
F-15E model outputs. Here again instructors have a considerable, and
even larger, advantage over non-instructors in terms of average hit
probability. In fact, non-instructors flying F-16Cs need to drop about
seven bombs per month to achieve an estimated probability of hit just
above 0.5 while F-16C instructors achieve an estimated hit probability
of about 0.75 with the same amount of practice

Estimated Impact of Deployments on F-16C Squadron Low Level Bomb

Delivery Performance

The estimated hit probabilities presented in table 5.12 were

combined with actual data on the average number of bombs instructor and
non-instructor pilots from the 4th Fighter Squadron dropped between
January 1, 1997 and February 28, 1998 to produce estimates of the
probability of hit for each group displayed in figure 5.13 below. The
 - 69 -

general pattern here is similar to the estimates for F-15E visual bomb
deliveries presented earlier. However, since non-instructor F-16C
pilots need considerable practice just to hit on average, the overall
estimated hit probabilities for the entire squadron are lower than for
their F-15E counterparts. The 4th Fighter Squadron made two deployments
to Saudi Arabia during the period under study here – one in March and
April of 1997 and another in December 1997 and January 1998.15 The
figure clearly illustrates the dramatic decline in LALD/HD hit
probabilities for both instructors and non-instructors during these
deployments.

____________
15 The estimated probability of hit of about 0.72 for November 1997
reflects a decreased level of training activity late in the month so
that necessary preparations for the up-coming deployment could be
completed. This period of decreased training tempo prior to a
deployment is necessary to allow squadron personnel time to take care of
personal and family business prior to a long absence, and to prepare the
aircraft and other equipment for the long trip to Saudi Arabia. A
similar “stand-down” period usually occurs at the end of a deployment as
well to allow squadron members some time off to reacquaint themselves
with their families, etc. As a result, 45-day deployments usually
result in 60 to 70 days of little or no training in the MTW combat
skills of interest here. 90-day deployments stretch this period to 120
days.
 - 70 -

F-16 Visual Bomb Ph

0.9
Predicted Ph

0.8
Non-Instructor
0.7
Instructor
0.6

0.5

0.4
Jan-97
Feb-97
Mar-97
Apr-97
May-97
Jun-97
Jul-97
Aug-97
Sep-97
Oct-97
Nov-97
Dec-97
Jan-98
Feb-98
Month

Figure 5.13 -- Predicted Probability of Hit (Ph) for F-16C Instructors

and Non-Instructors January 1997 Through February 1998.

Just as in the F-15E LALD/HD analysis presented earlier the

instructor and non-instructor hit probabilities were weighted to reflect
the relative proportions of instructors and non-instructors in the
squadron (again approximately 65 percent non-instructors and 35 percent
instructors) to produce an overall estimate of the proportion of targets
the squadron as a whole would have hit over the time period under study.
Figure 5.14 below presents the predicted probability the squadron would
miss LALD/HD targets (one minus the weighted average probability of hit)
for the 4th Fighter Squadron for 1997 and early 1998. It clearly shows
that for the squadron as a whole, the chances of missing a target
attacked from low altitude with visually aimed bombs increase 25 percent
during, and around, peace operations deployments.
Again, it is important to remember that this overall decrease in
low level bombing accuracy is a proxy for the loss of other important
tactical skills such as low level navigation, mutual support, threat
reactions etc. that almost certainly decline as well. It is worth re-
emphasizing that this decline in bombing accuracy probably represents a
lower bound for the negative impact of peace operations on MTW combat
 - 71 -

capability because the cumulative impact of relatively small decreases

in proficiency in a variety of mission critical skills generally not
practiced during peace operations sorties such as the ability navigate
at low altitude, employ electronic systems and countermeasures, evade
threats, maintain tactical formation, acquire and engage a target(s)
etc. is much greater than the decline in any single skill would
indicate. The cumulative impact of decreased combat skill proficiency
could easily result in even more missed targets and possibly increased
aircraft and pilot losses in the opening phase of a MTW for both non-
instructors and instructors.

F-16 Squadron Probability of Miss

0.5

0.4
Predicted Ph

0.3

0.2 F-16
Squadron Pm

0.1

0
Jan-97
Feb-97
Mar-97
Apr-97
May-97
Jun-97
Jul-97
Aug-97
Sep-97
Oct-97
Nov-97
Dec-97
Jan-98
Feb-98

Month

Figure 5.14 – Predicted Probability of Miss for F-16C Pilots 1997 and
Early 1998

SUMMARY OF F-15E AND F-16C LOW LEVEL BOMBING SKILL ANALYSIS

Figures 5.15 through 5.17 below present a graphic summary of the

low level bombing skill analysis for both F-15E and F-16C crews
presented above. Figure 5.15 presents a series of four predicted impact
 - 72 -

patterns for F-15E crews dropping visually aimed LALD/HD bombs. Each
pattern uses the average bomb miss distances and standard deviations
described above combined with random dispersion in azimuth to depict the
distribution of average miss distances for non-instructors and
instructors both with and without practice. The upper left pattern in
each figure presents results for non-instructors with no practice
deliveries in the preceding 30 days. The upper right pattern is for
instructors with no practice. The lower left pattern is for non-
instructors who have dropped the average number of practice deliveries
crews accomplished when not deployed, and the lower right pattern is the
same for instructors. In all cases the circle represents the maximum
limit of USAF hit criteria. Each of the 100 dots represents a bomb
dropped by a particular class of crew members (either instructors or
non-instructors) who have dropped a given number of bombs. For example,
suppose there are 100 non-instructors in a squadron. Further assume
none of them has dropped any practice bombs in the past 30 days. If we
send all of them to the bombing range, and have each drop one bomb we
would see a dispersion pattern similar to the pattern in the upper left
of each figure.16 For visual bombs the axes extend to + or – 100
meters, for radar bombs they extend + or – 300 meters.
In all cases, for both instructors and non-instructors, practice
deliveries decrease the average miss distance – drawing the impact
patterns closer to the center of the circle and decreasing the fraction
of bombs that fall outside the hit limit. It is worth noting the F-16C
non-instructor impact pattern is mostly outside the USAF hit criteria
for pilots with no practice in the past 30 days.
____________
16 Note: These figures assume regression error terms for average
bomb miss distance are normally distributed. The distribution of error
terms for all three regressions were examined using QQ plots to confirm
this assumption. In each case the error term distributions closely
approximated the normal distribution.
 - 73 -

100
90 100
F-15E Non- Instr No Practice LDHD
90 F-15E Instructors No Practice LALD/HD
80
70 80
70
60
60
50
50
40
40
30
30
20
F-15E Non- Instr No Practice LDHD
20 F-15E Instructors No Practice
10
10 LALD/HD
0
- 1 - 9 - 8 - 7 - 6 - 5 - 4 - 3 - 2- -110 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 0 1 0 0
00 0 0 0 0 0 0 0 0 0 0 - 1 - 9 - 8 - 7 - 6 - 5 - 4 - 3 - 2- -110 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 0 1 0
-20 00 0 0 0 0 0 0 0 0 0 0
-20
-30
-30
-40
-40
-50
-50
-60
-60
-70
-70
-80
-80
-90
-90
-100
-100

F-15E Non-Instructor No Practice LDHD F-15E Instructor No Practice LDHD

100 100
90 F-15E Non-Instr Practice LDHD 90 F-15E Instructors Practice LALD/HD

80 80
70 70
60 60
50 50
40
40
30 30
20 F-15E Non-Instr Practice LDHD 20 F-15E Instructors Practice LALD/HD
10 10
0 0
- 1 - 9 - 8 - 7 - 6 - 5 - 4 - 3 - 2- -110 0 10 20 30 40 50 60 70 80 90 10
- 1 - 9 - 8 - 7 - 6 - 5 - 4 - 3 - 2- -110 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 0 1 0
00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 - 200 0
-20
-30 -30
-40 -40
-50 -50
-60
-60
-70 -70
-80 -80
-90 -90
-100 -100

F-15E Non-Instructor Practice LDHD F-15E Instructor Practice LDHD

Figure 5.15 – Graphic Depiction of F-15E Low Level LALDHD Bomb Accuracy

300 300

F-15E Non-Instr No Practice LOFT F-15E Instr No Practice LOFT

150
150

F-15E Non-Instr No Practice LOFT

F-15E Instr No Practice LOFT

0 0
-300 -150 0 150 300 -300 -150 0 150 300

-150 -150

-300 -300

F-15E Non-Instructor No Practice LOFT F-15E Instructor No Practice LOFT

300 300

F-15E Non-Instr Practice LOFT 5E Instr Practice LOFT

150 150

F-15E Non-Instr Practice LOFT F-15E Instr Practice LOFT

0 0

-300 -150 0 150 300 -300 -150 0 150 300

-150 -150

-300 -300

F-15E Non-Instructor Practice LOFT F-15E Instructor Practice LOFT

Figure 5.16 – Graphic Depiction of F-15E Low Level LOFT Bomb Accuracy
 - 74 -

100 100
90 F-16C Non-Instr NO Practice LDHD 90 F-16C Instr NO Practice LDHD

80 80
70 70
60 60
50 50
40 40
30 30
20 20
F-16C Non-Instr NO Practice LDHD F-16C Instr NO Practice LDHD

10 10
0 0
- 1 - 9 - 8 - 7 - 6 - 5 - 4 - 3 - 2- -110 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 0 1 0 - 1 - 9 - 8 - 7 - 6 - 5 - 4 - 3 - 2 - -110 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 0 1 0
0 0 0 0 0 0 0 0 0 0 - 200 0 0 0 0 0 0 0 0 0 0 0 - 200 0
-30 -30
-40 -40
-50 -50
-60 -60
-70 -70
-80 -80
-90 -90
-100 -100

F-16C Non-Instructor No Practice LDHD F-16C Instructor No Practice LDHD

100 100
90 F-16C Non-Instr Practice LDHD 90 F-16C Instr Practice LDHD

80 80
70 70
60 60
50 50
40 40
30 30
20 F-16C Non-Instr Practice LDHD 20 F-16C Instr Practice LDHD
10 10
0 0
- 1 - 9 - 8 - 7 - 6 - 5 - 4 - 3 - 2- -110 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 0 1 0 - 1 - 9 - 8 - 7 - 6 - 5 - 4 - 3 - 2 - -110 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 0 1 0
0 0 0 0 0 0 0 0 0 0 - 200 0 0 0 0 0 0 0 0 0 0 0 - 200 0
-30 -30
-40 -40
-50 -50
-60 -60
-70 -70
-80 -80
-90 -90
-100 -100

F-16C Non-Instructor Practice LDHD F-16C Instructor Practice LDHD

Figure 5.17 – Graphic Depiction of F-16C Low Level LALDHD Bomb Accuracy

F-16C MEDIUM ALTITUDE LASER GUIDED BOMB DELIVERIES

Before closing this discussion of the relationship between

experience, practice and bombing accuracy it is important to examine
what happens to the ability of USAF aircrew to deliver precision guided
munitions from medium altitude as a result of deployments in support of
peace operations. While delivering laser guided bombs (LGBs) from
medium altitude is not a skill F-15E or F-16C crews are likely to be
called upon to use during the opening engagement phase of a MTW, the
ability to accurately deliver LGBs is important during the later
exploitation phase of a MTW. In addition, these skills may also be
useful in enforcing the no-drive zone portion of peace operation
missions or other contingency operations.
Figure 5.16 below illustrates a typical medium altitude LGB
delivery. To accomplish this type of bomb delivery the pilot (or in the
case of the F-15E the WSO) of the aircraft usually uses the aircraft’s
radar or computerized navigation systems to roughly identify a target’s
 - 75 -

location and slew the aircraft’s imaging infrared targeting pod sensor
to the approximate target location. The operator then uses a joystick
to center the sensor’s crosshairs on the target. When the aircraft’s
computer systems confirm the target is close enough for the LGB to reach
it the operator drops the LGB. Once clear of the aircraft the LGB is
guided by invisible coded laser signals transmitted by the aircraft’s
targeting pod and reflected from the target. A small computer takes
inputs from the sensor that detects the reflected laser signals and
transmits steering commands to guidance fins mounted on the nose of the
bomb. The bomb continues steering itself toward the target until
impact.

Medium Altitude LGB Deliveries

2 1

Target

Figure 5.16 – Typical Medium Altitude LGB Delivery

If the crosshairs of the targeting system remain centered on the

target, the bomb was within range when it was dropped, the laser is
activated at the proper time and functions properly, and the bomb’s
guidance system works as advertised a hit is almost assured. This
guidance system allows crews to train without actually dropping
 - 76 -

practice, or live, ordinance. All elements of the system and relevant

operator skills, can be assessed by viewing video tapes of the targeting
pod images and accompanying displays after the mission with the
exception of the functioning of the bombs guidance system. So, if post
mission review indicates the crosshairs were on the target, simulated
bomb release was within range, and the laser activated, then the
operator has done all that could be expected in an actual delivery, and
the simulated attack can be graded a hit. If, however, the operator
makes some sort of error in identifying or tracking the target, or in
dropping the bomb or activating the laser, the simulated attack is
graded a miss.
Beginning on October 1, 1997 the 4th Fighter Squadron began keeping
detailed records on all simulated LGB attacks by its pilots. Between
October 1, 1997 and February 28 1998 pilots of the 4th Fighter Squadron
accomplished 465 simulated medium altitude LGB attacks. In contrast to
low altitude LALD/HD, they were able to continue to train for these
types of attacks even while deployed to Saudi Arabia during December
1997 and January 1998 because they could accomplish this training while
flying missions in support of the Operation Southern Watch no-fly zone.
This was possible because pilots could lock their targeting pods onto
convenient objects and structures inside the no-fly zone and practice
simulated LGB deliveries while on patrol. There was no statistically
significant difference in the number of simulated attacks pilots
conducted per month while deployed to Saudi Arabia and the number
conducted when at home station. Table 5.6 below shows how much more
consistent LGB training was than LALD/HD training for the same pilots
over the same time period.
Table 5.6 illustrates how much more consistent LGB practice was
than LALD/HD for the 4th Fighter Squadron throughout this period. On
average pilots time between practice LGB deliveries was only one quarter
that of LALD/HD deliveries. In addition, the variation about the LGB
mean was much less than that around the LALD/HD mean. Pilots went a
week or less between practice LGB deliveries 50 percent of the time, two
weeks or less 75 percent of the time and, as the table shows, dropped a
practice LGB every four weeks or less 90 percent of the time. For
 - 77 -

LALD/HD practice comparable figures are 20 days, 68 days and 110 days,
respectively.

Delivery Mean Standard Deviation 90 percent <

LGB 10.9 11.4 28
LALD/HD 41.1 45.1 110
th
Table 5.6 – Mean, Standard Deviation, and 90 Percentile Days Between
F-16C LGB and LALD/HD Practice Deliveries October 1997 Through February
1998

The average frequency and consistency of practice with LGB

deliveries make it impossible to detect a practice effect with the
available data.17 This is probably because very few pilots went long
enough between deliveries for their proficiency to decline to the point
where a detectable decrease in their hit probability occurred. It might
be possible to determine this relationship with a much larger data set –
one containing several thousand practice LGB deliveries – but the 465
observations available do not have sufficient statistical power to
detect the very small differences in proficiency that result from the
relatively small variation in practice frequency.
However, just as with the low altitude bombing skills, there was a
large difference in the hit probabilities of instructor and non-
instructor pilots.18 On average, instructors were assessed to have hit
their targets in 87.5 percent of the 136 simulated LGB attacks they
accomplished. In contrast, non-instructors were judged to have hit in
only 79.3 percent of their 329 simulated LGB deliveries. This
difference is statistically significant19, and reinforces the advantage
____________
17 Here, again, a wide variety of model specifications were
evaluated including linear, log and other non-linear transformations of
the practice variables. None produced significant coefficients for the
practice terms.
18 As in the low altitude analysis other measures of experience
such as the number of hours a pilot had flying fighters, total flying
hours, instructor hours, etc (and various transformations of them) were
also evaluated. None performed as well in explaining the hit
probability difference as instructor status.
19 Pearson chi-squared test with 1 degree of freedom equals 4.2982
(p=0.038).
 - 78 -

superior talent, training and experience gives instructors as a group

where air-to-ground munitions delivery skills are concerned.

SUMMARY
This chapter has presented a short description of two different
types of low altitude bomb deliveries. In addition, it has described
the data on bomb delivery practice, accuracy and pertinent pilot
characteristics collected from an F-15E squadron from the 4th Fighter
Wing and an F-16C squadron from the 388th Fighter Wing. It summarizes
the extensive statistical analysis carried out to investigate the
relationship between various measures of experience, recent practice and
bomb delivery accuracy (and by extension other combat skills important
during the opening phase of a MTW). It also presents an analysis of F-
16C medium altitude LGB deliveries. The most important results of this
analysis can be summarized as follows:
• Instructor status is a significant and powerful predictor of bomb hit
probability for a variety of precision and non-precision bomb
deliveries
• Measures of recent practice, while not always as statistically
significant, are good predictors of low level bombing accuracy and
hit probability.
• Practice is especially important in maintaining the combat
effectiveness of non-instructors as measured by low altitude bombing
accuracy.
• Lack of practice during peace operations deployments has a dramatic
negative impact on individual, and overall squadron, bomb accuracy
and probability of hit.20
• The probability a squadron will miss assigned low level targets
increases by at least 25 to 30 percent during peace operations
deployments.
• Accuracy returns rapidly once normal training resumes.
____________
20 The same sort of erosion in MTW combat skill proficiency could
result from Air Expeditionary Force (AEF) deployments to various crisis
spots around the world if crews are not afforded the opportunity to
practice tactical skills.
 - 79 -

• Unlike low level bombing accuracy, medium altitude LGB hit

probability does not seem to be negatively effected by peace
operations deployments.

The next chapter will present an analysis of data on valid and

invalid simulated air-to-air missile shots from the 4th Fighter Squadron
to investigate the relationships between pilot experience, practice and
air-to-air combat skill proficiency.