Looking to the Skies: Operation Unified Protector and the Strategy of Aerial Intervention

Abstract What are the different ways in which an intervener can use airpower to enhance a rebel organization’s ability to capture government-held territory? Multiple studies have analyzed how foreign airpower can be employed as a counterinsurgency tool, intended to reduce the frequency and lethality of insurgent attacks. However, many civil wars are fought conventionally, and non-state actors can benefit from military interventions that helps them to overcome the advantage in capabilities often enjoyed by their government adversaries. I analyze how airpower can be used in support of a rebel organization engaged in a conventional civil war, contributing to its ability to produce salient battlefield information. I argue that that an intervener can employ direct attack against heavy weapons and anti-aircraft assets, and interdiction of command and control capabilities and logistics, in order to diminish the government’s advantage in conventional capabilities. In doing so, foreign airpower contributes to the rebels’ ability to capture territory, a crucial intermediary goal and source of battlefield information when attempting to impose defeat on the government. Through a quantitative case study of Operation Unified Protector and the 2011 Libyan Civil War, I find that coalition airstrikes against the Libyan government’s heavy weapons, logistics, and anti-aircraft assets contributed to the Libyan rebels’ ability to capture territory. By contrast, strikes against the government’s command and control capabilities had no effect. ¿Cuáles son las diferentes formas en las que una parte interviniente puede usar el poder aéreo con el fin de mejorar la capacidad de una organización rebelde para ganar territorios controlados por el Gobierno? Existen múltiples estudios que han analizado cómo puede emplearse el poder aéreo extranjero como una herramienta de contrainsurgencia, destinada a reducir la frecuencia y la letalidad de los ataques insurgentes. Sin embargo, muchas guerras civiles se libran convencionalmente, y los agentes no estatales pueden salir beneficiados de intervenciones militares que les ayuden a superar la ventaja a nivel de capacidades que, a menudo, disfrutan sus adversarios gubernamentales. Analizamos cómo se puede usar el poder aéreo en apoyo de una organización rebelde que esté involucrada en una guerra civil convencional, contribuyendo a su capacidad para producir información importante acerca del campo de batalla. Argumentamos que una de las partes intervinientes podría emplear un ataque directo contra armas pesadas y medios antiaéreos, así como la interdicción de las capacidades de mando y control y de la logística, con el fin de disminuir la ventaja del Gobierno en lo referente a las capacidades convencionales. Al hacer esto, el poder aéreo extranjero contribuye a la capacidad de los rebeldes para ganar territorio, lo cual resulta no solo un objetivo intermedio crucial sino también una fuente de información sobre el campo de batalla cuando se intenta imponer la derrota al Gobierno. Concluimos, a partir de un estudio de caso cuantitativo de la Operación Protector Unificado (Operación enmarcada dentro de la intervención militar en Libia de 2011) y de la Guerra Civil Libia de 2011, que los ataques aéreos de la coalición contra las armas pesadas, la logística y los medios antiaéreos del Gobierno libio contribuyeron a la capacidad de los rebeldes libios para ganar territorio. Por el contrario, los ataques contra las capacidades de mando y control del Gobierno no tuvieron ningún efecto. Comment un intervenant peut-il employer les forces aériennes pour améliorer la capacité d’une organisation rebelle de s’emparer de territoires détenus par le gouvernement ? De nombreuses études ont analysé l’emploi des forces aériennes en tant qu’outil de lutte contre l’insurrection, dont le but est de réduire la fréquence et la létalité des attaques des insurgés. Cependant, nombre de guerres civiles se sont déroulées de façon conventionnelle, et les acteurs non étatiques peuvent bénéficier d’interventions militaires qui les aident à contrer l’avantage en matière de capacité que les adversaires gouvernementaux possèdent souvent. J’analyse comment les forces aériennes peuvent être utilisées pour soutenir une organisation rebelle impliquée dans une guerre civile conventionnelle, en contribuant à sa capacité de production d’informations cruciales sur le champ de bataille. J’affirme qu’un intervenant peut avoir recours à une attaque directe à l’encontre d’armes lourdes et d’actifs antiaériens, ainsi qu’à l’interception des capacités logistiques, de commande et de contrôle, afin de réduire l’avantage du gouvernement en matière de capacités conventionnelles. Ce faisant, les forces aériennes étrangères contribuent à la capacité rebelle de capturer des territoires, un objectif intermédiaire et une source d’informations sur le champ de bataille qui s’avèrent déterminants quand on essaie de l’emporter face à un gouvernement. Grâce à une étude de cas quantitative de l’opération Unified Protector et de la guerre civile de 2011 en Libye, j’observe que les frappes aériennes de la coalition à l’encontre des armes lourdes, des capacités logistiques et des actifs antiaériens du gouvernement libyen ont contribué à la capacité des rebelles libyens de capturer des territoires. En revanche, les frappes visant les capacités de commande et de contrôle du gouvernement n’ont eu aucun effet.

disfrutan sus adversarios gubernamentales.Analizamos c omo se puede usar el poder a ereo en apoyo de una organizaci on rebelde que est e involucrada en una guerra civil convencional, contribuyendo a su capacidad para producir informaci on importante acerca del campo de batalla.Argumentamos que una de las partes intervinientes podr ıa emplear un ataque directo contra armas pesadas y medios antia ereos, as ı como la interdicci on de las capacidades de mando y control y de la log ıstica, con el fin de disminuir la ventaja del Gobierno en lo referente a las capacidades convencionales.Al hacer esto, el poder a ereo extranjero contribuye a la capacidad de los rebeldes para ganar territorio, lo cual resulta no solo un objetivo intermedio crucial sino tambi en una fuente de informaci on sobre el campo de batalla cuando se intenta imponer la derrota al Gobierno.Concluimos, a partir de un estudio de caso cuantitativo de la Operaci on Protector Unificado (Operaci on enmarcada dentro de la intervenci on militar en Libia de 2011) y de la Guerra Civil Libia de 2011, que los ataques a ereos de la coalici on contra las armas pesadas, la log ıstica y los medios antia ereos del Gobierno libio contribuyeron a la capacidad de los rebeldes libios para ganar territorio.Por el contrario, los ataques contra las capacidades de mando y control del Gobierno no tuvieron ning un efecto.
Comment un intervenant peut-il employer les forces a eriennes pour am eliorer la capacit e d'une organisation rebelle de s'emparer de territoires d etenus par le gouvernement ?De nombreuses etudes ont analys e l'emploi des forces a eriennes en tant qu'outil de lutte contre l'insurrection, dont le but est de r eduire la fr equence et la l etalit e des attaques des insurg es.Cependant, nombre de guerres civiles se sont d eroul ees de fac ¸on conventionnelle, et les acteurs non etatiques peuvent b en eficier d'interventions militaires qui les aident a contrer l'avantage en mati ere de capacit e que les adversaires gouvernementaux poss edent souvent.J'analyse comment les forces a eriennes peuvent être utilis ees pour soutenir une organisation rebelle impliqu ee dans une guerre civile conventionnelle, en contribuant a sa capacit e de production d'informations cruciales sur le champ de bataille.J'affirme qu'un intervenant peut avoir recours a une attaque directe a l'encontre d'armes lourdes et d'actifs antia eriens, ainsi qu' a l'interception des capacit es logistiques, de commande et de contrôle, afin de r eduire l'avantage du gouvernement en mati ere de capacit es conventionnelles.Ce faisant, les forces a eriennes etrang eres contribuent a la capacit e rebelle de capturer des territoires, un objectif interm ediaire et une source d'informations sur le champ de bataille qui s'av erent d eterminants quand on essaie de l'emporter face a un gouvernement.Grâce a une etude de cas quantitative de l'op eration Unified Protector et de la guerre civile de 2011 en Libye, j'observe que les frappes a eriennes de la coalition a l'encontre des armes lourdes, des capacit es logistiques et des actifs antia eriens du gouvernement libyen ont contribu e a la capacit e des rebelles libyens de capturer des territoires.En revanche, les frappes visant les capacit es de commande et de contrôle du gouvernement n'ont eu aucun effet.

Introduction
On 19 March 2011, French aircraft attacked a column of Libyan government armored vehicles outside the city Benghazi, a focal point of a burgeoning armed uprising against Libya's ruler Colonel Muammar Gaddafi.These strikes were the first of a military intervention that later coalesced into Operation Unified Protector, a NATO-led air campaign mandated to protect Libyan civilians from indiscriminate government attacks.However, the air campaign also prevented the Armed Forces of the Libyan Arab Jamahiriya to effectively fight the National Transitional Council (NTC) rebel coalition, enabling it to defeat Gaddafi (Mueller 2015b).The air campaign thus represents a successful implementation of a denial strategy, as it diminished and destroyed the Libyan government's military capabilities (cf.Pape 1996;Martinez Machain 2015).Furthermore, the NATO intervention is sometimes described as a stand-out case because of its overwhelming reliance on airpower (Mueller 2015a), but it conformed to the general trend that rebels defeat their government adversary when they receive direct foreign military support (cf.Gent 2008;Sullivan and Karreth 2015;Jones 2017).The Libyan case thus provides an opportunity to study the specific ways in which foreign airpower doesor does notcontribute to a nonstate actor's ability to defeat a nominally stronger government.
More specifically, what are the different ways in which an intervener can use airpower to enhance a rebel organization's ability to capture government-held territory?Any successful military intervention must help the supported actor to generate information about is relative advantage in power and capabilities compared to the government, which is partly derived from success on the battlefield (Slantchev 2003;Balch-Lindsay, Enterline, and Joyce 2008;Jones 2017).A non-state actor that seeks to decisively defeat a government must employ conventional tactics to take physical control over important parts of the state (Butler and Gates 2009;Sullivan and Karreth 2015), but governments are at the same time more likely to win when they can employ their conventional capabilities and combined-arms tactics against the rebels (Caverley and Sechser 2017).Pursuing victory thus push conflicts into a space that favors the government, but aerial denial strategies can significantly limit an actor's ability to conduct conventional operations (Pape 1996).Multiple studies have analyzed the micro-level effects of foreign airpower employed as a counterinsurgency tool (Kocher, Pepinsky, and Kalyvas 2011;Johnston and Sarbahi 2016;Gartzke and Walsh 2022;Newton and Tucker 2022), but less work has been dedicated to understanding the ways in which aerial force can support a non-state actor.Based on existing cross-case studies of aerial force and force employment in intrastate conflict, I argue that that an intervener can employ direct attack against heavy weapons and anti-aircraft assets, and interdiction of command and control capabilities and logistics, in order to diminish the government's advantage in conventional capabilities.In doing so, foreign airpower contributes to the rebels' ability to capture territory, a salient source of battlefield information needed to terminate the conflict (cf.Slantchev 2003).
Through a quantitative case study of Operation Unified Protector and the 2011 Libyan Civil War, I find that coalition direct attack and interdiction contributed to the NTC's ability to defeat the government in battle and capture territory.Direct attack against heavy weaponssuch as tanks, armored vehicles, and artilleryallowed the rebel forces to successfully take government-held territory.Similarly, strikes against anti-aircraft assets also facilitated rebel territorial captures, likely because adapting to the coalition air threat made the government forces more vulnerable to ground attacks.Lastly, the interdiction of logistics networks inhibited the government's defensive operations, enabling the NTC to seize towns and villages.By contrastand somewhat contrary to existing analyses of Operation Unified Protector (cf.Barry 2011;Chivvis 2012)interdiction of government command and control functions had no discernable effect on the NTC's likelihood of capturing territory.
Through this analysis, I make several contributions.First, while existing quantitative micro-level studies emphasize the utility of airpower as a counterinsurgency tool, I theorize and analyze how airstrikes can support a non-state actor in a mostly conventional conflict context.Second, I show that theoretical and empirical disaggregation can improve our understanding of denial strategy specifically, and military intervention in general.By focusing of specific targeting approaches, I for example reaffirm recent cross-case evidence that air superiority has a more immediate impact on ground combat than commonly assumed (cf.Saunders and Souva 2020).Third, I provide more general insights as to how coercive pressure and battlefield information is generated in practice, enhancing our understanding of how fighting battles is linked to conflict termination.I demonstrate how intermediary goalssuch as capturing territoryconstitutes a plausible metric for assessing how military intervention affects the generation of salient battlefield information.
This article proceeds as follows.First, I summarize relevant insights regarding how military interventions affect civil wars, and the existing knowledge about the utility of airpower.I then formulate a set of hypotheses on how direct attack and interdiction can alter the balance of conventional capabilities in a rebel organization's favor, enabling them to capture government-held territory.I then proceed to test these hypotheses through a quantitative within-case analysis of the 2011 civil war and NATO intervention in Libya.

Intervention Strategies and Airpower
Rebel organizations that receive direct foreign support are more likely to experience favorable civil war outcomes (Gent 2008;Sullivan and Karreth 2015;Jones 2017).Like all armed conflicts, civil wars can be conceived as episodes of "coercive bargaining" in which a government and one or more rebel organizations attempt to resolve information asymmetries that prevent conflict termination (cf.Slantchev 2003;Smith and Stam 2004).Intervening third parties can facilitate this by providing military support, enabling an actor to win battles and reach battlefield objectives that reveal important information about relative capabilities and power (Balch-Lindsay, Enterline, and Joyce 2008;Gent 2008;Sullivan and Karreth 2015;Jones 2017).Crucially, governments tend to be more successful when employing conventional military forcewhich includes heavy weapons, mechanized forces, and rudimentary combined-arms operationsagainst the rebels (Caverley and Sechser 2017).Rebels that seek to defeat a government must therefore overcome this oft-present preponderance of conventional military capabilities (Butler and Gates 2009;Sullivan and Karreth 2015).Consequently, rebels are more likely to win a civil war when an intervener commits its ownoften well-trained and well-equippedtroops to directly degrade or destroy the government's forces through a "direct-conventional" intervention (Sullivan and Karreth 2015;Jones 2017).The impact of an intervention is thus conditioned by its impact the primary actors' expectation of victory, derived from various battlefield events that foreign direct intervention can alter.
The deployment of modern aircraft and other aerial assets constitutes a potential strategic option for an intervener that seeks to bolster a rebel group's military capabilities.Intrastate conflicts are often associated with insurgency and other irregular tactics and means, but many contemporary civil wars are fought conventionally (Kalyvas and Balcells 2010;Biddle 2021).Conventional civil wars are characterized by set battles along relatively coherent frontlines, often involving the use of heavy weapons (Kalyvas and Balcells 2010, 419;Lockyer 2010, 93;Sullivan and Karreth 2015, 272-3).Evidence from interstate wars shows that airpower can be effectively used in such conventional contexts to produce coercive pressure and drive political bargaining.In particular, denial strategies that target the adversary's armed forces and military capabilities areunlike punishment strategies directed against civilian targetsassociated with a higher likelihood of victory and conflict termination (Pape 1996, 69;Horowitz and Reiter 2001;Allen 2007;Martinez Machain 2015;Allen and Martinez Machain 2019).As observed by Martinez Machain (2015, 543), important insights can be gained through more fine-grained categorization of aerial targeting strategies.For example, recent cross-case evidence demonstrates the importance of accounting for "mixed" denial and punishment campaigns.Denial can also be subdivided into selective or non-selective strategic interdiction of military-industrial and infrastructure targets, and operational interdiction and direct attack against fighting forces on the battlefield (Pape 1996;Martinez Machain 2015).More work is nevertheless still needed to explain exactly how airpower contributes to wider coercive processes (Allen and Vincent 2011;Allen and Martinez Machain 2019).
Multiple studies have demonstrated the analytical utility of theoretical and empirical disaggregation when studying foreign airpower used in civil war contexts.With the notable exception of Allen and Vincent's (2011) analysis of how the 1999 NATO intervention in Kosovo facilitated Serbian conciliatory statements, these quantitative within-case studies focus on the use of airpower against non-state actors as a counterinsurgency tool, and not in support of a rebel organization.For example, Kocher, Pepinsky, and Kalyvas (2011) find that the United States' punishment against civilians in South Vietnam increased the ability of the FNL insurgents to establish local control, as airstrikes increased local support to the insurgents.Similarly, Petersson (2023) finds that NATO airpower affected the Libyan government's ability to take and hold territory during the 2011 Libyan civil war.Furthermore, quantitative micro-level analyses of the wars in Afghanistan, Iraq, Pakistan, Somalia, and Yemen suggest that air-and drone strikes can both exacerbate and diminish the frequency and severity of insurgent attacks (Lyall 2014;Johnston and Sarbahi 2016;Tucker 2021;Newton and Tucker 2022;Gartzke and Walsh 2022).These studies focus on a broad range of mechanisms to explain how aerial force leads to increases or decreases in violence, such as reputation building (Lyall 2014), lack of restraint due to leadership loss (Tucker 2021), and need for retaliation (Gartzke and Walsh 2022).As argued by Allen and Vincent (2011), systematic micro-level studies provides more detailed accounts of how aerial force contributes to wider bargaining processes.
More qualitatively-oriented analyses have showed that airpower can be used to support a non-state actor on the battlefield.For example, the United States provided crucial air support to Northern Alliance militias fighting the Taliban during the 2001 war in Afghanistan (Andres 2006;Andres, Wills, and Griffith 2006;Biddle 2006).Similarly, Saudi Arabia used denial to effectively support groups fighting against the Houthis in Yemen (Shield 2018).Crucially, foreign airpower was found to have limited both the Houthis' and the Taliban's use of their available conventional forces and capabilities (Andres, Wills, and Griffith 2006, 133;Shield 2018, 473, 476).By contrast, comparatively little effort has been made to analyze the systematic micro-level dimensions of denial, despite evidence of its effectiveness.For example, Johnston and Sarbahi (2016, 206-207), Newton andTucker (2022), andPetersson (2023) utilize the causal logic of denialthat it degrades and disrupts military capabilitiesto explain how air and drone strikes affect and actor's battlefield performance, but without fully specifying the different ways in which denial can be pursued (recall Pape 1996;Martinez Machain 2015).As such, more work is needed not only to understand how airpower can systematically aid a non-state actor but also the specific ways in which denial can be employed in a civil war context to reduce a government's advantage in conventional capabilities.
Aerial Force, Battlefield Information, and Rebel Success War ends when the participants learn enough about each other to allow their expectations to converge.This convergence is achieved by fighting battles that produce non-manipulable battlefield information about their relative power and capabilities, which is then reconciled with manipulable information generated through purposive negotiation (Slantchev 2003;Smith and Stam 2004).Similarly, causal mechanisms linking military intervention to various political outcomes focus on the information generated by battlefield outcomes (Balch-Lindsay, Enterline, and Joyce 2008; Sullivan and Karreth 2015;Jones 2017).If intervener airpower is to affect a wider bargaining process (recall Allen and Vincent 2011), it must contribute to the generation of battlefield information that favors the rebels.The purpose of any direct military intervention is generally to bolster the rebels' "destructive capacity and war-fighting effectiveness" that allows them to beat the government in battle (Sullivan and Karreth 2015, 273;Jones 2017).However, Slantchev (2003, 621, 628) argues that battlefield information is "imprecise and noisy" because the "fog of war" makes battles and their outcomes difficult to interpret.Similarly, focusing merely on "how well bombs destroy targets" or how it affects the level of violence cannot fully account for the successful use of aerial force (Kocher, Pepinsky, and Kalyvas 2011;Allen and Martinez Machain 2019).
Nevertheless, battles are almost never directly linked to war's political outcome.In war, outcomes at one level constitute factors that cause outcomes at the superordinate level (Smith 2005, 131-132).As such, winning battles is an important tactical outcome, and these victories mayor may notcontribute to the achievement of so-called intermediary goals defined by an actor's strategy.Intermediary goals must just like battlefield information be reconciled with various political factors, and they constitute the link between battles and politics and "must be sequentially achieved prior to the fulfilment of the war's political purpose" (Smith 2005, 133;Herberg-Rothe 2014, 904).Understanding how airpower contributes to generating salient battlefield information is thus linked to how it systematically affects the rebel's ability to achieve its intermediary goals.
Rebels may seek many different intermediary goals, but if the political objective is to impose a decisive defeat on the government they must acquire enough military strength to seize important territory and infrastructure (Sullivan and Karreth 2015, 272-273).Capturing territory also has intrinsic signaling value that affects actors' bargaining positions (Smith and Stam 2004), and it bolster relative power by providing access to vital resources and safe spaces from which to launch attacks (Buhaug, Gates, and Lujala 2009;Cunningham, Skrede Gleditsch, and Salehyan 2009;Kocher, Pepinsky, and Kalyvas 2011;Sullivan and Karreth 2015;Petersson 2023).Furthermore, a recent study finds that changing territorial control patterns are within-conflict factors that can explain negotiation behavior during civil war (Pechenkina and Thomas 2020).As such, contributing to changing territorial control should affect the broader, politically driven termination process.Although airpower cannot seize tangible objectives directly (Meilinger 2003), it can facilitate rebel territorial captures by bolstering their battlefield capabilities, and thereby the generation of salient battlefield information.
When rebels are pursuing territorial conquest, the government cannot avoid direct combat and is forced to use conventional force employment to defend itself.However, a government generally has superior troop strength, equipment, and other operational capabilities compared to the rebels (Butler and Gates 2009, 331;Sullivan and Karreth 2015, 272-273).Furthermore, Caverley and Sechser (2017) provides compelling evidence that governments with combined-arms capabilities are more likely to decisively defeat a rebel group by using its superior mobility and logistics to produce decisive battles.Rebels that seek battlefield victories must also abandon irregular force employment for a more conventional approach that maximizes their ability to hurt the government, at the expense of increasing their own vulnerability to potentially devastating losses from government attacks (Butler and Gates 2009;Sullivan and Karreth 2015).However, a successful denial campaign can seriously undermine an actor's ability to fight its adversary successfully using conventional means (Pape 1996).Below, I outline a set of hypotheses regarding how intervener airpower can increase the rebels' prospects of winning battles by degrading and disrupting the government's conventional capabilities, and thereby capture territory.

Direct Attack
Denial can be achieved through direct attack, which entails using airstrikes to weaken the adversary's troops at decisive points on the battlefield.These weaknesses are not conclusive by themselves, but they can facilitate successful attacks from friendly ground forces.When used against a defending force, direct attack can contribute to weakening a part of a front or defended position, allowing friendly ground forces to break through (Pape 1996, 70-73, 78;Martinez Machain 2015).As such, we should expect that the effects of direct attack have an immediate impact before the government can recover and replace losses.When civil wars take on conventional characteristics, the warring parties tend to rely extensively on heavy weapons, including artillery, armored vehicles, and tanks (Kalyvas and Balcells 2010, 419;Lockyer 2010, 98).Caverley and Sechser (2017, 707) argue that governments can utilize heavy equipment to "flush out, surround, and contain [rebel] forces."While some non-state actors employ advanced heavy weapons, Biddle (2021, 60) concludes that "most states' weapon technology at any given time is more sophisticated and more lethal than most nonstate actors'."By striking against the government's heavy weapons the intervener can depriveat least in the short-termthe government of its material advantage in a contested area, allowing the rebels to win battles and capture territory: H1: Rebels are more likely to capture territory when government-heavy weapons are hit by airstrikes.
Direct attack can also be used to disrupt the government forces by making them more vulnerable to further air attacks.Saunders and Souva (2020) find that air superiority, which is achieved through the neutralization of the adversary's aircraft and/or air defense systems, can have a profound effect on battlefield dynamics.While Pape (1996, 58) argues that air superiority mainly enables other aerial strategies, others suggest that it also "conveys important advantages" through a specific form of denial (Mueller 1998, 214;Saunders and Souva 2020).In the civil war, governments are likely to enjoy the benefits of a friendly sky, as few rebel groups have historically possessed the capabilities needed for posing a credible air threat. 1  In any war, combat always entails balancing "[l]ethality (or firepower) and survivability (via cover and concealment)," with measures taken to protect oneself simultaneously reduces the ability to effectively hurt the adversary (Biddle 2021, 54).Prioritizing lethality entails adopting more conventional force employment, which leaves one's troops more exposed (Biddle 2021).Without a credible air threat, governments do not need to sacrifice lethality 1 Technological advances made regarding armed drones and loitering munitions could change this.

INTERNATIONAL INTERACTIONS
in order to avoid aerial attack.However, measures taken to reduce vulnerability to airstrikeslike dispersing or concealing troopsreduces force concentration and increases exposure to ground attacks (Biddle 1996, 162-163;Andres, Wills, and Griffith 2006;Saunders and Souva 2020, 746).By increasing the vulnerability of the government's forces to air attack, the intervener can effectively limit their lethality.Even if all hostile air activity may prompt dispersal and concealment, the loss of anti-air assets will exacerbate the vulnerability of the targeted government forces.We should expect that the loss of anti-air systems have an immediate, short-term effect on the government's forces in the targeted area, making the rebels more likely to capture territory: H2: Rebels are more likely to capture territory when government anti-air assets are hit by airstrikes.

Interdiction
The second major form of denial is interdiction of the adversary's combat support functions.Governments may enjoy superior conventional capabilities, but deploying heavy equipmentsuch as armor and massed artillery requires both leadership and coordination capabilities, and extensive logistics (Caverley and Sechser 2017, 704, 706-707).For example, defensive operations entail maintaining control over a relatively coherent frontline, and sustaining the cohesion and supply of one's troops (Biddle 2004;Kalyvas and Balcells 2010;Lockyer 2010).Government forces can also make use of their capabilities to uphold force mobility, enabling them to both respond to threats, and "evade or escape perilous engagements more easily" (Caverley and Sechser 2017, 704, 706).Interdiction can be used to undermine an actor's ability to effectively respond to the adversary's efforts to break its defenses, particularly by disrupting logistics networks and command and control functions (Pape 1996, 77;Martinez Machain 2015, 544).Rather than contributing to breaking the government's defenses, interdiction disrupts its ability to effectively coordinate responses to such breaks.When successful, interdiction induces "operational paralysis" in an adversary, which prevents it from using its forces effectively (Pape 1996, 72, 77;Martinez Machain 2015).Interdiction therefore has most utility when ground forces are in the process of concentrating and converging on a crucial section of the battlefield (Pape 1996, 77).We should therefore expect that interdiction efforts by an intervener is most effective when conducted prior to battle in a protracted effort to degrade the government's combat support functions.By degrading the government's command and control functions and logistics, the intervener can prevent it from effectively using its forces to their full potential, thus enabling rebels to capture territory: H3: Rebels are more likely to capture territory when the government's command and control functions are degraded by airstrikes.H4: Rebels are more likely to capture territory when the government's logistics networks are degraded by airstrikes.

Research Design
The hypotheses laid out in the previous section are tested through a quantitative case study of Operation Unified Protector, NATO's intervention in the 2011 Libyan Civil War fought between the Libyan government and the National Transitional Council (NTC).Even though the case has been subjected to previous analysis (Chivvis 2012;Mueller 2015a;Petersson 2023), the systematic effects of specific targeting approaches on the rebel's battlefield performance remain underexplored.The 2011 NATO intervention in Libya is sometimes described as "extreme" given its overwhelming reliance on airpower (Mueller 2015a, 6).However, from a broader perspective, the case is a relatively typical example of rebel victory facilitated by a direct military intervention by a highly capable third party (cf.Gent 2007Gent , 1090;;Gent 2008;Sullivan and Karreth 2015, 273;Jones 2017, 58).The case is also appropriate for testing the argument laid out above for three main reasons.First, unlike most existing micro-level studies of airpower, the Libyan case was one where airpower was used to support a non-state actor.Second, both the Libyan government and the NTC made extensive use of conventional force employment and heavy weapons (cf.Kalyvas and Balcells 2010;Bell andWitter 2011a, 2011c).For example, Gaddafi's "elite" paramilitary unitssuch as the 32nd and 77th brigadesemployed infantry, armored vehicles, tanks, and artillery (Bell and Witter 2011a, 22).Lastly, while the Libyan government was vulnerable to a systematic air campaign (cf.Barry, 2011;Mueller 2015b), the NTCwhich were the ones bearing the brunt of the fightingwas not a highly skilled fighting force.In fact, while it included professional soldiers, many NTC fighters were volunteers with limited training and equipment, lacking the skills required to coordinate effectively with friendly airpower (Chivvis 2012, 74).Modern airpower works best when integrated with a skilled ground force (Pape 1996;Andres 2006;Andres, Wills, and Griffith 2006;Martinez Machain 2015), which implies that the Libyan case did not exhibit all factors associated with the unproblematic execution of an air campaign.
I employ logistic regression to test the hypotheses, using a disaggregated dataset that covers the entire 2011 Libyan civil war.The unit of analysis is "district-day," which means that there is one observation for each of Libya's twenty-two districts (shabiya) from 15 February to 23 October (n ¼ 5,522).Direct attack and interdiction do not occur randomly but is driven by strategic decisions.NATO airstrikes focused on areas with a high risk of civilian harmwhich should correspond to intense fighting between the government and the rebelsand areas were strikes were requested by the NTC (Chivvis 2015;Wehrey 2015).As detailed below, I also account for a set of factors that can explain the spatial dimension of the airstrikes and the relative difficulty for rebels to capture territory.Furthermore, in order to mitigate potential endogeneity caused by simultaneity, I make use of qualitative illustrations to interpret and nuance any correlationsor lack thereofobserved in statistical analysis.This information is collected from various qualitative reports and analyses (for example Bell and Witter 2011a;Chivvis 2012;Mueller 2015a) and is mainly used to illustrate that airstrikes indeed affected the rebels ability to fight the government and that they preceded territorial captures in time.

Measuring Rebel Territorial Captures
The data on rebel territorial captures were primarily coded from Englishlanguage media reports retrieved using the online news archive Factiva.Major news agenciessuch as Reuters, AP, and AFPwere predominantly focused on events in eastern and western Libya, rather than the south of the country.To compensate for this uneven coverage, I included the BBC Monitoring Service in the Factiva searches, as it republishes news from local and regional agencies.I also conducted more open-ended searches for southern Libya as an additional complement.Multiple secondary sourcessuch as the Africa Research Bulletin, and the Armed Conflict Location and Event Data Project (ACLED, see Raleigh et al. 2010)were used to complement the media reports.In the dataset, a territorial capture is conceptualized as rebel forces seizing control over a town, village, or district capital.Settlements are not the only relevant territorial entity relevant, but they are easily identifiable and often used ask markers of fluctuating control in a conflict.When coding the data, self-reported rebel captures were verified by some other source, given recurring premature claims that a town or village had been seized from the government.As an example, there were multiple claims that the NTC was in control over Tripoli before the capital fell into rebel hands on 26 August.The geographical distribution of the rebel captures is displayed in Figure 1, and while all districts saw at least one territorial capturefor 109 in totalthe distribution across the country was relatively uneven, with the Jabal al Gharbi, Al Wahat, and Sirte districts seeing the most captures.
The dependent variable is Rebel Capture, a binary variable that denotes whether (1) or not (0) the rebels captured one or more towns, villages, or district capitals during any given district-day.A binary variable is unable to capture relative importance, but it precludes the need to assess the relative importance of each individual settlement, or the relative importance of capturing more than one settlement in a single day.For example, Gharyan in the Nafusa Mountains had more strategic value than most other district capitals as it overlooked the roads towards Tripoli, and its capture was more important than the simultaneous capture of the three villages Zawiyat al Baqul, Lawania, and Ghanymma.It is necessary to point out that some towns or villages were captured multiple times by the rebels, such as Ajdabiyah (21 February, 17 March, 26 March), since the government occasionally retook some territory. 2As argued by Cunningham, Skrede Gleditsch, and Salehyan (2009, 574-575), rebel power to target and power to resist the government are not the same thing.Focusing on rebel territorial captures constitutes a valid assessment of how rebels increase the pressure against the government in pursuit of victory, rather than just avoiding defeat (recall Sullivan and Karreth 2015).

Measuring Intervener Denial and Interdiction
Information about the types of targets struck by coalition airstrikes during each district-day was primarily collected from NATO's daily Operational Media Updates.The updates are available via the official NATO website (North Atlantic Treaty Organization, 2011) and contain information about those targets deemed "key" by the coalition.The information provided by the updates is disaggregated enough to determine the district, day, and type of target.To supplement missing data, the information in these updates were complemented by various sources, including operational updates from national defense ministries and media reports retrieved from Factiva using a standardized search string.Of the four target categories employed here, heavy weapons was the one most hit by coalition airstrikes.In total, 1,841 pieces of heavy equipment were struck, which is barely less than the other three categoriesanti-aircraft (815), logistics (879), and command and control (256)combined.As shown in Figure 2, the spatial distribution of the airstrikes was highly uneven between districts, which should be expected as some districts were more effected by fighting than others.

Direct attack
The variable Heavy Weapons includes the number of tanks, infantry fighting vehicles, armored personnel carriers, artillery, multiple-rocket launchers, technicals, unspecified military trucks, 3 and fixed firing positions hit by NATO airstrikes on any given district-day.This variable includes those resources that enhanced the Libyan government's ability to directly fight the rebels.The variable Anti-Aircraft assets include surface-to-air missile launchers and batteries, anti-aircraft guns, fixed and mobile radar sites, and anti-aircraft storage sites hit by NATO airstrikes on any given district-day.One potential issue here is that anti-aircraft weaponsparticularly selfpropelled anti-aircraft gunscan be used against ground targets.However, as it is not possible to identify such dual use, all asset labeled as "anti-air" are considered as such.The types of assets included here are all intended to directly engage an enemy force, and losing them would undermine the government's ability to resist rebel attack.I also include alternative operationalizations that measure the number of heavy weapons and anti-aircraft assets struck over the preceding seven days.

Interdiction
As theorized above, the degradation of command and control capabilities and logistics networks should constitute a more protracted targeting effort leading up to a decisive battle.I therefore measure the two types of interdiction as the sum of each target type hit over the preceding seven days, in order to account for both the protracted and transient nature of degradation. 4Firstly, Command and Control includes the number of command and control bunkers/nodes/sites, headquarters, and operational facilities hit over the preceding seven days.Second, Logistics measure the number of logistic nodes, ammunition depots/bunkers, missile storage facilities, 5 vehicle storage and maintenance facilities, and various miscellaneous military sites, such as military compounds/facilities and camps hit over the preceding seven days.I also include alternative operationalizations that measure the number of command and control, and logistics, targets hit during each district-day. 63 If a truck is specified as having transported munitions or other supplies, it would be considered a "logistics" target rather than a direct battlefield asset.

4
Robustness test reported in the Online Appendix also include alternative three-and 14-day measures.

5
Missile storages reported to contain anti-aircraft munitions are coded as anti-aircraft targets.

6
All variables measuring airstrikes are lagged one day to account for the time order of events expected in the theoretical argument.Measures of direct attack and interdiction do not overlap in time.

Control Variables
As summarized in Table 1, I include a series of control variables that could potentially affect both the rebels capturing territory, and the conduct of the air campaign.First, I include two control variables relating to the use of aerial force.Successful degradation and disruption of an actor's battlefield performance entails making resupply and reinforcements more difficult, something that could be achieved by striking targets in adjacent areas.The variable Contiguous Targets measures the number of heavy weapons, antiaircraft, command and control, and logistics targets hit in each districts contiguous districts. 7In order to gauge the relative importance of specific targeting approaches, it is necessary to account for the overall coalition aerial activity to ensure that any observed effects are not explained by the mere presence of foreign aircraft.The variable Sorties measured the overall number of sorties flown by the coalition each day.
A set of control variables that could account for both the conduct airstrikes and rebel territorial captures, are also included.The battle is operationalized as any event involving fighting between the Libyan government and the rebels that causes at least one fatality on any given district-day as reported by the Uppsala Conflict Data Program (UCDP) Georeferenced Event Dataset version 20.1 (Sundberg and Melander 2013), and all battleevents between government and rebel forces reported by ACLED (Raleigh et al. 2010).I also include an additional variable that covers the intensity of battle, measured as the number of fatalities reported by UCDP-GED and ACLED.The control variables also cover geographical factors, including the natural log of each district's Population Density, and a measure of Terrain Ruggedness (Biddle 2004;Shaver, Carter, and Shawa 2019).The variable Distance to District Capital measures the distance between each district capital and Tripoli and after the rebel capture of the capital 26 August, the distance between each district capital and Sirte (Buhaug, Gates, and Lujala 2009).I also include a variable that measures Conflict Duration in days (Allen and Vincent 2011).Lastly, to account for the diffusion of battlefield events, a variable that measures the Time between government territorial losses at the district level; squared and cubed variations of the last variable are also included (Carter and Signorino 2010).A binary variable that denotes whether or not the rebels captured territory in a contiguous district during the preceding three days is also included.

Empirical Analysis
Colonel Muammar Gaddafi seized power in Libya after a coup in 1969, but popular protests that began in early 2011inspired by similar protests sweeping the Middle East and North Africachallenged his rule.Gaddafi's security forces mounted a violent response to the country-wide protests, and the hostilities escalated quickly as local resistance movements were joined by defectors from Gaddafi's own ranks.As the violent protests evolved into a full-scale armed uprising, the National Transitional Council (NTC) was formed to coordinate the anti-Gaddafi factions.Even though the Armed Forces of the Libyan Arab Jamahiriya were in a relatively poor state, Gaddafi could also rely on well-equipped paramilitary units such as the 32nd Reinforced Brigade lead by his son Khamis Gaddafi to effectively fight the NTC (Bell and Witter 2011a).Following a succession of successful government offensives against rebel holdouts, and with the major antiregime strongholds in Misrata and Benghazi in immediate danger, the United Nations approved the establishment of a no-fly zone to prevent further civilian harm.On 19 March, aircraft from a NATO-led coalition began to strike forces loyal to the Libyan government.The air campaign enabled the rebels to turn the tide of the war, and the NTC routed the final pockets of government resistance in Bani Walid and Sirte in late October.Gaddafi himself was captured and killed by rebel forces on 20 October.The predominant assessment is that Operation Unified Protectorthe formal name given to the NATO-led air campaignmade a vital contribution to the NTC's victory over Gaddafi.The coalition's airstrikes allowed "the rebels to win against far better equipped and trained opponents" (Barrie 2012, 63), and "denied the regime forces the use of air assets and the ability to mass their weapons against the rebellion" (Chivvis 2012, 82).
The UN-mandated action to protect Libya's civilians from government attacks, but coalition aircraft provided both close air support and interdiction that weakened and attrited Gaddafi's forces (Barry 2011, 7;Chivvis 2012, 82;Chivvis 2015, 29;Mueller 2015b;Wehrey 2015).In line with these existing assessments, the statistical analysis presented below suggest that airstrikes against government heavy weapons and logistics made the NTC more likely to capture territory.Furthermore, I also find that airstrikes against anti-aircraft assets contributed to the rebels' ability to take towns and villages; this is a contribution that has received less attention in previous analyses.Lastlyand somewhat contradictory to existing assessments -I find that strikes against the government's command and control capabilities had no effect.

Direct Attack
Coalition airstrikes against the Libyan government's heavy weapons contributed to the NTC's territorial captures, which lends support to Hypothesis 1 (see Table 2).The main Model 1 shows that more heavy weapons hit during the preceding district-day made the Libyan rebels more likely to capture towns and villages held by Gaddafi's forces.When the number of targets hit by airstrikes in contiguous districts and contiguous territorial captures are excluded, strikes against heavy weapons that narrowly fails to reach conventional levels of significance (p <.10, see Model 2).Even though it is difficult to make strong claims based on this result, this is consistent with direct attack being more effective if there are concurrent effort to prevent building up and reinforcing heavy weapons in a contested areas.Furthermore, airstrikes against heavy weapons have a significant effect when controlling for the overall coalition air activitymeasured by the number of daily sorties flown (Model 3)and when the intensity of battle is accounted for (Model 4).
Interestingly, higher numbers of heavy weapons hit over the preceding seven days also had a significant, positive effect on the NTC's likelihood of capturing territory (see Model 5).While the theoretical argument forwarded here stipulates that strikes on heavy weapons should have a shortterm, immediate effect on the rebels' battlefield performance, this type of cumulative effect is not particularly surprising.For example, by striking heavy weapons repeatedly over time in a contested area will most likely inhibit the timely movement of defending forces (recall Pape 1996).When the analysis is restricted the districts most affected by the fightingthat is, the more densely populated coastal areas from the border with Tunisia in the west, to Benghazi in the east containing most the major cities (Wehrey 2015) the effect of strikes against heavy weapons remain significant   (Model 7).Likewise, strikes against heavy weapons have a significant, positive effect on the NTC's likelihood of capturing territory when the analysis is limited to the duration of the air campaign (19 March to 23 October, see Model 8).
The positive effect of airstrikes against heavy weapons on the NTC's offensive action is supported by existing assessments of the conflict.On a general level, Mueller (2015b, 377) argues that the NATO-led coalition "[paved] the way for rebel victories, particularly by striking regime armored vehicles, artillery, and other heavy weapons."There are also specific examples of how timely strikes on government tanks, armored vehicles, and artillery facilitated NTC offensives both along the coast east of Sirte, and in the Nafusa mountains (Bell and Witter 2011c, 13-14;Bell, Butts, and Witter 2011, 15).Furthermore, when rebel forces overran Ajdabiyah in late March, their attack was held back until coalition airstrikes "destroyed at least four T-72 tanks and an artillery piece" (Bell and Witter 2011c, 13).Using the main Model 1 as a baseline, the likelihood of the NTC capturing territory was 1.9 percent during any given district day.Increasing the number of heavy weapons hit from zero to the maximum sixty increases the likelihood that the NTC would capture territory from 0.7% to 7.6%.Furthermore, NATO hitting ten or twenty heavy weapons would change the likelihood from 1.1% to 1.6% respectively.Coalition aircraft hitting twenty heavy weapons would thus make the NTC more than twice as likely to capture territory.Hypothesis 1 is thereby supported.
Turning to Hypothesis 2, there is evidence that airstrikes against anti-aircraft assets made the NTC more likely to capture territory.Model 1 in Table 2 shows that the number of anti-aircraft targets hit during a given district-day had a significant, positive effect on the NTC's likelihood of capturing territory.The effect holds when contiguous factors are excluded (Model 2) when the overall sortie numbers and fatalities are accounted for (Models 3 and 4), and when the analysis is restricted to only included the districts most affected by the civil war (Model 7).However, the effect does not hold up when the alternative seven-day measure of anti-aircraft targets hit is employed (Model 5), or when the sample is restricted to the duration of the air campaign (Model 8).Consistent with these results, Chivvis (2012) observes that NATO' air presence prevented the government forces from operating freely, forcing them to seek cover in urban areas.Despite losing its fixed air defenses, the government forces used mobilealbeit aginganti-aircraft weapons, including anti-aircraft guns and mobile missile launchers (Barrie 2012;Wehrey 2015).NATO targeted such anti-air assets whenever they were located (Barrie 2012, 62), giving the government troops valid reason to believe that further strikes were incoming.Furthermore, Wehrey (2015) shows that coalition airpower had a nonkineticalbeit opaqueeffect on the cohesion of regime troops.For example, captured government artillery operators from the areas around Brega and Misrata were demoralized by "NATO's persistent bombardment and overflight."Accounts from the Nafusa Mountains also describe how regime troops fled at the mere sound of passing jets (Wehrey 2015, 59).While anecdotal, these accounts describe a sense of vulnerability to the presence of coalition aircraft, which likely would have been exacerbated if available anti-aircraft weapons were destroyed.Even though strikes against anti-aircraft assets were mostly intended to facilitate the conduct of the air campaign, the NTC appears to have exploited strikes against these assets to advance in areas where regime forces were forces to adapt to increased air threat.
Hypothesis 2 is therefore supported, but this result should be treated with some caution.When compared to the supplementary evidence of the strikes against heavy weapons, the situational accounts from Libya do not provide as clear-cut illustrations of the mechanism I propose for strikes against anti-aircraft targets, even if anecdotal evidence suggest an acute concern about exposure to coalition aircraft among government troops.Using Model 1 as the baseline, increasing the number of anti-aircraft targets hit from zero to twenty-six increases the likelihood of the NTC capturing territory from 0.8% to 3.6%.Increasing the target hit to ten, fifteen, or twenty corresponds to likelihoods of 1.4, 1.9, and 2.5 respectively.It is also necessary to reiterate that strikes against anti-air assets enable the safe conduct of further airstrikes (recall Pape 1996).The findings presented here validates recent claims by Saunders and Souva (2020) that air superiority is an underappreciated component of aerial force, which can have an immediate impact on an actor's battlefield performance.It also provides insights on a potential contribution not as clearly forwarded by existing analyses of Operation Unified Protector.
Even though Hypotheses 1 and 2 are supported by the main analysis, the theory forwarded here stipulates a relatively short-term effect of direct attack.I therefore conduct a temporal analysis that is reported in Table 3. 8Increasing the temporal lag on the number of heavy weapons hit from one dayused in the main Model 1to two and three days does not alter the effect.In both instances, higher numbers of heavy weapons hit during a district-day had a significant, positive effect on the NTC's likelihood of capturing territory.The effect disappears if the lag is increased to four days, suggesting that strikes on heavy weapons have an effect that dissipates after a few days.This temporal pattern conforms to Wehrey's (2015, 51, fn. 26) assessment that NATO direct attack in many instances was aimed at government forces striking civilian areas, rather than "in direct support" of the rebel forces.The temporal analysis supports the notion that the NTC could exploit situations where government heavy weapons were recently destroyed by airstrikes, even if multiple assessments conclude that coordination between NATO and the NTC was inconsistent at best, and dysfunctional at worst (cf.Chivvis 2012;Wehrey 2015). 9 Increasing the lag on the number of anti-aircraft assets hit from one to two, three, or four does not yield any significant results.This is not unexpected since rather than having a direct effect on the targeted actor, increased exposure to hostile aircraft will force the targeted forces to adapt and take various countermeasures that only temporarily reduces its ability to use for example heavy equipment.The Libyan government forces used multiple different tactics to reduce their vulnerability to coalition aircraft such as "shoot and scoot" tactics to minimize the time artillery was exposed in open terrain, and the use of armed trucks ("technicals") similar to those used by the NTC (Barry 2011, 6;Bell and Witter 2011c, 22;Barrie 2012, 63;Wehrey 2015, 55-56).This shows that the government forces were capable of adopting tactics that reduces vulnerability to air attack, which may explain the passing effect observed here.

Interdiction
I turn now to an assessment of the coalition's interdiction efforts.Contrary to Hypothesis 3, airstrikes against the Libyan government's command and control assets had no effect on the NTC's likelihood of capturing territory.As reported in Table 2, the number of command and control targets hit by coalition airstrikes did not have a significant effect in any of the reported models.These results do not conform to existing assessments that NATO targeting of the government's command and control capabilities was one of A series of robustness tests fail to uncover any significant interaction effect between battle and the targeting approaches analyzed here (see Online Appendix).
the air campaign's central contributions (Barry 2011;Bell and Witter 2011b, 29;Chivvis 2012).However, the pre-war state of the Libyan Army's chain of command may explain this lack of effect.According to Bell and Witter (2011a, 22), the regular Libyan units had unclear and "perplexing" command structures, while the paramilitary formations "had independent chains of command, with no horizontal hierarchy existing between commanders."The Libyan command and control capabilities thus appears to have been in relative disarray even before the air campaign, which could potentially damped the systematic effects of striking them.Robustness checks confirm that altering the time period over which strikes are measured to three or fourteen days, applying a cumulative measure of command and control assets hit in a district, or measuring the number of such assets hit each calendar dayrather than district-daydoes not yield any significant effect.In fact, the only conditions where strikes against these targets have a positive effect was when none of the other target categories were controlled for (see Online Appendix).In sum, Hypothesis 3 is not supported.Lastly, as stipulated by Hypothesis 4, coalition strikes against the government's logistics network contributed to the NTC's likelihood of capturing territory.The results in the main Model 1 (Table 2) shows that higher numbers of logistics target hit over the preceding seven days had a positive, significant effect on the NTC's likelihood of capturing territory.This effect is consistent across most alternative model specifications reported in Table 2.The only exception is that the number of logistics targets hit during a given district-day had no effect on the rebels' likelihood of capturing territory (see Model 6).This is not unexpected, since while operational paralysis could be achieved through a single debilitating attack, degrading the targeted government's logistics in a contested area should entail a more protracted targeting effort.In fact, robustness checks provide further evidence that the interdiction of logistics plays out over time.Substituting the sevenday measure for a three-day measure yields no significant results, but a higher numbers of logistics targets hit over the preceding fourteen days has a significant, positive effect on the rebel's likelihood of capturing territory (see Online Appendix).
This finding is largely in line with situational accounts from the 2011 Libyan civil war.In particular, multiple sources state that attrition of regime supply and logistics networks made a substantial contribution the NTC's victory over Gaddafi (Barry 2011;Chivvis 2012).Again, the battle for Ajdabiyah in late March corroborates the statistical results, as the NTC's capture of the town was facilitated by NATO aircraft cutting government supply lines from Sirte (Bell and Witter 2011c).Furthermore, after a protracted stalemate in the Nafusa Mountainsin part caused by insufficient airstrikes against the government troops' supply network -NTC forces were able to capture Bir al Ghanem after NATO began targeting regime logistics nodes in Gharyan, Mizdah, and al-Aziziyah (Wehrey 2015, 63).In substantive termsbased on the estimates in Model 1increasing the number of logistics targets hit during the preceding seven days from zero to the maximum seventy four increased the likelihood of the NTC capturing territory from 0.7% to 6.7%.Furthermore, fifteen, thirty, and forty five targets hit over the preceding days corresponds to a 1.2, 1.8, and 2.9 percent likelihood of rebel territorial capture.Hypothesis 4 is therefore supported.
To summarize, in line with Hypotheses 1 and 4, coalition strikes against the Libyan government's heavy weapons and logistics network increased the likelihood that the NTC would capture territory.These two findings conform to existing assessments of how coalition airpower facilitated the rebels' victory over Gaddafi and mirror the claim that both close air support and interdiction were important parts of the air campaign.Furthermore, I find that direct attack against the regime's anti-aircraft assets also correlated with the NTC territorial captures, which confirms Hypothesis 2. While striking these targets were undoubtedly a prerequisite for conducting the air campaign itself, depriving the government forces of their protection against air attack made them more vulnerable to opportune attacks by the rebels.Lastly, contrary to both existing analyses and Hypothesis 3, coalition airstrikes against the Libyan government's command and control capabilities had no effect on the NTC's likelihood of capturing territory.It is plausible that the present analysis cannot accurately capture the effects of this targeting approach, as command and control may operate in ways that are difficult to capture using the methods employed here.The results reported here should thus not been seen as general evidence that targeting of command and control assets is ineffective.
It was the NTC that defeated the Libyan government (Mueller 2015b), and Figure 3 exemplifiesnot all captures are displayedthe changing rebel territorial control.In the east, rebel territorial control expanded westward from Benghazi, along the coast towards Sirte.In the west, rebels in Misrata were eventually able to march against Tripoli via Zliten and Khoms, and against Sirte via Tawergha.Rebel forces in the area around Nalut and Zintan pushed towards the coast, in the directions of Zawiyah and Tripoli, through Gharyan and Bir al Ghanem.After the fall of Tripoli in late August, NTC main efforts concentrated on capturing remaining government holdouts around Bani Walid and Sirte.There were some unsuccessful attempts to negotiate an end to the war (cf.Chivvis 2015, 37), but the Gaddafi regime was mainly defeated through the NTC's effort to systematically capture its remaining holdouts.The government's defeat was nevertheless not absolute, illustrated by a clash between Gaddafi loyalists and NTC forces near Bani Walid in late November 2011 (International Crisis Group 2011).The end to the 2011 civil war is thus best described as an implicit bargain, driven by the NTC's ability to defeat government forces in battle and capture territory, causing loyalist resistance to eventually fizzle out.Analyzing how battlefield information is generatedfor example through the capture of territoryand how foreign military intervention affects this process is thus a crucial aspect of understanding the link between military intervention and civil war termination, particularly in the absence of extensive deliberate bargaining.

Conclusion
In this paper, I have analyzed the different ways in which an intervener can use airpower to enhance a rebel organization's ability to capture government-held territory.Based on existing cross-case research on airpower and force employment in civil war, I argued that that an intervener can employ direct attack against heavy weapons and anti-aircraft assets, and interdiction of command and control capabilities and logistics, in order to diminish the government's advantage in conventional capabilities.In order to test this argument, I conducted a quantitative case study of Operation Unified Protector and the 2011 Libyan Civil War.The results suggest that coalition airstrikes contributed to NTC territorial captures by striking heavy weapons to weaken government defenses, by increasing vulnerability to ground attacks by systematically attacking regime anti-aircraft assets, and through degradation of the Libyan armed forces' logistics networks.By contrast, I was unable to find any significant effects of strikes against the government's command and control networks.The coalition airstrikes were both used to directly support the NTC, but the rebel forces also appear to have been able to exploit the effects of the strikes even when coordination with NATO was insufficient.Even though governments benefit from their advantage in conventional capabilities when fighting a rebel organization, the Libyan case indicates that foreign airpower can be used to effectively remove this advantage.
The findings have several important implications.First, while existing quantitative micro-level studies emphasize the utility of airpower as a counterinsurgency tool, I theorize and analyze how airstrikes can support a nonstate actor in a mostly conventional conflict context.Many contemporary civil wars fought conventionally (Kalyvas and Balcells 2010), and the proliferation of different aerial weapon systemsboth conventional aircraft and armed UAVsmakes it is crucial to understand how airpower may, or may not, offset the advantages held by governments and other actors that can employ conventional means and tactics to fight against non-state actors.This is illustrated by cases such the United Arab Emirate's interference in the post-Gaddafi conflicts in Libya, the Western support to SDF fighting IS in Syria, the Saudi intervention in the Yemeni civil war, and the Armenian involvement in the 2020 Nagorno-Karabakh War.Existing studies of air support to "local allies" also tends to assume an extensive integration between foreign airpower and a non-state actor (Andres 2006;Andres, Wills, and Griffith 2006).The Libyan case suggests that the local actor also can take opportunistic advantage of airstrikes, even in the absence of complete coordination.This can have important implications for the decision to employ aerial force, and the form of future operations.
Second, I show that theoretical and empirical disaggregation can improve our understanding of denial specifically, and military intervention in general.Existing micro-level studies of airpower utilize the causal logic of denial (Johnston and Sarbahi 2016;Newton and Tucker 2022), and recent cross-case studies of airpower highlights the importance of accounting for "mixed" denial and punishment strategies (Allen and Martinez Machain 2019), but the results presented here demonstrates the utility of differentiating the precise ways in which denial can be pursued.For example, by singling out strikes against anti-aircraft targets, I reaffirm recent cross-case evidence that air superiority has a more immediate impact on ground combat than commonly assumed (cf.Saunders and Souva 2020).Since governments fighting rebels have historically faced minimal aerial threats, the proliferation of cheap armed UAVs makes this finding relevant for future conflicts.Furthermore, while I could not find any effect of strikes against command and control assets in Libya, the results reported here should not be seen as an assessment of whether certain aspects of denial are "better" than others.Instead, this is a clear indication that novel insights can be gained from break down large, otherwise static categories of action.As another example, all four targeting approaches analyzed here constitute what Jones (2017) calls "direct-conventional" intervention, but substituting cross-case for within-case analysis reveals important variations in terms of effectiveness within an established category of military action.
Lastly, I provide some general insights regarding how coercive pressure is achieved in practice.Both airpower and military interventions seeks to affect wider bargaining processes, and conflict termination is drivenat least in partby the generation of salient battlefield information (recall Slantchev 2003;Jones 2017;Allen and Martinez Machain 2019).Airpower cannot seize tangible objectives on its own, but analyzing the ways in which foreign aerial force enabled the Libyan rebels to achieve a set of intermediary goalstaking government-held territoryillustrates how an aerial military intervention contributed to the generation of this information.As shown by Pechenkina and Thomas (2020), within-conflict outcomes such as changing territorial control can impact subsequent negotiating behavior.This highlights the importance of treating intermediary goals as the main object of analysis since they constitute salient conveyers of battlefield information.Understanding the ways in which an intervener systematically contributes to the factors that subsequently acts as drivers of negotiation behavior complements existing studies of both airpower and military interventions.This is also one way in which we can improve our understanding of how fighting affects bargaining processes, and thereby conflict termination.

Figure 2 .
Figure 2. Targets hit by type and district.
7A contiguous district is any district that shares a land border with another district.

Table 2 .
Logit estimates of rebel territorial captures.

Table 3 .
Temporal analysis of direct attack.