Why should I use THOR? What does it provide?

Never before has there been a means of swiftly referencing where and when air strikes have been conducted across the globe. Vague mentions of strategic bombing such as “in Germany” have never allowed a detailed examination of effects generated or joint fires created when combined with land and sea forces. Conversely, anecdotal stories “from the cockpit” usually fail to tell the larger operational or strategic story of how that one plane or one squadron fit into the larger campaign narrative. For the first time, THOR provides a quick and reliable way to view air strike activity from 1915 to 1975, and provides the capability to generate historical summaries based on user-selected criteria to answer basic questions of who (callsign, service, country) how many (strikes, weapons, etc) what kind (aircraft, weapons, etc), when, and where. Additionally, by using Google Earth, this information can be presented geographically, and it allows for time series presentations (movies), as well as strike density and weapon intensity comparisons, to be made. Other sources of information, such as population density, etc, can be graphed on Google Earth to compare with air strike information.

Doesn’t this capability already exist? Why did you reinvent the wheel?

If it exists, we haven’t found it. We didn’t invent THOR because we thought it would be neat, we built it to answer recurring questions about how airpower has been used in war. If you know of a similar or better capability, please contact us at thor@us.af.mil so we can compare ideas.

What does your database currently contain? How current is it?

THOR contains data from 1915 to 1975, covering World Wars I and II and the Vietnam War. It is as complete as we’ve been able to make it. We suspect there are some gaps in WWI – particularly in the Italian records, and are compiling the French bombing record in WW1. Vietnam is missing data prior to September 1965, and also the month of November 1967 is missing (due to water damage to its magnetic tape archive). To the best of our ability, the rest of the information is as complete as records document. The Korean war records are being hand transcribed from paper mission reports, and will be posted as soon as that action is complete.

Why bother including past wars in THOR? The current war is completely different than ones we’ve fought before.

Postwar assessments do not necessarily contribute to the current war but may still prove important to employing effects-based operations (EBO) and predictive battlespace awareness (PBA) in future wars. Correctly anticipating enemy responses and the ways airpower can best be used to exploit them might be based on modeling. One way to calibrate predictive models would be to understand how adversaries responded to previous airpower applications. (See “Slow Airpower Assessment: A Cause for Concern?” Lt Col Paul D. Berg, USAF. Air and Space Power Journal, Fall 2004, pp. 75-83) Additionally, some doctrine and strategies may be timeless. By providing the data, this idea can be more fully explored.

Why don’t you include Weapon System Video(WSV) or post strike(BDA) imagery?

We look to add new features as our capability grows. We are excited about WSV and BDA imagery, but we would ask for help to host that information in a way we can reliably link to it--including it inside the THOR database would seriously bog down performance, as we are currently sorting through 4 million+ records to process a query.

Why do you report weapons dropped by weight? is that really the best metric?

Post-combat assessments from World Wars I, II, and GWAPS all compared weapons dropped by weight. To provide an “apples-to-apples” comparison, we have included that functionality in THOR. Additionally, a “by weight” analysis allows a comparison of the intensity of attacks by comparing gun, rocket, missile, and bomb attacks side-by-side without skewing the results with the large numbers of ammunition expended (when compared to numbers of bombs). We make no claim this is the best metric. If there is a comparison you need on a regular basis, please contact us at thor@us.af.mil so we can see if it is feasible to include in THOR.

Why do some entries have more data than others?

We have done our best to faithfully transcribe the information provided from the source documents. To gather and record all of the pertinent data has presented a recordkeeping challenge since World War I. The only editorial input we provide is to clean up Lat/Long location (DMS, DD MM.MMM, MGRS, etc), altitude(FL230, 12k ft, 3000m), and weapon type data (GBU-12 vice GBU 12, etc) to be properly plotted and tabulated. We have also posted the unaltered source data on the site for comparison.

Why are you using Google Earth? Why not Falconview (PFPS)?

The formatting of data to be consistent with Google Earth does not constitute a Federal Gevernment endorsement or support of Google Earth. Google Earth was chosen for several reasons, the primary one is that the community of users is constantly growing and innovating new ways of using Google Earth, and presenting new ways of displaying data on it. Two, it is available free from Google, which puts powerful GIS tools in the hands of folks without any outlay of funds. It allows easy sharing of information between multiple users (you can e-mail your analysis via .kmz file to another user), and can be effortlessly shared across the military and academic communities. PFPS did not seem to offer these advantages. However, we have included a spreadsheet output mode for anyone wanting to export the data into another database or software package.

Why do other sources report different numbers of weapons employed or sorties flown than THOR?

One of the functions THOR performs is to check for duplicate entries - if the exact same plane with the exact same callsign drops the exact same weapon at the exact same time(TOT, ATO Day and Date) on the exact same coordinates, we flag it as being a duplicate and remove it from the database. We researched this during the THOR development cycle to make sure we were catching actual duplicates, and we found that if the duplicates were, in fact, true, the plane would be dropping more than its maximum-capable bomb load. All of the numbers in THOR are determined by the records that are in the database. It could be that THOR does not have some of the records in question due to records being lost over time. If you do see a discrepancy, please drop us a note us at thor@us.af.mil so we can look into the matter.

What percentage of the total records do you have?

An interesting question that is nigh impossible to answer accurately. In some cases we record more sorties flown or bombs dropped than the approved and recognized sources. For example, the USAF Statistical Digest for WW II has long been considered one of the most trusted sources for operational air data on WW II. However, reading the fine print, the Digest notes that 8th AF bombing data in Europe is not included prior to 1 October 1943! That means that the Digest excludes the first 104 missions flown over 14 months by the 8th AF, which would include the first Schweinfurt raid, amongst others. Frustratingly, no explanation is given for this editorial decision. THOR contains data from the first raid by 8th AF (Rouen, 17 Aug 1942) through the end of the war, so the numbers in THOR will not necessarily match up with those that are based on the Statistical Digest.

Are there mistakes or inaccurate data in THOR?

Yes. Data collected during combat conditions is never as pristine or complete as desired. There are four main sources for inaccuracies that exist in the database:
1. Errors in the source material – the pilot misreported the facts: misspellings, location, time, etc
2. Errors in the encoding process – while completing the punchcard or typing the data, the wrong value is selected
3. Errors in the OCR process – the computer misread the scanned text and substituted an incorrect value (transposing an “8” in place of a “5” for example)
4. Errors by the database author – introducing errors by incorrectly calculating values or making mistakes while entering data by hand
In the course of building THOR, all forms of these errors have been detected and corrected when identified. For example, the longitude of the Nagasaki atomic bomb strike was recorded on its respective punchcard 550 miles off from the target (and occurring on the Chinese coast) because the punchcard value was transposed as a “0” and not a “9” (one value up on the punchcard). That error was identified and fixed. That doesn’t mean that there aren’t additional errors that still lurk in the data. If you should find an error, please use the contact forms to alert us to the problem. If you have source material that will allow us to document and correct the error, please provide that as well. Our goal is to get this right.

Why are you missing data from (Fill in the blank)?

The short answer is that we have what we have. We hope there are additional records out there that we can add to the database. If you have access to additional records, and are authorized to share them with us, we welcome their addition to THOR. Please use the contact forms to let us know what we can add to the datasets.

When I choose the altitude display in Google Earth, does that show the bomb release point?

No. The bomb release point depends on numerous factors – altitude, speed, ballistic characteristics of the weapon, guidance system, winds, etc. We have included the bomb release altitude display option, (which shows the altitude the bomb was released, but displays that over the impact point) to help untangle the confusion caused by multiple attacks on the same or nearby locations, when all of the strike indicators stack up on top of each other on the ground. Additionally, we have included an Excel output module for anyone wanting to export the data into another database.

Why is the SOF underreported? Where are all of the AC-130 strikes? Where are the WWII fighter strikes in Europe?

SOF MISREPS have traditionally failed to consistently include coordinates of where they have employed their weapons. The best we have usually uses, for example “IVO (in vicinity of) Baghdad.” Since Baghdad easily covers greater than 81 square miles, inaccurately plotting the AC-130 data could inadvertently lead researchers to draw false conclusions. The same is true for WWII fighter information. When we have coordinates, we do plot those engagements, as seen in WWII aircraft data in the Pacific.

Do you have bomb impact coordinates for World War I? How did you plot those strikes?

While we lacked latitude and longitude data from World War I, the original 1918 Raid Reports and Daily Operations summaries made a point of describing the location their bombs impacted in some detail. Combined with maps and the few aerial photographs we have located from World War I, we feel comfortable that the locations used are consistent with the accuracy of target plotting of later wars.

Why do you include World War I at all? What possible relevance could it have?

World War One is included for two main reasons:
1. It allows us to provide (eventually) a full accounting of air power employment back to the dawn of air combat tactics and doctrine.
2. There may be lessons still unlearned hiding in the data for either current or future conflicts. In short, we can, so we will.

What is the target location accuracy?

The target location accuracy varies through a number of factors. One is the level of latitude and longitude resolution. WWII US Strategic Bombing Survey data is reported in Degree Degree Minute Minute resolution (DDMM), which equates to 1 nautical mile (1nm) accuracy. While that sounds quite inaccurate to an audience raised on GPS-guided munitions having 1 meter or less accuracy, the size of the bomb load must also be factored in. For WWII, the best accuracy achieved was approximately 30% of bombs within a 1,000 ft radius of the target, and 97% within a mile of the target. A bad bomb run using H2X (radar) to guide to a completely cloud-obscured target could have as few as 50% of its bombs within FIVE miles of the target. With these factors taken into account, the 1 nm accuracy seems appropriate for WWII. Korea and WWI share a similar accuracy to WWII. For Vietnam, the accuracy of target plotting is closer to 100 meters for the best coordinates, but can vary to as many as 1,000m for some of the less well documented strikes.

Where are all of the USMC and Army strikes? Why is THOR Air Force-centric?

The goal of THOR is to include all airpower strikes, regardless of service or Allied country of origin. We would be glad to incorporate USMC and USA strikes into the database, but we have been unable to locate an authoritative source. When we have data, we have included it. Please contact us at thor@us.af.mil so we can see if it is feasible to include additional sources in THOR.

What about other facets of airpower? What about Airdrops?

The focus of THOR has been on strike data – not that we claim strike data alone is the full measure of airpower; the strike records were historically the most complete ones to include for analysis. We have been able to recover 3000+ records of strafing data from WW2 (approximately 30,000 aircraft and 12.9M bullets, 97K rockets), and continue to look for more data to add to the database. We have not yet explored airdrops or other exercises of airpower. Please contact us at thor@us.af.milso we can see if it is feasible to include additional sources in THOR.

What about smaller raids and operations - OSW, ONW, Panama, Grenada, Eldorado Canyon, etc?

We would be glad to include them in THOR, although they are on the lower end of our priority list for the time being. We would need to gather all of the pertinent data, and identify authoritative sources for each of these operations. Please contact us at thor@us.af.mil so we can see if it is feasible to include additional sources in THOR.

Who have you worked with on this?

THOR would not have succeeded if it weren’t for the efforts of many. The THOR team is deeply indebted to the Air Force Historical Research Agency (AFHRA), AF/A5XS “Skunks”, Air Force Lessons Learned (A9L, specifically Mr. John Georgiou, who kept a lonely vigil collecting strike data for years), the consummate professionals at the National Archives, the AF/AO-AOA NCO’s for being the beta testers on this project for over a year, the RAND Corporation, Air Force Historical Support Office, 9th AF History Office, ACC History Office, ACC/A2, AF/AFIAA, the OCR wizards at Docucon, and most importantly, the warfighters in the forward CAOCs, who are writing this first draft of history. There are many more that I am inadvertently leaving out. To those not mentioned, I apologize and thank you for your help.

How do I cite THOR in my project?

Please use the following citation:
Lt Col Robertson, J. A., Burr, R., and Barth, B. (2013) USAF THOR Database; www.afri.au.af.mil/thor retrieved on (insert day, month year). Data compiled from multiple sources.

What is the security classification of the information in THOR?

All data available on the NIPRnet is unclassified. Please let us know if you have any concerns.