Operation Desert Storm:
Evaluation of the Air Campaign
(Letter Report, 06/12/97, GAO/NSIAD-97-134)


Appendix I

The data we analyze in this report are the best information collected
during the war.  They were compiled for and used by the commanders
who managed the air campaign.  These data also provided the basis for
postwar Department of Defense (DOD) and manufacturer assessments of
aircraft and weapon system performance during Desert Storm.  We
balanced the limitations of the data, to the extent possible, against
qualitative analyses of the system.  For example, we compared claims
made for system performance and contributions to what was supportable
given all the available data, both quantitative and qualitative.  In
the subsequent appendixes, we use these data to describe and assess
the use of aircraft and weapon systems in the performance of
air-to-ground missions.  And to the extent that the data permit, we
assess the claims for and relative effectiveness of individual
systems.  Finally, we use these data to discuss the overall
effectiveness of the air campaign in meeting its objectives. 

--------------------------------------------------------- Appendix I:1

In this report, we assess the effectiveness of various U.S.  and
allied air campaign aircraft and weapon systems in destroying ground
targets, primarily those that fall into the category of "strategic"
targets.  In Operation Desert Storm, some targets were clearly
strategic, such as Iraqi air force headquarters in Baghdad, while
others, essentially the Iraqi ground forces in the Kuwaiti theater of
operations, could be considered both strategic and tactical.  For our
purposes, we concentrated on the effects achieved by the air campaign
before the start of the ground offensive, including successes against
ground forces in Kuwait.  Unlike most previous large-scale conflicts,
the air campaign accounted for more than 90 percent of the entire
conflict's duration.  Therefore, what we have excluded from our
analysis is the role of air power in supporting ground forces during
the ground offensive ("close air support"), as well as such
nonstrategic missions as search and rescue. 

We evaluated the aircraft and munitions that were deemed to have had
a major role in the execution of the Desert Storm air campaign by
virtue of their satisfying at least one (in most cases, two) of the
following criteria:  the system (1) played a major role against
strategic targets (broadly defined); (2) was the focus of
congressional interest; (3) may be considered by DOD for future major
procurement; (4) appeared likely to play a role in future conflict;
or (5) even if not slated currently for major procurement, either was
used by allied forces in a manner or role different from its U.S. 
use or used new technologies likely to be employed again in the
future.  These criteria led us to assess the A-6E, A-10, B-52,
F-111F, F-117A, F-15E, F-16, F/A-18, and British Tornado (GR-1).  We
examined both guided and unguided munitions, including laser-guided
bombs, Maverick missiles, Navy cruise missiles, and unguided "dumb"
bombs.  (We did not examine Air Force cruise missiles because so few
were used.)

We focused our analysis on strategic targets in part because they
received the best-documented bomb damage assessments (BDA), although
there was very substantial variation from target to target and among
target types in the quantity and quality of BDAs.  Twelve categories
of strategic targets in Desert Storm are listed in table I.1.  With
the exception of mobile Scud launchers and ground forces, each type
of target was a fixed item at a known location on which battle damage
assessments were possible. 

                               Table I.1
                 Twelve Strategic Target Categories in
                     the Desert Storm Air Campaign

tion      Target category
--------  ------------------------------------------------------------
C\3       Command, control, and communication facilities

ELE       Electrical facilities

GOB       Ground order of battle (Iraqi ground forces in the Kuwait
          theater of operations, including the Republican Guard)\a

GVC       Government centers

LOC       Lines of communication

MIB       Military industrial base facilities

NAV       Naval facilities

NBC       Nuclear, biological, and chemical facilities

OCA       Offensive counterair installations

OIL       Oil refining, storage, and distribution facilities

SAM       Surface-to-air missile installations

SCU       Scud missile facilities
\a In our database, GOB targets are in the kill box target set. 

--------------------------------------------------------- Appendix I:2

------------------------------------------------------- Appendix I:2.1

To examine how the different types of aircraft and munitions
performed and were used to achieve the air campaign objectives, we
required data on the aircraft missions flown and missiles launched
against each type of target.  To assess the effectiveness of the
aircraft and munitions, we needed data on the outcome of each
aircraft and missile tasked (what was dropped or launched and where
it landed) as well as the physical and functional impact of the
munitions on the targets.  We had to review DOD and manufacturers'
Desert Storm claims for selected weapon systems and seek out data to
validate their assertions. 

To assess the relative costs of the systems employed, we needed
various cost measures of the systems and sufficient data on their
effectiveness to be able to relate cost and performance.  To examine
operating conditions of the air campaign, we required data on the
characteristics of the Iraqi threat, political and military operating
conditions in the theater, and the environmental conditions in which
combat occurred. 

To determine the degree to which air campaign objectives were met
with air power, we required, first, data that described the campaign
objectives and the plans to achieve those objectives and, second,
data that addressed the outcome of air campaign efforts in pursuit of
air campaign objectives. 

We obtained descriptive data on objectives and plans from a series of
interviews and a review of the literature.  We interviewed 108 Desert
Storm veteran pilots, representing each type of aircraft evaluated,
with the exception of British Tornados.\1 We also interviewed key
Desert Storm planners and analysts from a wide spectrum of
organizations, both within and outside DOD.  (See table I.2.)

We also conducted an extensive literature search and reviewed
hundreds of official and unofficial documents describing the planning
for, conduct of, and performance by the various aircraft and
munitions used in the campaign, and we searched for documents on
Desert Storm operating conditions. 

To examine the nature and magnitude of Desert Storm inputs employed
against strategic target categories, as well as outcomes, we needed
two types of databases.  We needed the "Missions" database generated
by the Gulf War Air Power Survey (GWAPS) to assess inputs.  And we
needed the Defense Intelligence Agency's (DIA) phase III battle
damage assessment reports to assess Desert Storm outcomes. 

                                    Table I.2
                       Organizations We Contacted and Their

Organization                                 Location
-----------------------------------  ------  -----------------------------------
Air Combat Command                           Langley Air Force Base, Va.
Center for Air Force History                 Washington, D.C.
Center for Naval Analyses                    Alexandria, Va.
Central Intelligence Agency                  Langley, Va.
Defense Intelligence Agency                  Washington, D.C.
Department of Air Force,                     Washington, D.C.
Embassy of the United Kingdom                Washington, D.C.
Foreign Science and Technology               Charlottesville, Va.
Grumman Corporation                          Bethpage, N.Y.
Gulf War Air Power Survey (research          Arlington, Va.
Institute for Defense Analyses               Alexandria, Va.
Lockheed Advanced Development                Burbank, Calif.
McDonnell Douglas Corporation                St. Louis, Mo.
Naval A-6E Unit                              Oceana Naval Air Station, Va.
Naval F/A-18 Unit                            Cecil Naval Air Station , Fla.
Navy Operational Intelligence                Suitland, Md.
 Center, Strike Projection
 Evaluation and Anti-Air Research
 (SPEAR) Department
Office of the Chief of Naval                 Washington, D.C.
Office of the Secretary of Defense           Washington, D.C.
Rand Corporation                             Santa Monica, Calif.
Securities and Exchange Commission           Washington, D.C.
Survivability/Vulnerability                  Wright-Patterson Air Force Base,
 Information Analysis Center                  Ohio
Texas Instruments                            Dallas, Tex.
U.N. Information Center                      Washington, D.C.
U.S. Atlantic Fleet, Headquarters            Norfolk, Va.
U.S. Central Air Forces,                     Shaw Air Force Base, N.C.
U.S. Central Command, Headquarters           MacDill Air Force Base, Fla.
U.S. Space Command                           Cheyenne Mountain Air Force Base,
4th Tactical Fighter Wing                    Seymour Johnson Air Force Base,
48th Tactical Fighter Wing                   RAF Lakenheath, U.K.
49th Fighter Wing                            Holloman Air Force Base, N.Mex.
57th Test Group                              Nellis Air Force Base, Nev.
363rd Fighter Wing                           Shaw Air Force Base, S.C.
926th Fighter Wing (reserve)                 New Orleans Naval Air Station, La.

\1 We did not select pilots randomly, given constraints on their
availability, travel, and time.  The only requirement was that a
pilot had flown the relevant type of aircraft in a Desert Storm
combat mission.  In most cases, the pilots had flown numerous
missions.  The purpose of interviewing pilots was to receive as
direct input as possible from the aircraft and munition user rather
than views filtered through official reports.  In Operation Desert
Storm:  Limits on the Role and Performance of B-52 Bombers in
Conventional Conflicts (GAO/NSIAD-93-138, May 12, 1993), we assessed
the B-52 role in detail.  Where they were relevant, we incorporated
the data and findings from that report into our comparisons.  The
British government denied our requests to interview British pilots
who had flown in Desert Storm.  However, we were able to obtain some
official assessments of the British role in the air campaign, and we
questioned U.S.  pilots about their interactions with British pilots. 

------------------------------------------------------- Appendix I:2.2

The Missions database represents a strike history of air-to-ground
platforms and ordnance in the Persian Gulf War.  GWAPS researchers
compiled a very large computerized database on aerial operations in
the Gulf War from existing records.  It documents aircraft strikes on
ground targets, number and type of ordnance, date, and time on target
(TOT) information, target names and identifiers, desired mean point
of impact (DMPI), and additional mission-related information.  It
contains strike history information across the duration of the air
campaign for most of the air-to-ground platforms that participated. 
There are data on 862 numbered targets that together comprise more
than 1 million pieces of strike information. 

The Missions database also contains strike records across the
duration of the air campaign for most of the air-to-ground platforms
that participated in the Gulf War.  This database includes platforms
from the U.S.  military services and some non-U.S.  coalition
partners.  The Missions database was intended to provide information
not on aircraft sortie counts but, rather, on aircraft strike counts
and associated target attack information.  Further, it was not
intended to provide information on platform or munition

The selection criteria that guided our use of the database records
required us to select targets that were designated by a unique basic
encyclopedia (BE) number and an associated target priority code
(target category designation) and that were records of identifiable
U.S.  aircraft strikes or strikes conducted by the British Tornado,
GR-1 (interdiction variant).\2 We did not include records that did
not meet these criteria.\3 Also, we did not include A-10 records
because the majority of A-10 strike events as represented in the
database are unclear.\4 Finally, we did not include strike events
that were designated as ground aborted missions or headquarters
cancellations.  Unless indicated otherwise, the data we reviewed on
strategic target categories, the nine platforms, and their munitions
originate from this data set. 

Targets were assigned to target categories based on the AIF
functional target category designations.  (See table I.3.)

The AIF target category designations indicate broad categories of
strategic targets (for example, offensive counterair) as well as
provide more specific examples of individual target types within the
broad target categories (for example, hardened aircraft shelters). 
The AIF strategic target category referred to as ground order of
battle (GOB) was expanded to include all "kill box" targets that had
an assigned BE number, and it is subsequently identified in our
database as the KBX category.\5

                               Table I.3
                 AIF Target Categories and Target Types

Target category     Target type
------------------  --------------------------------------------------
Government control  Government control centers

                    Government bodies, general

                    Government ministries and administrative bodies,
                    nonmilitary, general

                    Government detention facilities, general

                    Unidentified control facility

                    Trade, commerce, and government, general

                    Civil defense facilities (in military use)

Electricity (ELE)   Electric power generating, transmission, and
                    control facilities

Command, control,   Offensive air command control headquarters and
and communications  schools

                    Air defense headquarters


                    Electronic warfare

                    Space systems

                    Missile headquarters, surface-to-surface

                    National, combined and joint commands

                    Naval headquarters and staff activities

Surface-to-air      Missile support facilities, defensive, general
missiles (SAM)

                    SAM missile sites/complexes

                    Tactical SAM sites/installations

                    SAM support facilities

Offensive           Airfields (air bases, reserve fields, helicopter
counterair (OCA)    bases)

                    Noncommunications electronic installations (radar
                    installations, radars collocated with SAM sites,
                    ATC/Nav aids, meteorological radars)

                    Air logistics, general (air depots)

                    Air ammo depots (maintenance and repair bases,
                    aircraft and component production and assembly)

Nuclear,            Atomic energy feed and moderator materials
biological, and     production
chemical (NBC)

                    Chemical and biological production and storage

                    Atomic energy-associated facilities production and

                    Basic and applied nuclear research and
                    development, general

Military            Basic processing and equipment production
industrial base

                    End products (chiefly civilian)

                    Technical research, development and testing,

                    Covered storage facilities, general

                    Material (chiefly military)

                    Industrial production centers

                    Defense logistics agencies

Scuds (SCU)         Guided missile and space system production and

                    Fixed missile facility, general

                    Fixed, surface-to-surface missile sites

                    Offensive missile support facilities

                    Medium-range surface-to-surface launch control

                    Fixed positions for mobile missile launchers

                    Tactical missile troops field position

Naval (NAV)         Mineable areas

                    Maritime port facilities

                    Cruise missile support facilities, defensive

                    Shipborne missile support facilities

                    Cruise surface-to-surface missile launch positions

                    Naval bases, installations, and supply depots

Petroleum, oil,     POL and related products, pipelines, and storage
and lubricants      facilities

Lines of            Highway and railway transportation

                    Inland water transportation

Ground order of     Military troop installations
battle (GOB)\a

                    Ground force material and storage depots

                    Fortifications and defense systems
\a In our database, GOB targets are in the kill box target set. 

While the Missions database contains an abundance of Desert Storm
strike history information, it has its limits.  Different reporting
procedures adopted during Desert Storm and the use of different
terminology and language, within and among services, have resulted in
more or less detailed data for particular platforms.  These
limitations in the final form of the database transfer to all users
of the database.  For example, in some instances, database records
documenting Air Force aircraft strikes may be more complete with
fewer missing observations than the same data for other service
platforms because services may have adopted different methods of
tracking and identifying outcomes during the war.  As stated
previously, GWAPS indicates that A-10 data are difficult to summarize
and interpret because of the way the data were initially recorded. 
Where relevant and necessary for this research, we consulted with the
appropriate GWAPS staff regarding limitations and usage of the
Missions database. 

Studies using the database for different purposes should not be
expected to generate identical data.  For example, the number of
strikes conducted by a particular platform against strategic targets
may not be equivalent across studies because of the degree of
specificity in the question being posed.  One study may be concerned
with strategic targets regardless of any other delimiting factors,
while another may be concerned with strike counts against strategic
targets, discounting those strikes where some mechanical failure of
the aircraft was reported to have occurred over the target area. 
Therefore, differences among studies that rely on the use of the
Missions database, in some form or another, should be interpreted
considering differences in research questions, methodologies, and

We also used the Missions database to create the variables to measure
air campaign inputs.  These variables are used to measure either the
weight of effort (WOE) or the type of effort (TOE) expended and are
defined in
table I.4. 

                               Table I.4
                 Definition of Composite Variables for
                          WOE and TOE Measures

Measure         Variable
--------  ----  ------------------------------------------------------
WOE             Quantity of BE numbers to which platforms were tasked
                Quantity of strikes that platforms conducted
                Quantity of bombs that platforms delivered
                Quantity of bomb tonnage that platforms delivered
TOE             Quantity of bombs that were guided bombs
                Quantity of bombs that were unguided bombs
                Quantity of bomb tonnage that was guided
                Quantity of bomb tonnage that was unguided
Other           Quantity of day and night strikes
The only variable in the list above that was directly accessible from
the Missions database was the number of BEs to which aircraft were
tasked.  All other variables were derived by us from the raw data
provided in the Missions database. 

\2 Designating targets by a BE number is a method of identifying and
categorizing target installations for target study and planning. 

\3 In several instances in which records met all selection criteria
except for a missing target category designation, we used all
available target-identifying information and assigned the target to a
target category based on automated intelligence file (AIF) target
category designations. 

\4 At least one-third of the A-10 strike data could not be accurately
determined from the original records, and GWAPS researchers were not
able to reconcile the inconsistencies. 

\5 Kill boxes were areas where the Republican Guard (RG) and other
Iraqi troops were dug in.  According to GWAPS, the vast majority of
kill box strikes were directed against GOB targets.  However, GWAPS
did not include the universe of BE-numbered kill boxes in the GOB
target category.  Therefore, we expanded the GOB target category to
include all BE-numbered kill boxes and subsequently identified it as
the KBX category.  GWAPS indicates that approximately 8 percent of
kill box strikes were conducted against targets other than GOB
targets.  Examination of the database indicates that these other
target types include SAM sites, artillery pieces, and some bridges. 

----------------------------------------------------- Appendix I:2.2.1

Quantity of BE Numbers.  BE numbers are a method of categorizing and
identifying various types of target installations for target study
and general planning.  The number of BEs are only considered an
approximation of the actual number of targets or desired mean points
of impact (DMPI) that aircraft were assigned to and may have struck. 
The quantity of BE numbers can only be considered an approximation
because a single BE number can encapsulate more than a single DMPI. 
For example, an entire airfield may be assigned a single BE number,
yet there may exist multiple DMPIs on that airfield (hardened
aircraft shelters) that could potentially inflate the actual number
of targets.\6

Quantity of Strikes.  We used the GWAPS method of assessing strike
counts based on Missions data.  We excluded only those strike efforts
that were most likely not to have expended some actual weight of
effort against targets.  For example, we included strike events from
the database that were signified as weather-aborted or canceled,
without reference to why or whether or not the cancellation occurred
over the target or on the ground before takeoff.  Aircraft that
arrived at the target area, and then the strike events were canceled
because of weather, still represented a part of the weight of effort
that was expended on a target.  This is because numerous resources
are required simply to get the aircraft safely to the target (for
example, tankers, planning time and resources, airborne warning and
control system (AWACS) resources, and possibly escort and SEAD
aircraft).  As concluded by GWAPS researchers, their database has
inconsistent abbreviations and meanings attached to the codes for
canceled missions.\7 This lack of consistency and clarity suggests
that using mission cancellation codes as a filter for strike summary
information is not reliable, and therefore, we did not use them. 

Quantity of Bombs.  The quantity of bombs was determined from those
database fields that provided some information on the number of bombs
that an aircraft delivered and the number of aircraft that delivered
it.  If the database fields listing the quantity of bombs were empty,
bomb quantities for those strike events were not determined.\8 The
quantity of bombs measure does not include clearly designated
air-to-air ordnance, aircraft gun ordnance, decoys, or psyop delivery

Quantity of Bomb Tonnage.  The quantity of bomb tonnage was
determined by entering a new variable into the database representing
the weight of air-to-ground bombs (in pounds), summing these weights,
and then dividing the sum by 2,000 to determine the overall amount of
bomb tonnage.  The quantity of bomb tonnage could only be calculated
for those entries in the database where a verifiable type and
quantity of bomb actually appeared.\9

\6 The lack of consistently detailed DMPI indicators in the database
does not permit a reliable estimate of the actual number of targets
represented by individual BE-numbered targets within all target
categories.  Because the database contains at least two fields to
capture information on DMPIs, there could be at least two DMPIs per
BE number.  This would effectively double the number of targets. 
Therefore, at most, the 862 BE-numbered targets in our database may
be the lower bound of the actual number of targets. 

\7 Gulf War Air Power Survey, vol.  V, pt.  I:  Statistical
Compendium and Chronology (Secret), pp.  425-26. 

\8 Approximately 2 percent of the database records used in the
analysis, and which provide designation of the primary type of
aircraft ordnance, were blank. 

\9 The quantity of bomb tonnage is obviously a function of
information on the quantity of bombs.  Thus, the baseline percentage
of database records where information on bomb tonnage could not be
calculated is 2 percent--as noted in the previous footnote. 

----------------------------------------------------- Appendix I:2.2.2

Quantity of Guided and Unguided Bombs.  The quantity of guided and
unguided bombs was calculated in the same manner as the quantity of
bombs described previously; however, ordnance was categorized
according to whether it was precision-guided or unguided. 

The ability to determine guided and unguided bomb categorizations was
dependent on the way that ordnance was designated in the database. 
If the type of bomb was clearly indicated in the Missions database,
then the category to which it belonged--guided or unguided--could be
determined.  In many cases, if bomb types were unclear or missing
(thus not permitting clear categorizations), those bombs would not
have been categorized.\10

However, in those instances in which a bomb type was unclear but
additional information permitted a categorization, bomb
categorizations were done.  For example, it was not unusual to see an
entry like `27X' in the database field that was supposed to contain
the primary type of aircraft ordnance.  In many cases, examination of
the type of aircraft that was associated with the ordnance would
indicate what type of ordnance it was.  Using the example above,
aircraft ordnance entries like `27X' had other data indicating that
the delivery platform was an F-117; thus, the bomb was assumed to be
a GBU-27 and a guided categorization would have been provided. 

Quantity of Guided and Unguided Bomb Tonnage.  The method and
restrictions for calculating guided and unguided bomb tonnage are the
same as those described previously under the WOE Variables section. 

\10 Estimates are approximately the same as noted previously--about 2
percent of the database records used in the analysis. 

----------------------------------------------------- Appendix I:2.2.3

The time at which strikes occurred was determined from the time on
target variable provided in the Missions database.  TOTs, designated
in Zulu time, were translated to an air tasking order (ATO) time to
determine whether strike events were occurring during daylight or
night hours.  A key provided by GWAPS indicated the ATO hours
associated with daylight and night hours.\11

\11 GWAPS, vol.  V, pt.  I (Secret), p 558. 

------------------------------------------------------- Appendix I:2.3

The Defense Intelligence Agency (DIA) generated battle damage
assessments during Operation Desert Storm in support of U.S.  Central
Command (CENTCOM).  The DIA's phase III reports detailed the extent
of physical and functional damage on strategic targets based on
multiple intelligence sources.\12 DIA prepared phase III BDA reports
only for targets identified by CENTCOM.  These targets were of
special interest to CENTCOM and lent themselves to data collection
from national sources.  The phase III analyses reported the degree to
which campaign objectives were met at a BE-numbered target at a
specific point in time.\13 These reports did not necessarily assess
the impact of any one mission or strike package; rather, they
assessed the effect of the cumulative efforts of the air campaign on
the function and capability of a specific target.  After assessing
all sources of intelligence to determine the functional damage
achieved at a target, DIA made a summary recommendation of whether a
restrike was needed. 

Phase III reports were written for 432 fixed strategic targets.  The
number of strategic targets assessed by DIA is only somewhat over
half the number of strategic targets CENTCOM identified by the end of
the war (772) and half the number of the BE-numbered targets
identified in GWAPS' Missions database (862).  In addition, these
targets were not necessarily representative of the entire strategic
target set.\14 However, they do represent the targets of greatest
interest to CENTCOM planners.  CENTCOM's level of interest is
reflected in the repeated assessments requested for and conducted on
some key targets; several of the targets were assessed over 10 times. 

The phase III reports do not provide strike-by-strike functional BDA
for each strategic target, but they represent the best cumulative
all-source BDA available to planners during the course of the war.\15
Though a few agencies produced postwar BDA analyses on narrowly
defined target sets, no other agency or organization prepared BDA
reports comparable to DIA's, which drew upon multiple sources and
assessed hundreds of diverse targets throughout the theater.\16

\12 Intelligence sources included imagery from national sources,
human intelligence, signal intelligence or electronic intelligence,
and tactical reconnaissance. 

\13 DIA also produced phase I and II reports during the war.  Phase I
reports identified whether a target was hit or missed on a specific
mission.  These reports contained the initial indications from the
imagery and were transmitted orally to the theater.  Phase II reports
were more detailed than phase I reports, describing the extent of
physical damage as well as functional impact based on imagery. 
Phase III reports also provided functional BDA to the theater but
required more time because they were based on a fusion of all
available intelligence sources rather than imagery alone. 

\14 Our data sources did not provide us with some detailed target
information such as number and characteristics of DMPIs, threat
environment, campaign objectives, or Iraqi adaptations or
countermeasures that would enable us to compare targets assessed by
DIA and those that were not. 

\15 Gulf War planners who were frustrated with the timeliness,
coverage, and occasionally the conclusions of BDA based primarily on
imagery increasingly relied on aircraft video to assess strike
success.  One blackhole planner stated that strike BDA was assessed
in theater based on F-117, F-15E, and F-111F video (taken during the
delivery of laser-guided bombs) and restrikes were postponed until
phase III reports confirmed or refuted the cockpit video.  Thus,
during the campaign, for some targets, BDA and restrike
determinations were supplemented by--but not wholly replaced
by--cockpit video. 

\16 See Central Intelligence Agency, Operation Desert Storm:  A
Snapshot of the Battlefield (Sept.  1993); Defense Intelligence
Agency, Vulnerability of Hardened Aircraft Bunkers and Shelters to
Precision Guided Munitions (Apr.  1994); Foreign Science and
Technology Center, Desert Storm Armored Vehicle Survey/BDA
(Charlottesville, Va.:  Joint Intelligence Survey Team, Jan.  1992). 

----------------------------------------------------- Appendix I:2.3.1

We used phase III reports on fixed strategic targets to determine the
extent to which the functional capabilities of the target had been
eliminated.\17 Using the final BDA report prepared during the
campaign on each target, we assessed whether the campaign against
that target had been fully successful or not fully successful.  We
based our judgments on the phase III report's (1) physical damage
summary, (2) cumulative summary of intelligence data on functional
damage, and (3) restrike recommendation, if provided. 

We rated the campaign against a target as fully successful (FS) if
phase III report stated following: 

  The target was destroyed or so damaged as to be unusable or
     nonfunctional, and the diminished condition of the target was
     because of the physical damage of air strikes or indirectly
     attributable to the air campaign, such as the threat of strikes. 

  The restrike recommendation was "no."\18

We rated the campaign against the target as not fully successful
(NFS) if the phase III report stated the following: 

  The target was not destroyed or so damaged as to be unusable or

  The facility had been struck and suffered only partial (or no)
     damage or degradation and remained on the target list. 

  Insufficient data were available to confirm that the objective had
     been met, and the target therefore remained on the list.\19

  The restrike recommendation was "yes."\20

Table I.5 illustrates examples of the phase III BDA information
reported by DIA and our FS or NFS determinations. 

                                    Table I.5
                     Examples of Phase III BDA and Our FS or
                                 NFS Assessments

Target                                                              Our
category      Target type         BDA summary                       assessment
------------  ------------------  --------------------------------  ------------
C\3           Air defense radar   50 percent degraded;              FS
                                  nonoperational; restrike: no

              Air defense radar   Radar and command capability      NFS
                                  remain; restrike: yes

\ELE          Power plant         Turbines not operating;           FS
                                  restrike: no

              Power plant         Installation 70 percent           NFS
                                  operational; switchyard must be

LOC           Highway bridge      Direct hit, bridge                FS
                                  nonoperational; traffic rerouted

              Highway bridge      Bridge still operable; no damage  NFS

NBC           Munitions storage   All bunkers out of operation;     FS
                                  restrike: no

              Chemical warfare    Laboratory intact; restrike: yes  NFS
              production and

OCA           Airfield            Limited operations possible;      FS
                                  restrike: no--unless flight
                                  operations resume

              Airfield            50 percent hardened aircraft      NFS
                                  shelters intact; airfield
                                  operational; restrike: yes

\17 DIA generated 986 phase III reports covering 432 separate
targets.  We used the final phase III report when more than one
report was produced on a target. 

\18 Additional strikes on a target were recommended by DIA to CENTCOM
when the results of their BDA indicated that military activity or
capability remained at the target site.  Restrikes may or may not
have occurred for a number of reasons (for example, changing or
conflicting priorities in-theater, constraints imposed by the
weather, or limited dissemination of BDA results). 

\19 It was standard procedure during the air campaign to retain
targets on the daily air tasking order and the Master Target List
(MTL) and retask aircraft to the target if BDA was absent or

\20 By categorizing a target as NFS, we are not implying that the
strikes (or other actions of the air campaign) did not have an
adverse impact on the enemy at that location.  In many instances,
strikes resulted in the partial destruction of the targets and may
have affected the tactics and level of enemy activity.  An NFS rating
implies only that the complete destruction of the target or the
elimination of its function had not been achieved (or could not be
confirmed) and additional strikes were necessary. 

----------------------------------------------------- Appendix I:2.3.2

Although DIA's phase III reports were by far the most comprehensive
compilation of BDA for strategic, fixed targets produced during or
after the campaign, there were several limitations to these data. 
These include

  Not all strategic targets were assessed.  DIA issued phase III
     reports on
     432 BE-numbered strategic targets, which was a total lower than
     either the final number of strategic targets identified by
     CENTCOM during the war or the number of BE-numbered targets in
     the Missions database, and which was a set of targets that were
     not necessarily representative of the universe of strategic

  No effort was made after the campaign to update or verify the vast
     majority of the reports.  The accuracy of some analyses without
     ground verification is very difficult to determine. 

  Imagery limitations can hinder analysis.  Imagery collection may at
     times have preceded strikes because combat missions were delayed
     or postponed.  Imagery may not have been taken from the optimal
     side of a target or at an inappropriate angle for assessment

  According to DIA, the reliability of assessments grew over the
     course of the war with the increased experience of the analysts. 
     Thus, the assessments later in the conflict may be more reliable
     than those made earlier because analysts learned more about the
     capabilities of the aircraft and munitions through the course of
     the war. 

         OTHER DATA
----------------------------------------------------- Appendix I:2.3.3

We obtained aircraft and munitions cost data from Air Force and Navy
documents and costs as identified in DOD's periodic Selected
Acquisition Reports to the Congress. 

------------------------------------------------------- Appendix I:2.4

To analyze the use of aircraft and munitions in achieving air
campaign objectives, we used the Missions database to determine
weight-of-effort and type-of-effort measures at two levels.  First,
we calculated WOE and TOE at the broad level of the target category
for each of the 12 strategic target categories shown in table I.1. 
Second, we calculated WOE and TOE for each aircraft and TLAM across
the 12 categories. 

We used phase III reports on 432 fixed strategic targets to determine
the extent to which the functional capabilities of the target had
been eliminated.  To correlate outcomes on targets with the input to
them, we matched phase III data with data in the Missions database. 
For 357 strategic targets (where both BDA and WOE/TOE data existed),
we sought to assess the relationship between the WOE and TOE data
representing campaign inputs with phase III BDA representing campaign
outcomes at the target level.\21

We conducted our work between July 1992 and December 1995 in
accordance with generally accepted government auditing standards. 

\21 This methodology was discussed with DIA analysts who were
familiar with both the Missions database and the phase III reports. 
They identified no reason why this methodology would not result in
valid comparisons of inputs and outcomes.  In addition, they believed
that the use of WOE and TOE variables would alleviate data problems
previously encountered by analysts conducting strike BDAs. 

--------------------------------------------------------- Appendix I:3

This analysis of campaign, aircraft, and munitions use and
effectiveness benefited from our use of the most comprehensive strike
and BDA data produced from the Persian Gulf War; a previously untried
methodology to match inputs and outputs on targets; additional
qualitative and quantitative data obtained from Desert Storm veterans
and after-action reports to corroborate information in the primary
databases; and the results of other Desert Storm analyses, such as
the Gulf War Air Power Survey. 

This study is the first to match available Desert Storm strike and
BDA data by target and to attempt to assess the effectiveness of the
multiple weapon systems across target categories.  Despite the data
limitations discussed below, our methodology provided systematic
information on how weapon systems were employed, what level and types
of weapons were required to achieve success, and what was the
relative cost-effectiveness of multiple platforms.  The reliability
and validity of these findings are strengthened by our use of
interviews, after-action reports, and other Desert Storm analyses to
better understand platform performance variables and place the
results of our effectiveness analyses in the appropriate context. 

Our analyses of campaign inputs (from the Missions database) and
outcomes (from the phase III reports) against ground targets have
limitations of both scope and reliability imposed by constraints in
the primary Desert Storm databases.  Systematically correlating
munition inputs against targets to outcomes was made highly
problematic by the fact that the phase III BDA reports did not
provide a comprehensive compilation of BDA for all strategic targets
and could not differentiate the effects of one system from another on
the same target.\22

We sought to work around data limitations through a qualitative
analysis of systems, based on diverse sources.  Claims made for
system performance were assessed in light of the most rigorous
evaluation that could be made with the available data.  We have
explicitly noted data insufficiencies and uncertainties.  Overall,
data gaps and inconsistencies made an across-the-board
cost-effectiveness evaluation difficult.  However, there were
sufficient data either to assess all the major claims made by DOD for
the performance of the major systems studied or to indicate where the
data are lacking to support certain claims. 

\22 Such assessments, system by system, were not the goal of these
reports.  Since targets were generally assessed only episodically
and, in most cases, after being hit by numerous diverse aircraft and
munitions over a period of time, it was impossible to know which
munition from which aircraft had caused what amount of damage.