“They Bombed Everything that Moved”: Aerial military attacks on civilians and humanitarians in Sudan, 1999 – 2011 | Part III: A guide to the data
III. A guide to the data:
The database in this report contains all identifiable, publicly available data, including for incidents referred to in the Introduction to this report and in the preceding schematic history. The data are concentrated in the years 1999 to 2011, although some data from earlier years are included. Sustained and organized collection of data only began in 1999, but it is likely that previous years would show similarly extensive aerial assaults if comparably complete data were available.
It is likely that most attacks, both in South Sudan and Darfur, were not reported. Often there was insufficient or no reporting presence in much of the South, and the same is true of Darfur. Thus the more than 1,400 incidents reflected in the data do not constitute a complete record of the extent of Khartoum’s war crimes. Indeed, in analyzing Khartoum’s campaign of aerial destruction, it is essential to understand that the majority of attacks, both in the South and Darfur, were almost certainly not reported. A reporting presence was too often simply non-existent in much of the South. And a common attitude on the part of civilians was grimly noted in one human rights report previously cited: “There are reports of frequent bombing in southern Blue Nile, particularly around Geissan and Demsaid, but local people are so accustomed to it that they see no point in keeping records.” Thus the more than 1,400 incidents reflected in the data spreadsheet are by no means a full reckoning of the extent of Khartoum’s war crimes. Even so, I have been parsimonious in using the data available, and have always chosen to exclude rather than include reports that seem doubtful, redundant, or contradictory. This has resulted in the exclusion of more than 200 reported incidents originally included in the data spreadsheet.
There have been many challenges in assembling these data: eliminating redundancy of reporting; identifying as specifically as possible the locations of attacks; establishing a threshold for “confirmed” attacks; ascertaining that given attacks are directed against civilian or humanitarian targets. These challenges are sometimes rather different for South Sudan (as well as South Kordofan, including the Nuba Mountains, and southern Blue Nile) and Darfur; this warrants brief discussion.
October 15, 2002 marks a partial terminus ad quem for Khartoum’s aerial assault on the South. This was the date on which a “cessation of hostilities agreement” was signed by the regime and the Sudan People’s Liberation Movement. And though violated on a number of occasions— sometimes in highly consequential fashion—reported attacks dropped off dramatically. By the first half of 2003 the scene of military assaults on civilian and humanitarian targets had moved to Darfur, and there was little in the way of international presence that could record the thousands of attacks on non-Arab or African villages that marked the terminus a quo for Khartoum’s genocidal counter-insurgency efforts. We are not without reports from this period, but without the presence of international humanitarian organizations, and with only a modest human rights reporting presence, the vast majority of aerial attacks were not reported. At the same time, many reports by survivors continue to accumulate, and this part of the data set will continue to expand in the years to come.
Establishing the criteria for a confirmed report of a bombing or strafing attack on a civilian or humanitarian target has been a great challenge, and I have occasionally been guided by context as much as by the specific details of a reported incident. But for the overwhelming number of incidents, the following serve as the basis for confirmation:
 Confirmation by a UN or nongovernmental humanitarian organization;
 Confirmation by a credible human rights organization, including Sudanese national organizations;
 Confirmation by Sudanese church sources;
 Eyewitness accounts by journalists;
 Reports by civilians, if there is more than one witness; for Darfur, this includes reports from Radio Dabanga, which often cites a specific witness, but one who has been determined by Radio Dabanga to be representative of the targeted community;
 Forensic investigation confirming a bombing or strafing attack (for example, in South Sudan the Civilian Protection Monitoring Team that began work in late 2002 reported effectively for over half a year);
 Confirmation by UNAMID or its predecessor AMIS in Darfur, or Operation Lifeline Sudan officials in South Sudan;
 Attacks reported by the SPLM: these were so consistently confirmed when investigated that it seems unwarranted to exclude the very few cases where the Movement is the only source of information.
I have included only a few reports by the Darfuri rebel groups as confirming evidence, although they have been consistently used when the primary report comes from a humanitarian representative; rebel spokesmen have on occasion exaggerated and misled, and they lack a track record comparable to that of the SPLM. But if an attack is confirmed by sources of the sort indicated above, then rebel military reports seem appropriate in augmenting an account of the attack (precise location and time, number of casualties—though this with skepticism—and kinds of aircraft involved).
Some reported attacks appear to have a military purpose, but are typically mounted with complete disregard for civilian casualties. They are not simply indiscriminate but deliberately so. Yet again, this is conspicuous evidence of the contempt with which the Khartoum regime regards the lives of African tribal groups, whether in the South or Darfur. I have consistently included reports that establish the use of Antonov “bombers” near civilian population concentrations.
Evidence used to ascertain whether an aerial attack was directed against civilians as opposed to military targets includes:
 Use of Antonov aircraft (“bombers”) near areas with significant civilian populations. The Antonov is inherently very inaccurate and thus indiscriminate in all its bombing attacks: flying at roughly 5,000 meters to avoid ground fire, lacking a bomb guidance system or a bomb rack or bay, Antonovs used in populous civilian areas are ipso facto attacks on civilians.
 Use of helicopter gunships—which fly low, and fire with considerable accuracy—can be judged to be attacks against civilians if they are the primary victims of an assault.
 Eyewitness accounts of the purpose of particular missions by Antonovs, helicopter gunships, or military jet aircraft.
 Photographic or forensic evidence indicating civilians or humanitarians were the targets of an aerial attack.
Column for Casualty Figures:
Inevitably, for most aerial attacks the number of casualties is simply registered as “unknown”; this is always the case when there are no specific casualty figures. But this doesn’t mean that we know nothing about the civilian casualties after a given aerial attack; indeed, the observations column (H) very often refers to “civilian casualties”; sometimes the references are to heavy, but unquantified, mortality and injury, or extensive damage to hospitals and health clinics, or to the interruption of critical humanitarian services. Observations about the destruction of schools, churches, cattle herds, indeed entire villages are frequent. Similarly, the forced flight of hundreds of thousands of civilians by virtue of aerial attacks has caused massive morality, especially among the young, the old, and the infirm. But because the casualties are not quantified, they do not figure in calculations for Column G.
Even so, the accompanying observational notes often make clear just how many casualties resulted from attacks, and the totals can be shocking—sometimes there are literally hundreds of civilian casualties. Sometimes all casualties in a confirmed report are of civilians “killed,” and are so designated; sometimes all casualties in a confirmed report are “injuries,” and are so designated. A figure for “total casualties” indicates that this represents a combination of killed and injured; it should not suggest that this is a definitive figure, only the one offered in a credible contemporaneous report (many would later succumb to their injuries). In a great many cases, there would have been deaths subsequent to the only report that was made or preserved. A “+” sign indicates that the number is actually greater, or that there is further information on casualties in the adjoining explanatory column.
What else can be said about the total number of casualties derives from a simple statistical calculation. Of the 1,414 incidents confirmed and represented in the data spreadsheet, 307 incidents have a known figure for casualties, and 1,107 do not. The average casualty total for the 307 incidents for which we have figures is approximately 14 (4,261 total casualties).
Multiplying this average casualty figure by the number of incidents for which we do not have a figure (1,107) yields a total of approximately 15,500 casualties. The two figures together suggest that on the basis of present confirmed reports for South Sudan, the Nuba, Southern Blue Nile, Darfur, and other locations, 20,000 have been killed and injured during Khartoum’s aerial military attacks; this is the best statistical estimate of total casualties without more data.
But there are other statistical implications to this estimate, given the fact that significant understatement of the actual number of attacks conducted is extremely likely. If the presently established number of reported and confirmed instances of aerial attacks understates by a factor of two actual attacks, then the estimated total rises to 40,000 killed and injured. And if the reported and confirmed incidents understate by an order of magnitude—a distinct possibility, if we consider also the years before 1999—then the estimated figure for total casualties rises to 200,000 killed and injured.
Every effort has been made to eliminate duplicate reporting; this is, however, an extremely difficult—and at times impossible—task. Khartoum has frequently bombed the same site twice or more in the same day; it has attacked the same site on consecutive days; it has attacked in the close vicinity of immediately preceding attacks. And very often Khartoum has attacked a series of villages and towns on the same bombing run, though not all are accounted for in the reports we have. My strategy in eliminating duplication has been to look for obviously identical features in reports—dates, locations, number of bombs, number of casualties—and then to exclude those that seem to have excessive statistical similarity, even if authoritatively reported by another source. For this reason, during a given period of frequent bombings I rely as much as possible on a single highly authoritative source (e.g., John Ashworth of the Sudan Ecumenical Council), reasoning that a single source reporting seriatim is less likely to produce redundant reports. But these are imperfect solutions.
While quantitative totals and representation are certainly important, the observations that accompany the data in Part IV are just as important. And in the many cases where only “unknown” appears in the casualties column, the accompanying notes are critical to any real understanding of what these aerial attacks represent: they often point to large, but indefinite losses of life and serious injuries; the destruction of villages; the killing of cattle and livestock; large-scale civilians displacement; the targeting of water sources; and the forcing of humanitarian withdrawal. That so many aerial attacks in Darfur, including very recent ones, have only an “unknown” number of casualties is all too revealing of the international level of concern.
Geographic identification has proved enormously challenging, despite working with substantial cartographic assets, and extensive communication with experts on both Darfur and South Sudan, as well as people from these regions. Herewith some examples of what has made identification difficult when geographic location is not tightly specified:
 Repetition of names: in both Darfur and South Sudan, there is a tremendous repetition of names, sometimes in close geographic proximity, sometimes widely separated. In Darfur, for example, there are no fewer than 16 iterations of Hashab(a). And sometimes there are no names (a location is indicated on a map as “unidentified,” or an identifying name is described as “not applicable”). Sometimes the name of a location occurs in a neighbouring country (Uganda, Democratic Republic of Congo, Chad).
 Sometimes the same location has more than one spelling or even more than one name. For example, a location may have one Arabic name and one Nilotic name in South Sudan (especially Abyei); sometimes cartographers have chosen different names for the same location, evidently
on an arbitrary basis. In Darfur the same problem exists: Kuma in North Darfur is also referred to as Khurayt; also in North Darfur, Basao is also referred to as Berediq. Arabic names are transliterated in a wide variety of ways by different sources.
 Sometimes coordinates for particular locations vary significantly from one map to another, although the most recent digital maps are largely in agreement on nearly all locations. Very occasionally, especially in South Sudan, certain names cannot be located more specifically than by state, despite many hours of research attempting to confirm precise location. To a lesser degree the problem also exists for Darfur.
 Sometimes, especially in South Sudan, all available evidence puts a given site of attack almost precisely on a county border. If the site itself does not appear on a map, and a location can be inferred only by means of using references to nearby villages or towns, a best estimate has been made as to which side of the county border to designate in the spreadsheet.
The goal for South Sudan has been to identify a town or village location by county and state; since the geographical divisions in South Sudan have changed, this has created reference problems. Some reports speak of “Western Upper Nile”; more current references are to “Unity State” (the present geographic designation). In general, I have decided to use the nomenclature of current cartographic resources, retaining the original only very occasionally for clarity’s sake. One village has been located only by state; one is indicated by “payam” (a sub-county designation); one by only the general designation “Equatoria.”
The goal for Darfur has been to identify in which of the three states an attack has occurred, and then to identify the specific “locality” (roughly the equivalent of “county”) for the village or town. Only a couple of locations do not include “locality” information. (Rural Council boundaries have not been considered.)
Different reports have different degrees of specificity. Many simply list attacks by date, indicating “Antonov bombing.” On the other hand, some—such as the hideous example of Bieh—have detailed eyewitness accounts by multiple UN aid workers. Some attacks have been recorded on videotape (e.g., the bombing of Yei market, and the Comboni school in Kauda, Nuba Mountains). There are quite literally thousands of photographs of the aftermath of aerial attacks on civilian and humanitarian targets….