U.S. military vehicles in Syria’s northern city of Manbij on Dec. 30. (Delil Souleiman/AFP)

A Jan. 13 tweet by President Trump revived the idea of imposing a safe zone — an area meant to protect civilians and minimize refugee burdens within a war zone — in Syria. Syria today looks very different from 2016, the last time there were serious public calls for a safe zone. Russia and Turkey have expanded their support for opposing sides of the conflict with ground troops and airstrikes. The Islamic State has lost much of its territorial holdings and has returned to terrorist tactics. The United States is drawing down its military presence, and it seems that a main U.S. goal for a safe zone would be to protect Kurds, although other civilians would probably seek refuge there. As negotiations between the United States and Turkey over a safe zone intensify, details of what they are considering are still hazy.

My research into the military requirements of safe zones identifies two main forms that one in Syria could take. Proponents of safe zones may underestimate the gaps an air-only safe zone would leave or the number of ground troops required to fully complete its humanitarian mission. What’s most intractable is that no safe-zone option meets every criterion the United States has for a safe zone at the moment — one that protects the Kurds, requires no U.S. ground troops and avoids confrontation with Turkey. What would it take for these safe-zone options to work?

What it would take: In the air

Perhaps the most commonly proposed safe-zone option is one enforced from the air, as employed over Iraq from 1991 to 2003 and Libya in 2011. In this option, intervening aircraft restrict access to a geographic area by enemy aircraft (no-fly zones) or enemy forces (exclusion zones).

Establishing an aerial safe zone typically requires two steps. First, the intervener destroys antiaircraft capabilities. Then it launches regular air patrols to surveil and respond quickly to threats. In Syria, however, that first step would require targeting Syrian and Russian radars and air defenses, bases that usually also house Iranian advisers — a recipe for escalation. However, no American or NATO air force has operated a no-fly or exclusion zone over an area with modern air defenses such as in Syria without first taking them out, making this option a hard sell.

Even if interveners took this unprecedented step, defending an area from ground forces is still a challenge. My model, adapted from other research, suggests that air patrols covering a 25-nautical-mile radius would require roughly 12 sorties and 10 aircraft a day. Such an aerial safe zone would draw a line in the sand that enemy aircraft and ground forces would have to decide whether to cross. However, airstrikes do not fare as well against adaptive enemies that disperse their ground forces or hide among the population — both tactics the Islamic State has employed in the face of coalition airstrikes.

What it would take: On the ground

An aerial safe zone with a ground presence could defend against terrorist attacks and facilitate humanitarian convoys. A ground-enforced zone would take the form of “pockets,” or discrete protection zones set up around existing population centers or newly assembled camps for displaced people. The alternative — manning the full border of a wide swath of land — would waste troops on a front line that would see little action, given how approaching air or ground assaults would be eliminated by friendly aircraft patrolling overhead.

Protecting these sites would require a standard package of tactics employed by U.S. and Western militaries when defending bases, which often resemble mini-towns. Al-Taqqadum air base in Iraq — one of the few bases in modern U.S. counterinsurgency wars for which base security details are readily available — required a U.S. battalion, or 400 to 800 troops, to defend a perimeter of roughly 10 kilometers (6.2 miles). Such a force posture could defend the Kurdish-held town of Kobane, roughly 10 kilometers in perimeter and 5.5 square kilometers (2.1 square miles) and with a prewar population of about 40,000.

My model-building from these rules of thumb suggests that a minimum of 13,600 troops would be needed to protect 500,000 people. To protect 1.5 million civilians — as might be required in a safe zone repatriating refugees or resettling the internally displaced — the number rises to 40,000 troops, or about 10 U.S.-size brigades, not including necessary support troops.

An air-and-ground safe zone would protect civilians from the range of threats they face in Syria, encouraging more to seek refuge in the zone instead of flee. However, the massive footprint it would require — apart from contradicting Trump’s desire to withdraw and being beyond the capacity of many European countries acting alone — may encourage the Islamic State and a bitter Syrian regime and its allies to find creative ways to spoil the international effort.

Who would do it?

Given these requirements, the reality of a truly safe zone for civilians in northern Syria turns on questions of who would operate it — and whom these operators would allow to stay.

Turkey has revived an offer to be the primary guarantor of a safe zone that would more accurately be termed a “buffer zone.” Unlike safe zones, which pursue humanitarian protection as the primary goal, a buffer zone physically distances a threat primarily for security reasons — and Turkey has made no secret its goal would be to distance Syrian Kurds, whom it considers terrorists, from its southern border. Indeed, the Turkish military is preparing a campaign to push back Syrian Kurdish forces and establish a new front line. Although there’s no reason this buffer zone could not assume some humanitarian purposes, such as repatriating refugees, Turkey has stated it would not allow Kurds to be protected by its zone.

Syrian Kurds have offered to take the daunting responsibility of ground enforcement off Western armies’ hands. As my research shows, however, to be meaningfully “safe,” they would require a no-fly zone overhead — and the Kurds have no air force. However, unlike past calls for a safe zone protecting against the regime or Russia, an American- or European-led no-fly zone to protect Kurds would have to also exclude Turkey, a NATO ally, risking an intra-alliance crisis.

Sara Plana is a PhD student in the Department of Political Science at MIT, where her research focuses on military effectiveness, proxy warfare and political violence. Before MIT, Plana worked at the Department of Defense in Washington.