It’s strange how things sometimes come together – or collide and crash. Two weeks ago I wrote about satellite imagery and ‘remote violence’, and over the summer I discussed several projects that mapped Israel’s military assault on Gaza and its people, including the Gaza Crisis Atlas produced by the UN’s Office for the Coordination of Humanitarian Affairs (which makes extensive use of hi-res satellite imagery).
The links between those two posts are obvious enough, but today I have something altogether different in my sights. ‘Judge Dan’ (Dan Smith), who blogs for Israellycool, has used the Gaza Crisis Atlas to construct a series of maps to geo-wash Israel’s military campaign.
Working from the data in the Atlas, Smith produces four maps to disaggregate the severity of destruction (the four levels, increasing from left to right, are based on OCHA’s own assessment):
His conclusions from this exercise are the following:
The attacks are in no way “random” or “indiscriminate”. One can clearly see the spatial distribution of the damage in several aspects. We find 8,952 of the 12,433 total points (72%) are within a 3 KM buffer abutting the border with Israel. The main objective of Operation Protective Edge was to find and destroy dozens of terror tunnels dug from Gaza into Israel.
That the most intensive damage was caused to the area where the tunnels naturally originated is thus perfectly understandable. Furthermore, of the 4,441 destroyed structures, 3,481 of them (78%) are within the 3 KM buffer, as are 2,531 of 3,303 (77%) of the lowest intensity damage (simple craters), which are mostly strikes on rocket launchers and tunnels.
Most of the attacks are grouped around certain neighborhoods or villages, such as Shuja’iyya, Johur ad-Dik, Sureij, and Khuza’a. These were probably the result of the ground operations that took place in dense urban areas also within the 3 KM buffer that housed multiple tunnel entrances and shafts, as well as launch sites for mortars and rockets.
Smith then takes the Israeli military’s map of ‘terrorist infrastructure’ in Shuja’iya and overlays this on what he calls ‘OCHA’s damage points': ‘the correlation is uncanny.’
Smith’s next manoeuvre is to sweep aside OCHA’s focus ‘on the civilian aspect’ because it ‘misses the big picture’ (really – or perhaps Israelly): ‘the overall intensity of the strikes’. So he constructs a kernel density map or a ‘damage intensity heat map':
It now becomes very clear that most of the damage was caused to 5 locations right on the border with Israel. The rest of the Gaza Strip was, for the most part, undamaged. The main population areas of Gaza city, Jabaliya, Khan Yunes, Rafah and Deir el-Balah were disproportionately undamaged (sic).
If we do a rough estimate of the damage area, it is once again clear the vast majority of the Gaza Strip was unscathed. With a fairly generous estimation that a damage point has a 25 meter radius – the footprint of a house, or the blast radius of a bomb – the total damage area of the 12,433 impacts was in the order of 15 KM2. The land area of the Gaza strip is 360 Km2. In other words, less than 5% of the land was affected.
There’s a follow-up post on ‘damage clusters’ here, but in this commentary I’ll focus on Smith’s ‘big picture’. I take the basic points to be these:
(1) Smith’s approach makes an appeal to the supposed objectivity and even facticity of the map (and, by extension, the satellite image), but there is a substantial body of scholarship that goes back 25 years and underscores the multiple ways in which mapping is an exercise in the production of power. For a depressingly relevant example of the ways in which maps can speak power to truth, taken from Israel’s attack on Gaza in November 2012, see my discussion here (scroll down to the maps).
(2) Appealing to the map and its manipulations as the single source of authority is designed to disavow the testimony of witnesses on the ground: precisely the point sharpened by Andrew Herscher in his timely critique of ‘Surveillant witnessing’ (see my discussion here) and a far cry from the incorporation of photographs in (say) the Gaza War Map.
(3) Smith’s methodology reduces Gaza to an object space of structures and buildings, craters and points; he constructs a kernel density map (more on this in a moment) but provides no population density map that would at least gesture towards the people killed, wounded and traumatised by the Israeli offensive and who are wholly absent from his account. Here, by contrast, is a map I posted previously showing deaths in Gaza to 6 August:
The Gaza Crisis Atlas focuses on damage to buildings and infrastructure because it is a tool directed explicitly towards reconstruction, so the same criticism doesn’t apply (particularly if you look at OCHA’s work more generally, including the information it provided for the map I’ve just reproduced). But if we are to limit ourselves to gazing on structures from space, UNOSAT’s analysis of satellite imagery provides a sharp reminder that these buildings included schools and hospitals (see also here and here); the report also provides a telling comparison between the intensity of destruction in 2009 and 2014:
(4) Smith prefers to construct his own generalised map of damage density using kernel density estimation, a smoothing algorithm that converts point data into a continuous surface. I discussed the way in which the US military uses this technique in ‘Seeing Red’ (DOWNLOADS tab), and what I said there bears repeating:
The maps are known for their dramatic visual impact, and the desired message can be engineered into the production process. One of the most influential handbooks on KDE is published by the US National Institute of Justice and describes how to map crime ‘hot spots’… The authors of the NIJ handbook acknowledge that ‘map production is an iterative process’ and that ‘the first map produced is very rarely the one presented to the target audience.’ They continue: ‘The intended message should also be seen as the driving force behind what the map should look like’ (US National Institute of Justice, 2005: 26, 33).
So let’s turn to the rest of Smith’s message.
(4) Smith justifies the pattern of destruction shown on his maps by claiming that ‘the main objective of Operation Protective Edge was to find and destroy dozens of terror tunnels dug from Gaza into Israel.’ In fact, the stated objectives of the Israeli assault changed throughout the campaign. The attacks were supposedly sparked by the kidnapping and murder of three Israeli teenagers – in the West Bank not Gaza – and as the mission was ratcheted up so their central objective changed: according to the IDF the aim was to put an end to Hamas rockets being fired into Israel (for a radically different view, see Graham Liddell‘s more general discussion at Mondoweiss here). The rhetoric of ‘terror tunnels’ came later. And while Smith is right to draw attention to the swathes of destruction to the east of Gaza’s central spine, he never addresses the human consequences of successive Israeli expansions of this so-called ‘buffer zone’ until it covered more than 40 per cent of Gaza: see my post here for more details.
(5) Towards the end of his analysis, Smith concedes that destruction is not punctiform. Bombs are not ‘pinpoints’ (cf. Nathan Guttman‘s report on Smith’s work and ‘the pinpoint accuracy of Israel’s strikes’), not only because they rarely land exactly on target but also because their blast radiates outwards from the point of impact. But Smith’s ‘fairly generous estimate that a damage point has a 25 meter [82 feet] radius‘ – is in fact a serious underestimate that at the very least halves the blast radius of a 155 mm shell. Here is Mark Perry‘s report that I cited previously, which includes testimony from senior US military officers about the shelling of Shuj’aiyya:
Artillery pieces used during the operation included a mix of Soltam M71 guns and U.S.-manufactured Paladin M109s (a 155-mm howitzer), each of which can fire three shells per minute. “The only possible reason for doing that is to kill a lot of people in as short a period of time as possible,” said the senior U.S. military officer. “It’s not mowing the lawn,” he added, referring to a popular IDF term for periodic military operations against Hamas in Gaza. “It’s removing the topsoil.”
“Holy bejeezus,” exclaimed retired Lt. Gen. Robert Gard when told the numbers of artillery pieces and rounds fired during the July 21 action. “That rate of fire over that period of time is astonishing. If the figures are even half right, Israel’s response was absolutely disproportionate.” A West Point graduate who is a veteran of two wars and is the chairman of the Center for Arms Control and Non-Proliferation in Washington, D.C., he added that even if Israeli artillery units fired guided munitions, it would have made little difference.
Even the most sophisticated munitions have a circular area of probability, Gard explained, with a certain percentage of shells landing dozens or even hundreds of feet from intended targets. Highly trained artillery commanders know this and compensate for their misses by firing more shells. So if even 10 percent of the shells fired at combatants in Shujaiya landed close to but did not hit their targets — a higher than average rate of accuracy — that would have meant at least 700 lethal shells landing among the civilian population of Shujaiya during the night of July 20 into June 21. And the kill radius of even the most precisely targeted 155-mm shell is 164 feet. Put another way, as Gard said, “precision weapons aren’t all that precise.”
(6) Finally, let’s take Smith’s central claim that ‘less than 5% of the land [of Gaza] was affected’ – and reverse it. If Hamas were to say that less than 0.00005% of Israel had been hit by its rockets – to be fair, it’s a difficult calculation to make because Israel has never fixed its borders and so it’s not possible to determine its area – and that the rest of Israel was ‘disproportionately undamaged’, would Israelis have simply shrugged them off?