Who is counting Yemen's casualties?

Between 29 and 45 people were reported killed by an apparent air attack at the al Mazraq camp – some were said to have been burned beyond recognition. Depressingly, the victims also included children.

Although the attack came shortly after Saudi Arabia had launched an aerial bombardment of Yemen, Yemen’s foreign minister, speaking from Riyadh, blamed artillery fired by the Houthi militia which stormed the country’s capital Sanaa late last year. A Saudi spokesman meanwhile said that rebels had been firing from a residential area in response to a question about the bombing.

Before answers could be found about what happened at al Mazraq, the violence quickly moved on elsewhere. Dozens of civilians were were reported killed by an attack on a dairy factory in Hodeida the following day. Again, the details are still unclear.

The potential for more such atrocities to be carried out with impunity is increasing as the Yemeni state collapses and regional powers pile in.

Saudi Arabia, convinced that the Houthis are backed by Iran, is leading an air campaign with a coalition of 10 Gulf neighbours and North African allies. Warships thought to belong to Egypt have reportedly shelled Yemen from the sea. US officials have pledged that drone strikes will continue. And al Qaeda is taking advantage of the chaos, springing hundreds of criminals from a prison in the east of the country.

It has never been more important, nor more challenging, to ensure proper mechanisms for the recording of casualties are in place.

The UN’s Office for Coordination of Humanitarian Affairs (Ocha) is doing the best it can. It is collecting, aggregating and publishing reports of casualties from any sources it can find. Its situation reports are providing vital insights into the bloodshed.

But it remains extremely difficult to record casualties from armed conflict accurately, even for the UN, which actually has its own staff in Yemen.

As Erich Ogoso, public information officer for Ocha Yemen, told the Bureau: “The biggest challenge is verifying information coming out of Yemen.”

The challenges stem, in part, from the lack of independent on-the-ground reporting to corroborate the proliferation of videos and tweets.

Perhaps paradoxically in the digital age, the fog of war has never been denser.

These are also the challenges the Bureau has faced over the past four years in recording deaths from the US’s ongoing war on terror in Pakistan, Somalia and Yemen itself.

We have tallied more than 400 US drone strikes in Pakistan since they began in 2004. And we have marked each reported drone, air and cruise missile strike in Somalia since 2007, and in Yemen since 2002.

The Bureau’s casualty data includes information gleaned from court documents and leaked government databases. But it is built on open source information harvested from thousands of media and NGO reports. Our work relies on the brave work of journalists and researchers risking their liberty and lives.

In conflict situations, it’s often simply not possible to get on-the-ground information. When US and Yemeni forces slowly forced al Qaeda out of its enclave in southern Yemen in 2012 for example, the ferocity of the fighting meant few details of specific drone attacks escaped the maelstrom.

And even when investigators have the opportunity to go back to the site of a strike after it has occurred, it can be difficult to pinpoint responsibility for it.

After a catastrophic air strike in the southern al Bayda governorate in September 2012, it took Human Rights Watch months of work to identify 12 civilians killed.

Blood money was paid in secret to the victims’ relatives and a US official anonymously admitted Washington was responsible. But he wouldn’t say if a drone or a jet carried out the strike – an important factor in determining who should be held to account. The airstrikes with fast jets in Yemen were under the control of the US military, whereas drone strikes hit Yemen at the behest of the CIA and the military command.

For all the frustrations, casualty recording is essential when confronted by official obfuscations or outright denials. Marking the deaths of people in conflict gives families and communities the chance to seek justice.

But gathering empirical evidence of deaths in conflict is also essential for scrutinising policy and challenging the government line when it deviates from the evidence.

When now CIA director John Brennan claimed in 2011 that the CIA had not killed a single civilian in Pakistan for almost a year, the Bureau’s credible record of casualties enabled us to investigate this claim and demonstrate it was untrue.

A common code of practice in the recording of casualties by civil society organisations has evolved since Iraq Body Count first started in 2003, and the Every Casualty’s International Practitioners network can offer practical help to organisations looking to do this work.

The UN’s efforts to record casualties as part of its overall reporting on Yemen are important and should be supported. But to produce an authoritative record, they need to be complemented with more information, and with cross-referencing and fact-checking.

One contribution has recently been launched by the Global Voices network, which uses Checkdesk, an open-source tool that enables journalists to sort, analyse and verify the torrents of material produced by traditional and citizen journalists.

Scores of people have been killed already in Yemen and it appears many more may perish before stability is returned. These lives need to be marked, and the manner of their deaths recorded. Only in this way, can those responsible be held to account.

The Bureau’s Naming the Dead project is supported by Joseph Rowntree Charitable Trust.

Image of demonstrations against Houthis in Yemen in January 2015 by Mohammed Hamoud/Anadolu Agency/Getty Images)