195
1 INTRODUCTION
Inrecentconflicts,themainexamplesofmajorports
subject to armed conflict are all exclusively located
withincountriesaffectedbycivilwarsresultingfrom
the collapse of regimes following the Arab Springs
(2011): Syria (Lattakia, Tartus), Libya (Benghazi,
Misrata, Al Khums, Tripoli) and Yemen (Aden, Al
Hudaydah, Al Mukalla). Among them, only Al
Hudaydah is directly located on a front line since
2018.ThisisasaresultoftheinterventioninYemen
byanArabcoalitionledbySaudiArabia(2015—…).
As far as we know, this is also the first container
terminal—i.e., specifically dedicated to
their
handling—toexperiencethisintriguingsituation.This
leads us to wonder about the spatial evolution
generated at the scale of the container terminal in
timesofconflictandwiththemeansavailableatfirst
sight:easilyoropenaccessdataandliteraturemostly
fromtheinternet.
Studying a port (or
a terminal) is an
interdisciplinary opportunity according to the last
paragraphs pointed by Ng et al. in the article Port
geographyatthecrossroadswithhumangeography:
between flows and spaces (Ng, Ducruet, Jacobs,
Monios, Notteboom, Rodrigue, Slack, Tam,
Wilmsmeier, 2014). Authors say: “They are locally
embeddedwithinaparticularregion,
andwillalways
be affected by the local socio‐cultural environment;
but simultaneously they are the major outlet for
external connections, and so will always be affected
by global spatial development. They are the
intersection and interaction points between people,
cargoflowsandmarkets.They arethearenaswhich
attract
substantial research interests from many
Assessing the Effects of a War on a Container Terminal:
Lessons from Al Hudaydah, Yemen
F.H.Guiziou
FrenchNationalCenterforScienticResearch(CNRS),Paris,France
ABSTRACT:ThispaperisfocusedonthecaseofthecontainerterminalofAlHudaydahport(Yemen)andon
its situation on the battlefront in a country struggling with a civil war since 2014, despite the military
interventionofaSaudi‐led
coalitionsince2015.ThegoalistoproposeacomprehensiveOSINTmethodologyto
evaluatethecaseofthisterminalwhichisthefirstcontainerterminalhardlyimpactedbyamodernconflict.
Thisstudybelongstothefieldofportgeographystudiesandaimstocontributetoabettercomprehensionof
transport and security/development nexus, build on an easy‐to‐use GIS methodology based on open access
data.Lessons learntfrom AlHudaydahcontainer terminalarea stepinthedescriptionofthe relations and
dynamics between war and terminals/ports. The main results show that war and battle have two different
effects:
warleadstolong‐termeffectsprobablylinkedtohinterlandattritionandchangeinporthierarchy;battle
lead to short/medium‐term effects linked to battlefront distance and road blockade. Finally, the port is a
valuable asset for both sides involved in the Yemeni Civil War, but it is difficult to
evaluate further the
importanceoftheterminalthanasignoftheportdynamics.
http://www.transnav.eu
the International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 18
Number 1
March 2024
DOI:10.12716/1001.18.01.20
196
disciplines, which facilitate intra‐ and
interdisciplinary” (Ng, Ducruet, Jacobs, Monios,
Notteboom,Rodrigue,Slack,Tam,Wilmsmeier,2014,
94–95). Therefore, ports, as crossroads and complex
objects, could be a new frame of reference to study
security anddevelopment nexus. AlHudaydah port
fulfills all the characteristics because of its status of
interface between the container global system and
Yemen Civil War complexity at local, regional and
internationalscales.Thatisalsoaborderlinecase:Al
Hudaydah port and container terminal are on the
battlefront sinceJune2018andthepoliticalsituation
seemsdeadlocked.
While this work islargely exploratorydueto the
absence of other comparable situations, two
questions/objectivesunderlieit:(1)One concerns the
dynamics and evaluation of the container terminal
during the Yemeni conflict. It is then possible to
questiontheroleofaterminalasanindicatorofthis
conflict; (2) The other raises the question of the
feasibility
ofananalysisofportgeographyusingthe
process and methods of open‐source intelligence
(OSINT). It means understanding the effects on the
terminalfromopenaccesssourcesbutalsotodevelop
a GIS methodology to overcome the lack of reliable
data in times of war or in ports outside the
main
globalmaritimesystem.
This paper focuses on the following sources, the
vastmajorityofwhichcomedirectlyfromtheinternet
and are open access: shipping data from the United
Nations Conference on Trade and Development
(UNCTAD);data fromThe Armed ConflictLocation
&EventDataProject(ACLED;Raleigh,2010);
public
portauthoritydatafromAdenPortAuthority;Google
Earth Pro satellite images (2004–2020); National
Geospatial Agency gazetteer; and a thick amount of
grey literature was processed (peer reviewed and
non‐peer reviewed scientific articles, press articles,
reports and UN documentation, websites,
photographs, etc.). For the development of
methodology in Section
5, the QGIS 3.16.7 software
(qgis.org) and the mmqgis addon
(michaelminn.com/linux/mmqgis/) were used. SRTM
andWorldPopdatawerealsousefultoassessYemeni
context.Aparticularattentionwaspaidtothequality
ofthedatausedandthemetadata.
The rest of the paper is structured as follows:
Section 2
draws the context of the Yemeni conflict
whichledtothebattleof AlHudaydah;Section 3is
dedicated to a critical analysis of Yemeni ports
dynamics according to UNCTAD data; creation of
new spatialized data at the scale of the container
terminal is treated in Section 4 and the analysis
is
proposedinSection 5;finally,Section 6serves asan
assessment.
2 THEYEMENICIVILWAR(2014—…)ANDTHE
BATTLEOFALHUDAYDAH(2018)
2.1 Thecontext
ThedifficultyinanalyzingtheYemeniconflictliesin
the multiplicity of local, regional and international
actors, but also opportunistic and
predatory local
actors that result from a socially and economically
fragilenationalcontext.Thefollowingtwopointsare
abriefreminderofthissituationandoftheconflictual
dynamics.
Yemen (pop. 27 million) does not have the same
stability as its neighbors, Saudi Arabia and the
Sultanate of Oman. Its strategic
location is ideal: it
borderstheRedSea,theGulfofAdenandtheBabel‐
Mandeb international strait. Yet it does not benefit
fromtheglobalshippingroutesbecauseofthelackof
alargetransshipmentportthatcancompetewith,for
example,theportsofJeddah(SaudiArabia)
orSalalah
(Oman).Theconstitutionofaunifiedstate(1990)led
toapoliticalandeconomicmodernizationattemptby
President Ali Abdullah Saleh’s regime. This
modernization did not prevent a cycle of conflicts
gradually stratifying over the past three decades:
nepotism, old divide between South and North,
tribalism and sectarianism, economic
difficulties,
water tensions, etc. (Burgat, 2006; Bonnefoy and Al
Rubaidi,2018;Clausen,2019,1–5).
These serious problems are worsened by the
strategies of neighboring regional actors as Saudi
Arabia, United Arab Emirates and Iran, that regard
Yemen as a security issue and by the war on terror
policy led by
the United States against Al Qaeda in
the Arabian Peninsula (AQAP) and Islamic State
(provinceofYemen).ThefallofPresidentSaleh(2012)
didnotfundamentallychangetheregimeandtherise
of the long‐suppressed rebel forces of Ansar Allah
(also called Houthis) from the Sa’dah region in the
north
ofthecountryreacheda criticallevel. In2014,
Houthis succeeded in taking control of the capital
Sana’a from the new president Abdrabbuh Mansur
Hadi’sforces.TheoffensiveoftheHouthisandtheir
allies undermined the presidential forces and
threatened the major cities of southern Yemen, Taiz
andAden.At
theendofMarch2015,anoperationled
by an Arab coalition from the Gulf Cooperation
Council (except Oman) intervened and quickly
federatedstatesfromtheArabworldandtheMiddle
East (Saudi Arabia, United Arab Emirates, Sudan,
Bahrain,Kuwait,Qatar,Egypt,Jordan,Morocco)with
anotableWesternsupport(Sharp,2020,
15).
2.2 InternationalizationofYemeniCivilWar
Due to the absolute air superiority of the Saudi‐led
coalition(SLC),inthedaysfollowingMarch25,2015,
mostofthemajormilitarytargets,includingballistic
missile sites, Houthis bases, and airports, were hit
hard,andcontinuouslybombeduntilApril21,2015.
ThisoperationwasthenreplacedbyRestoringHope,
supposedtoaimatrestoringsecurityforallYemenis.
Among the targets of the coalition forces were
numerous civilian facilities and infrastructures
(bridges, electrical equipment, pumps and dams). In
response to the strikes, the Houthis engaged in a
missilewar,targetingSauditerritory
andtheYemeni
maritime approaches (anti‐ship missiles but also
underwatermines).Themissilesanddronesusedby
the Houthis and supplied by the Iranians arrived
mainlyinsparepartsfromIran(WilliamsandShaikh,
2020; 29). These deliveries justified the blockade of
Yemeni ports, and then, progressively, the taking
of
Aden,MochaandalltheislandsinthesouthernRed
Sea(WilliamsandShaikh,2020;29–31).
197
On the ground, in addition to the Yemeni forces
supportedbythecoalitionforces,whichmakeupthe
majority of the forces facing the Houthis, the SLC
deployed heavy equipment (tanks, howitzers, anti‐
missilesbatteries),aswellastroops,mainlymadeup
of African auxiliaries. The front line of
the border
areasbetween Saudi Arabia andYemen, Ma’rib and
Adenstabilizedin2016andthesouthfrontlineruns
from Al Bayda in the east to the port of Mocha. In
June 2018,this lineis changedsignificantlywith the
recapture by government and coalition forces of a
narrowstrip
ofterritoryfromthesmallportofMocha
tothelargeportofAlHudaydah.InDecember2018,
thesituationwasfrozenonlandatAl Hudaydahand
atseawiththeSLCblockade.
Ataregionallevel,theSLCoperationconfirms:(1)
the will of Saudi Arabia to impose
itself as the
regionalleader capable ofguaranteeingpeace inthe
peninsula, the maritime spaces and the flows
borderingit(GuzanskiandShavit,2017);(2)thegreat
firmness of countering Iranian influence at the
Kingdom’sdoorstep.Thisinfluencetakestheformof
Iranian support to Houthis, a Shiite movement that
occupies
the region bordering Saudi Arabia but a
limited scale compared to Lebanon or Syria
(Guzansky, 2015). Iran is seizing the opportunities
providedbytheShiitemovementsintheArabworld
to increase its regional influence (Kam, 2016;
Guzansky,2012).TheconflictinYemenis,atthesame
time,a Yemeni
civilwar, a Saudi‐Iranian proxy war
andapointofinterestfortheUSandtheinternational
community(antiterrorism).Ithasentereditsseventh
year with no military outcome in sight. With each
passing day, the failure to re‐establish the
internationallyrecognizedgovernmentlegitimizesthe
SupremeCouncilofthe
Houthisandtheirallies.
TheSLChasbeenweakeningsincethesummerof
2019: (1) almost all Emirati forces are withdrawing
from the South in order to be redeployed to their
territory and turned directly towards Iran and the
Strait of Hormuz; (2) the American election of 2020,
won by
the Democrat Joe Biden, calls into question
thesupportandtheverynatureoftherelationshipof
trustwiththeSaudiregime(Sharp,2020,20–21).This
results in a blocked configuration and in a
humanitarian crisis: a large part of the civilian
population issubjected to the investors of war
(Clausen, 2019,
3) and to a disastrous situation
between health crises (cholera, dysentery and other
waterbornediseasestowhichwecannowaddSARS‐
COV2),hungerandlackofeverything(Varisco,2019,
325;17;MousaviandAnjomska,2020,1–2).
2.3 Thelastmajorbattle:AlHudaydah(June2018‐
december2018)
AlHudaydah
(pop.400,000),Yemen’smajorRedSea
port, is of vital importance as the gateway of a
transport corridor running to Sana’a and northern
Yemen.Before1994andthereunification,thePeople’s
Democratic Republic of Yemen, asocialist regime in
thesouth, andthe YemenArabRepublic, supported
bySaudiArabia,in
thenorth,hadeachaportforthe
twoformercapitals:Aden,forTaiz(pop.620,000)and
Al Hudaydah for Sana’a (pop. 2.5 million). This
political and commercial divide, which has never
beenerasedsince,hascreatedtwodefactocorridors.
It should be noted, however, that Aden has a
less
exclusive position due to its more diversified
hinterlandwithallthegovernoratesofSouthYemen
and up to Sana’a. The port of Al Hudaydah has a
hugedependenceonSana’a,whichislessreciprocal.
However, determining the boundaries of the
hinterlands is a challenge (De Langen, 2007;
Rodrigues,2020).
The
port and the city were regularly bombarded
bytheSLCforces,includingseveralbuildingsinthe
port and terminal (August 2015). In 2018, this last
majorportinthehandsoftheHouthisandtheirallies
became a first‐class objective and the target of an
offensive launched by Yemeni
government forces
supportedbytheSLCandEmiratigroundforcesfrom
Assab (Eritrea) and South Yemen. Moving up from
MochainJune2018,themainoffensivewaslaunched
onJune13,2018,afteranintensivebombingcampaign
ofthemaindefenselinesinthesouthofthecityand
thecity
itself.For7months,interspersedwithperiods
ofceasefire,thelastbattleofthecivilwartookplace
inthreephases(Fig.1):(1)June2018withthecapture
of the airport and defensive positions south of the
city; (2) September‐October 2018 withthe capture of
road number 3
towards Sana’a at the level of
kilometer16(Kilo‐16); (3)November‐December 2018
withtheattempttocompletelybesiegethecity.
The fortifications surrounded the city in several
lines with the main defense including the container
terminal.Itbecameamajorstakeintheconflictona
national and international
scale as well as the first
exampleofacontainerterminalstuckonafrontline.
Figure1.ThebattleofAlHudaydah(June‐December2018)
When, in November 2018 SLC backed forces
struggled to besiege the city, two political events
stopped their offensive: (1) the international
community denounced the murder of the Saudi
journalistJamalAhmadKhashoggibyofficialsofthe
Kingdom (October 2018); (2) the Yemeni
humanitarian drama put forward by the United
Nationsand
relayedby theinternationalmedia. The
international community pressured the coalition not
to cut the last major supply infrastructure for aid
198
delivery. On December 13, 2018, the Stockholm
Agreement (including theAgreement on the City of
Hodeidah)endedthebattle.AccordingtotheACLED
database,in the entirebattle, onlyone strike, by the
SLC,hitdirectlytheport,whichremainedlargelyout
ofthe clashes.The evacuationof the
portby Houthi
troops (May 2019) validated the application of the
Agreement but led to a frozen battlefront where
fightingandbombingremainepisodic.
3 TRENDSFROMUNCTADSTATISTICSAND
PROBLEMSATSUBNATIONAL/LOCALSCALE
3.1 AlHudaydahinUNCTADDatabase
Thedynamicsofaportduringawar,andevenmore
soduringabattle,canbeassessedbytheevolutionof
thenumberofshipsusingitaswellasbythetonnage
and quantities handled. However, statistics on the
port’susearenoteasilyavailableinthiscase,mainly
becauseoftheconflict.Inaddition,theproductionof
data
bythemajorinternationalorganizations(United
Nations agencies, World Bank, etc.) is often a
compilation/aggregationofdataproducedbyprivate
operators(consultancyfirms,intelligencecontractors,
etc.) whose collection methods are based on data
producedbyportcompaniesandoperatorsaswellas
onmoreorlesspreciseevaluations.Atthesame
time,
sites dedicated to collect AIS data (MarineTraffic,
Vesselfinder,etc.) produceanexceptionalamountof
information on port traffic and vessel typology, but
the acquisition of these studies is most often
unaffordable for preparatory work with limited
resources. Moreover, these data are declarative and
methodsexisttodistortthem(Boudehenn
etal.,2021).
Thus,assessingatrendattheglobalorregionalscale
is possible, but assessing the trajectory of
infrastructure at the state, subnational, and local
scalesischallenging.
Thissectionfocusesonmaritimedataprovidedby
the United Nations Conference on Trade and
Development (UNCTAD), one of the most
widely
used and valuable database in the transportation
geography and economics community. The statistics
onmaritimetransportarebrokendowninto16tables
of temporal data, with annual to quarterly
granularity, organized around groups of economies,
nationaleconomies,attheportandflaglevels,onthe
main types of traffic
(dry bulk, liquid bulk, RoRo,
containers, etc.), their tonnage (DWT) and port
performance.Amongthesetables,italsoproposesthe
LSCI (Liner shipping connectivity index), LSBCI
(Liner shipping bilateral connectivity index) and
PLSCI(Portlinershippingconnectivityindex)which
make it possible to establish the trends for the
international (LSCI and
LSBCI) and
national/international (PLSCI) port hierarchy and
their integration in the world maritime economy,
drivenbythecontainermarket.
Asaworkfocusedonthenationalandlocallevels
ofa containerterminal,thispapermostlylooksattwo
tables: the PLSCI (a quarterly index of nearly 900
ports handling
containers) and the Container port
throughput(annualdata).Thechoiceofanindexand
statisticsoncontainerswasmadeforthreereasons:(1)
containers are present in large numbers, easily
identifiable and sometimes countable through a
standardizedunit(TEU);(2)theyarekeyelementsof
maritimeglobalizationandofhinterlandstructuration
(Frémont, 2007, 15; Frémont and Soppé, 2005, 187;
Rodrigue, 2013, 66–67, 113–120; Guerrero, 2014, 92;
Guerrero and Rodrigue, 2012, 2–7); (3) while it is
difficulttoknowwhatiscontainedinashed,asiloor
a container (from the sky), the latter has the
advantage of being mobile, small
size, multimodal
and each unit can be considered as an intermediary
storage unit(halfway between the goods in and the
stackorshipwhereitisstored).Thelargenumberof
observablecontainersandclusterscanbeestimatedas
asufficientquantityforastudyoftheevolutionofa
container
terminal (with the limitations presentedin
Section5).
3.2 PLSCIandContainerportthroughput
It istherefore possibleto compare the trajectories of
the three main Yemeni ports (Fig. 2): (1) for Al
Hudaydah, the PLSCI indicates a significant
downwardtrendovertheperiod2006–2020interms
of its maritime
integration into the global
containerizedsystem. Thisis nolonger calculableas
of2018.Theestablishmentofthefrontlinenexttothe
port that same year is undoubtedly the cause: the
terminal is emptying; regular lines are disappearing
andcontainershipsnolongertaketherisktousethe
port.
The terminal where all container traffic is
concentratedis losingitsfunctionastheheadof the
Sanaʹa corridor. This downwardtrend could also be
linkedtocompetitionwithAden,whoseportfacilities
aremoredeveloped,efficientandlessimpactedbythe
war.(2)For Aden,there isno
visibletrend overthe
same period. This is probably due to Adenʹs less
exclusivehinterlandanditsstrongdependenceonthe
securitycontext.Thisisverifiedinrecenthistorywith
the inability of Aden to become a transshipment
platform and the strong subsequent unrest of the
Arab Spring such as
the multiplying number of
political actors and groups in Yemeni politics. The
greatvariationintrafficalsodependsontheterrorism
threatsonalongertime(Guiziou,2014,208–209).(3)
Al Mukalla, a general‐purpose port away from the
main fronts and dedicated to the eastern region of
Yemen (Hadramawt)
cannot compete with the two
others. The PLSCI thus reflects trends which seem
possibleforYemeniportstrajectorieswhencompared
tothehinterlandsandthesecuritycontext.
Figure2.Portlinershippingconnectivityindex(PLSCI)for
thethreemajorportsinYemen
199
The Container port throughput analysis, a
compilation of data from port authorities and data
providers (Lloydʹs List Intelligence in particular), is
more critical because its limitations are more easily
identifiable (Fig. 3). Subtracting the Aden Port
Authority (APA) data from the Container port
throughput, we see that over
the period 2015–2018,
thedataforportsotherthanAdenareexactly200,000
TEU annually. Most of the container traffic of these
portsishandledbyAlHudaydahwhich,alongwith
Aden, is the only one equipped with cranes and
adequateequipmentforthemassificationofcontainer
traffic. Thus, there are
at least 4 years where the
activity of the AHCT is a virtual estimation. In
additiontothebiasthatthisimplies,thedistribution
between Al Hudaydah and other small ports
(including Al Mukalla where the imagery shows a
modest container activity) remains problematic: the
main assumption is the arbitrary
allocation for the
periodof2015–2018forfiguresinthecontinuityofthe
year2014(200,003 containers).Fortheyear2018,for
example, the imagery shows the AHCT empty.
Therefore, the assessment of 200,000 containers
outsideAdenisobviouslynotrealisticasAlMukalla
can in no way absorb several hundred
thousand
containersduetolackofequipment,storagecapacity
andsufficienthinterlanddepth.However,thefigures
for2019seemmorerealisticwiththereturnofsmall
numbersofcontainerstoAlHudaydahandavisible
developmentoftrafficinAlMukalla.
Figure3.UNCTADStatisticsTEUhandledinYemeniports
(2010–2019)comparedtoAPAstatistics
Having made these observations, the question is
howtogetaroundtheobstacleofthelackofdatafor
theperiod 2015–2018 inorder tohave, ifnot precise
data,atleastanideaofthetrendsincontainertraffic
withintheport.Theanswermaylieinthefollowing
methodology,
whichdoesnotcountTEUsbutfocuses
onspaceoccupationbyTEUs.
4 ASSESSINGTHETRENDSOFACONTAINER
TERMINAL:STACKOCCUPATION
METHODOLOGY
4.1 Introductiontothemethodology
Assessingthe evolution of thenumber of containers
handled could be challenging, even with a broad
approachandthatisthecase
inthisexample.Remote
satelliteimagery,nowmoreandmoreavailable,may
helpbecausethecontainerisastandardunitwhichis
countable and a metric for logistic performances at
local and global scale. A lorry is generally carrying
one or two TEUs, a ship can carry hundreds up to
several
thousands and a terminal can store up to
hundredsofthousandsonitsstacks.
This standardization should help to evaluate the
number of TEU a terminal handle/store, but the
realityofaclassiccontainerterminaleverydaylifeis
muchmorecomplex(Dekker,2006):
1. Containerscouldbestackedonone
leveltoseveral
levels.Evaluatefromapicturethesizeofapileisa
complex operation, because of the quality of
photos,shadows,hugevariationoflevelsfromone
tiertoanother,etc.;
2. Containerscouldbeloadedoremptywhichisnot
seeablefromremotesensors/photosbecauseofthe
nature of the totally closed box. Administrative
files, precise localization on the stack and the
number of each container are barely available
outside companies and terminal authorities and
even then, it is virtually impossible to have a
preciseideaofthetonnageandthevalue;
3. Containers are logistic objects
which move but
could also be stored for a long time at the same
place. The best example of this mobility is
articulated in the terminal, where the connection
between ship and lorries is particularly visible
before and after the port of call and
loading/unloading operations. Even in the
terminal, a
container may move several times
between stacks (stacking management) or stay
storedforalongtime.
These principal limits are problems for assessing
the evolution of number of containers, especially in
countries when stacking strategies are based on
poor/mediummanagement.Themainchallengehere
is to create data as reliable
as possible via a
standardized spatial approach of binary
occupied/non‐occupied grid cells. This stacking
occupation methodology is then indirect: the goal is
nottoaccountTEUbutcellsoccupiedbycontainers.
There is a potential link between TEU number and
occupiedcellsateachdatebutitisnotdeveloped
in
thispaper.
The methodology is based on available Google
Earth data from 2004 to 2020 and treatment by GIS
trying to show the main dynamics during AHCT
experienced war, a difficult situation on the
battlefrontandanapparentlydecreasingtraffic.This
approachisbasedonfollowingfactsandestimations:
1. Morethan3leveltiersseemtoberareaccordingto
shadowsandlackofstackingcranes,andnomore
2leveltiersarevisibleatthestartof2018;
2. Obviously, the AHCT is not a first‐class terminal:
stacking strategy and management are allegedly,
at best, medium. The
containers are handled by
lorries and with the help of four quay cranes
(gantrycranes),atleastfourreachstackers,anda
dozen of straddle carriers. The storage of these
machines was bombed in 2016, but most of the
equipmentseemsstilloperable.Twogantrycranes
werealsodamagedby
bombingin2016andareno
longerusable;
3. Containersarestackedinmanyplacesintheport,
broaderthantheonlyasphaltedterminal.Thisfact
probablyshowslacksinthelogisticchains(taxes,
corruption,lackofdemandduringthewar,goods
atthedestinationofoccupiedterritories,difficulty
tomanage
emptycontainers,etc.).
Notethatthemainlimitationofthismethodisthat
itreliesonsatellitedatathatisnotcontinuousintime.
200
Theportofcallsofcontainershipsarealsodistributed
in a regular way but not necessarily cyclic and
obviously notcoordinated with theimages.Another
limit is that containers are black boxes, but their
dynamicsandspatial distribution(clusters)could be
an indication of the activity of the
terminal (or the
port).
4.2 Implementation
The methodology applied here consists in precisely
evaluating the occupation rate of each stack in the
terminal (10 entities) and informal “stacks”
surrounding(5entities).Thethreecategories[M],[M’]
and [m] are applied to describe a comprehensive
approach of port dynamics linked with containers
handledinaformalmanagedareaornot.Theseareas
were estimated from satellite images without any
preconceivednotions oftheir dynamics. We canjust
estimate that containers in [m] areas are empty, in
long transit, or/and less important than the ones in
[M] and [M’] areas because of their
poor staking
conditions. Areas are defined by these factors: (1)
containerdensity;(2)geometricdisposition;(3)visible
use of reach stackers and straddle carriers; (4)
asphaltedareas/non‐asphaltedareasandroads/alleys
between.Inaddition,wealsodefinedthetwomajor
stacks[M]intheimmediate vicinityofthe container
berths. Finally,
we can estimatewhat are the AHCT
limits and the informal areas because of the lack of
space.Theelementsofthisclassificationarepresented
inthemap(Fig.4)behind.
Figure4.PortofAlHudaydah(2018)and[M],[M’]and[m]
stackingareas
Basically,theaccountingmethodologyappliedfor
[M],[M’]and[m]areasarethesame:
1. it needs 2 TEU per 12m×12 m grid cells to be
considered as an occupied cell. This choice of
12m×12misdefinedbythedimensionsoftheTEU:
1TEU=6.1×2.44m;2TEU(most
ofthecontainers)
=12m×2.44m;
2. intheory,eachgridcellcouldcontain,ononelevel,
around 10 TEU. In some cases, due to the
photographicquality (dust,clouds, shadows)and
contestable visual counting, the choice has been
madenottovalidatesomeoccupiedgridcells.
Thus, the visual accounting
has been made with
therulespresentedinthefollowingfigure(Fig.5).
The principal goal is to identify, with a
standardized methodology, changes in container
occupationofeachcellandclustersofoccupiedcells
to express these changes as a percentage for each
stacking area (percentage expressed with respect to
themaximumoccupationontheperiodfortheentire
terminal: 1,250 cells on 21 September 2014) and the
temporal evolution of this occupation rate. This
methodology was used to analyze of 120 photos on
the 2012–2020 period, by visual classification. This
classification has been double‐checked. It assumed,
despitethe
limitsseenbefore,thattheanalysisoffersa
realistic view or, at least, reliable tendencies. The
resultsarepresentedinthefollowingparagraphs.
Figure5.Stackingareasandgridcellsvisualclassifications
5 RESULTSANDANALYSIS
5.1 Long‐termanalysis(2012–2020)
Themainquestionhereis tocompareourdatawith
the trends identified with the PLSCI. The general
trend ([Total], see Fig. 6) of the strong decline of
AHCT seems to be largely confirmed on the period
(R²=0.84): the port seems to
be under pressure from
thewar,especiallysince2018andevenmoreduring
the battle. We could also observe the existence of a
small growth of occupied cells which incarnates the
return of a small container activity, confirmed by
imagerywherewecanobservesomefeedersback.But
thisgeneral
trendneedstobedecomposedatthescale
of [M], [M’] and [m] areas to establish fine‐grained
trends.Thus,wecannotetwospatialdynamics:
1. [M] and [M’] are in a constant decline from 2016
and[M],typicallythebusiestareaonthepre‐battle
period(andintheory
foragoodstackingstrategy),
wastotallynon‐occupiedshortlyafterthestartof
the last phase of the battle (September 2018). A
linear regression between these two areas shows
that there is a strong spatial link between them
(R²=0.88): when[M]isoccupied,[M’] isgenerally
occupiedtoo.It
showsthatthelimitsoftheformal
terminal are quite well assessed but that, during
the battle more than during the war, there is no
activity and probably no traffic at all. The closer
201
thebattleis(andthebattlefront),themorereduced
is the traffic is. Activity was back in mid‐2019,
whichisroughlythedateoftheapplicationofthe
Stockholm Agreement. It is then very likely that
thepressureofthebattleisakeyfactor;
2. [m]is
moredifficulttoanalyzebecauseofitsnon‐
formal and messy nature. The link between the
terminal([M]+[M’])and[m]islesswellestablished
(R²=0.65) and many containers, particularly those
indistinctclustersofprobablyemptyones,canbe
stacked there for a long time. The year 2018 is a
turning
point for [m], where the percentage of
occupiedcellsdropsdrasticallyfrom35%to15%a
fewmonthsbeforethebattleandthendeclinesto
itslowestlevelofaround10%untilthemiddleof
2020.Thisispossiblyduetothecombinedeffects
of the classic dynamics of empty
containers, the
urgent need of containers to maintain a trade by
lorry and the fossilization of old and damaged
containers.Thesatellitephotosof[m]areasshow,
until the end of 2020, numerous stationary
containersandclusters.Ahypothesisistoconsider
thatdropnotonlyasamajorchangein
theactivity
of the terminal but also as a possible
announcement of the upcoming battle by well‐
informedlocaloperators.
Figure6.[Total],[M],[M’]and[m]percentagesofoccupied
gridcellsincomparisontothemaximum,2012–2020
Thislong‐term analysishas tobe completedbya
pre/post‐battle study, focused on the 2018 year and
thesurroundingmonths.
5.2 Pre‐Battle/Post‐battleanalysis
Focusingonthefollowingheatmapandmaps(Fig.7)
of thecoefficient of determination(R²) from alinear
regression(linearfunctionf(x)=ax+b;R²
=corr(x,y)²)
betweenoccupationrateofstacksontheperiod2012–
2020 (based on the maximum 100% rate for the
terminal on 2021 September 21), it is possible to
determinesomecharacteristicsoftheterminalandits
surroundingspaces.Theselinearregressionscouldbe
interpretedasa spatialdependence(R²tends
to1)or
independence (R² tends to 0). This materializes
adequationtothemodelf(x)=ax+band,therefore,the
strengthofthelinkbetweentwostacksontheperiod.
Itisnotalinkbetweenthenumberofcontainersand
even less container mobility stack to stack, but a
spatial indicator
based on the occupation by
containers in spatially localized stacks (difference
between correlation and causality). For instance, the
linkbetween[M1]and[M2](R²=0.59),explainedthat
thevariationofoccupiedcellsbycontainersof[M2]is
linked for 59% of the variation of occupied cells of
[M1] according to the linear
model. This doesn’t
describethefullnatureofthisrelation.Thefollowing
observations,inadditiontophotoobservations,could
describetheterminalstructuration:
1. Generally,[M]and[M’] arequitewell linked,but
withavariationofR²(67%>R²=0.5/49%>R²=0.6/
13%>R²=0.7).Italsoseemsto
haveastrongerlink
betweenstacksinthenorthernareaoftheterminal
([M1], [M’1], [M’2], [M’3]) and, on the other,
between stacks inthe southernarea ([M2], [M’4],
[M’6],[M’7],[M’8]);
2. Generally, [m] stacks are poorly linked both with
[M] stacks and other [m]stacks which shows the
random
andnon‐formalnatureof[m]stacks;
3. [M1] and [M2] (R²= 0.59) are moderately linked
because there are two distinct berths, and could
alsoreinforcetheobservation(a1);
4. [M’5](generallyR² around0.4with [M]and[M’],
around0.1‐0.2with[m])hasaninterestingposition
andseems
tobelargelyindependentandcouldbe
aspecific/privatestack;
5. [M’7] and [M’8] (R²=0.78) are strongly linked
probably because these stacks are dedicated to
reefers (fridge containers) and need electricity
connectionstobeoperable;
6. [m2] and [m5] (R²=0.80) are strongly linked
probablybecauseofthevicinityoftheexit
roadof
theterminal;
7. Note that inall cases weobserve that a(slope) is
positive (so the models show a positive relation
betweenstacks)andbisoftencloseto0.
Figure7. Coefficient of determination heatmap (linear
regression)andspatialdistribution,2012–2020
Usingtheseobservations,it ispossibletoadvance
thatthereisnothingtoodifferentbetweentheAHCT
202
and other container terminals: specialized areas are
dedicated to different types of containers (a4, a5),
thereisprobablyquiteadecentstackingstrategy(a1,
a5,a6)andthecontainerberthsarewellidentifiedby
theirrelation withstacks(a1,a3).Moreover(a1)and
(a2) confirm that assessed limits of
the formal
terminalareagainwellidentified,evenif(a4)asksthe
questionof thelocalizationand situationof [M’5]as
partornotoftherealterminal.
Wecannotunfortunatelygofurtheronthepre‐war
period because of lack of data (only 11 photos
between 2012–2014), but we can
focus on the pre‐
battle and battle/post‐battle period. The same work
wasdonewiththetwoperiodsApril2016‐May2018
(pre‐battle) and June 2018‐July 2020 (battle/post‐
battle), as presented inthe followingheatmaps (Fig.
8):
1. during the pre‐battle period, the model described
before is
efficient in spite of war. The only
difference is the high spatial independence
between[M&M’]and[m],exceptforthe[m2]and
[m5]asseenin(a6).Thisindependenceisprobably
linkedtowarandlackofcontainertraffic(noneed
tousethe[m]areas);
2. during the battle/post‐battle
period, a very high
independence is now seeable for near all the
couplesofstacks.[M1]and[m5](R²=0.63)arethe
only exception but it is certainly due to the
fossilizationof[m5]andtheemptinessof[M1].
This change could be related to some factors: (1)
strong diminution (and
disappearance during the
battle) of port calls and container traffic; (2)
modificationsofthestackingstrategywhichtendsto
become more and more daily/messy because of the
battle situation and the troubles after the battle
(episodicfightsandshelling);(3)spaceintheterminal
whichcanleadtoamorepermissive
stakingstrategy;
(4)useofonlyoneofthetwocontainerberthsbecause
ofthediminutionofportcallsandthehighersecurity
andproximityofberth6(Fig.5)andendoftheuseof
gantry cranes replaced by on‐board cranes (Lift‐
On/Lift‐OfforLoLo)fora
smallcontainertraffic,etc.
The battle is a turning point in the history of the
AHCT because of the fights around the ports and
perturbationslinkedtothem,butalsobecauseofthe
changesinthetrafficdynamicsatalllevels:localwith
alesseffectiveterminal;national,withthechange
of
port hierarchy; international with the changes of
servicestotheport.Thesedifferentchangescouldbe
problematiciftheireffectsshouldlastinthelongrun.
Figure8. Coefficient of determination heatmaps (linear
regression)2012–2020,April2016‐May2018,June2018‐July
2020
Thetwoanalysesofdataproposedintheprevious
methodology could explain that the decline of the
terminalisatwo‐levelphenomenon:thewarandthe
hinterland attrition (Sana’a and Sa’dah) moreover
withtheSLCmaritimeblockade,isalong‐termeffect;
thebattleandtheestablishmentofthebattlefront
are
a middle to short‐term effect, as it is difficult to
managetheterminalbecauseoffightsinthevicinity
andthelackofaccessibilitytooutlets becauseofthe
closebattlefront.The combinationof both ishard to
overtakedespitesomesignsofrecoveryofportcalls
and
returnofcontainertraffic.
5.3 Fortification,portmanagementandstrategicvalue
Theselastparagraphsbringsomefacts(fromsatellite
photos) and a hypothesis to put further the debate
and the analysis. A first fact is that the large
fortification of the port and the AHCT. As a line of
defense of
the city, the port fortification started in
May 2018, few weeks before the battle. Starting
trenches and sand walls, the defenses were fast
reinforced during the battle by a lot of containers,
both for roadblocks and heavy walls (at least 400
TEU).Containersplayedabigroleintheimpossibility
for SLC‐backed troops to take Al Hudaydah. This
atypicalfunctionof containersshows,first,the large
number of fossilized containers easy to use for
defensivepurposeand,second,thevalueoftheport
andterminal,bothintermsofdefenseandmaybeasa
strategic object. At the end of
2020, even after the
withdrawal of Houthis from the port, fortifications
were still standing. Another important fact is the
restartofadecentportmanagementjustmonthsafter
the battle. The two damaged gantry cranes were
deconstructedbetweenMay2019andSeptember2019
andstoredclosedtotheterminal.The
tworemaining
cranesseemtobeinoperable,buttheLoLoshipsare
backatberth6andalittletrafficisgoingon.Evenin
poorconditions,withthelackofoilandelectricity,as
wellasthe additionaldifficultyto paytheport staff
(due to conflict between the two
branches of the
Central Bank of Yemen), the terminal and the port
had been restarting an activity with feeders and a
lorrytraffictoSanaabythenorthernroad.
This raises questions about the future of AHCT
andthestrategicvalueofacontainerterminal.Forthe
future, the answer
will probably depend on four
factors:thepoliticalsolutiontotheYemeniCivilWar;
thechangesinthehinterland limits;theconcurrence
with Aden and the place of AHCT in the container
trade hierarchy; the confidence of companies and
level of prices to visit the terminal (particularly
additional costsas insurances,
e.g., risk of war). For
the strategic value of a container terminal, the
question is delicate and could be seen from several
frames of reference or scales. It is impossible to
generalize:theabsenceofcomparisonsdoesnotallow
ustodeviatefromthiscasestudyandwilltherefore
limit
us to making assumptions. A port is always a
strategicasset.The maritimeblockade aswell asthe
takeofYemeniportsbySLC‐backedtroopsillustrate
that well. At the terminal level, it is difficult to
evaluate this strategic importance. Blockade and
hinterland attrition undeniably affect the container
traffic in
the terminal but, on the other hand, it is
difficult to assessthe complex local geopolitical and
economic situation because of a lack of sources.
Moreover, with gantry cranes non‐operational and
trafficlimitedtoLoLoandRoRoships,anypartofa
port could be considered as a container
terminal if
203
there is enough space and depth. Another good
argument could be the low level of destruction and
damagesduringthebattle.
However,wecanconsidertheterminalaspartof
theportwhichisastrongpoliticalcardforbothSLC
andtheHouthisinaprocessofa
politicalnegotiation
towardtheinternationalcommunity.Houthisexploit
the vital importance of the port for the population.
SLC says they bring pacification and aid to all the
Yemenis by ousting the Houthis and attempting to
avoid missile attacks from them. Is this terminal an
indicatoroftheYemeniCivilWar?We
thinkthatthe
answer is a (weak) yes: (1) because of the trends
which could show an evaluation of hinterland
attrition;(2)becausethebattlefrontestablishedonthe
terminalmarksthelastmajorbattleofthewarandthe
blocked situation which continues until today; (3)
because it is part
of the port from which the attack
triggered international intervention to limit the SLC
operations.But,intheendandinastrategicframe,it
isdifficulttoseparatetheterminalfromtheportand
wecanonlyarguethatthecontainerterminalactivity
couldbeastrongindicationofport
dynamics.
6 CONCLUSIONS
This paper proposes an unprecedented case of a
container terminal struggling with a conflict at
nationalscale(civilwarandSaudi‐ledinterventionin
Yemen)andatlocalscale(battlefront).Theeffectsof
thishighintensityarmedconflictare:(1)theconstant
degradationoftheterminalactivity
whichresultsina
near complete disappearance of container shipsport
ofcallsandTEUtrafficinthefirstmonthsof2019;(2)
achangeinYemenporthierarchyand(3)maybealso
a change in historical hinterlands. The effects of the
warontheterminalarelong‐termeffects(attrition
of
the hinterland, maritime blockade) but the effects of
the battle are stronger (distance to the battlefront,
roadblockadeandfortificationoftheterminal).This
is not a surprise and these two temporalities could
drastically change the status of the AHCT. In
addition,thefortificationof theportandthe
limited
damages during the battle are signs of a strategic
interest of both factions confronted to a blocked
battlefront and situation. This interest is probably
more politic than operational. A light recovery is
noticeable both in port management and in traffic
since mid‐2019 but this trend is difficult to evaluate
andthelong‐termeffectsareunpredictable.
ThepaperalsopointsouttheinterestofanOSINT
methodology, basically the multiplication of sources
and sub‐methodologies, to assess the situation of a
port in a lacking data context. The exploitation of
open access data from varied actors/producers
permitstounderstand
thesituationatmanylevelsof
a terminal in time‐related, geographical and
conflictual approach. An effective and diversified
OSINTstudyisagoodfirststeptoassessasituation
before putting more efforts (e.g., time and money).
There is no surprise mainly because of the
interdisciplinarynatureof portsstudies
as described
byNgetal.(2014).
The GIS methodology proposed in this paper,
based on a grid cell occupation of stacks by
containers,isanindirectspaceandtime‐relateduseful
process.Obviously,itneedstoberefined,reproduced
and developed, but it offers, at least in this case,
reliabletrends,which show theterminal,iftheport,
dynamics. From totally open access data, this new
databaseallowstocompletetheonesfrompublicand
privatedataproviders.
Duringaconflictorwhendataislacking,assessing
the dynamics of a terminal or a port is also a key
factor
in modern logistic operations, for instance to
identifyif itcould manage theinternationalaid and
itsabilityandcapacitytoconnectwithitshinterland.
Itwillalsobeanadvantagefortheevaluationofthe
dynamics of container traffic at the range and
hinterland scales as proposed by Rodrigue
(2020).
This methodology is also a first step for a most
sophisticated tool or model. The limits of it are the
availabilityand qualityof primarydata,and, inthis
case,mainlythesatelliteimagery.Moreover,itisnot
possibleto compare atthescaleof TEU,but onlyto
see
thespatialchanges.
Finally, the case of Al Hudaydah container
terminal,asabrand‐newsituation,ishardtodescribe
andtoanswermorethanthefacts,interrogationsand
hypothesis presented before. However, this terminal
could be considered as an indicator in the Yemeni
CivilWarandshowthehigh
economicandstrategic
importanceofcontainersfromtheglobalscale(global
maritimesystem)tothelocalscale(terminal),evenin
armed conflicts and their local development. Maybe
another evidence that containers are a paradigm of
globalizationevenatlocalscale.
REFERENCES
[1] “Aden PortAuthority, Statistics,” portofaden.net., 2021
[Online]. Available:
http://www.portofaden.net/en//statistics/list
[2] S. Brom, Y. Guzansky, “The conflict in Yemen: a case
study of Iran’s limited power,” Institute for National
Security Studies, 747, 2015. [Online]. Available:
https://www.inss.org.il/publication/the‐conflict‐in‐
yemen‐a‐case‐study‐of‐irans‐limited‐power/
[3] L. Bonnefoy, “Les
faux‐semblants de la guerre au
Yémen,”(inFrench),Esprit,vol.9,no.9,pp.7‐10,2016.
https://doi.org/10.3917/espri.1609.0007
[4]C.Boudehennetal.,“Navigationanomalydetection:an
added‐valueformaritimecybersituationalawareness,”
Cyber Science 2021, Ju. 2021.
DOI:10.1109/CyberSA52016.2021.9478189
[5]F.Burgat,“LeYémenaprèsle11septembre2001
:entre
construction de l’État et rétrécissement du champ
politique,“(inFrench),Critiqueinternationale,vol.3,no.
3,pp.9‐21,2006.https://doi.org/10.3917/crii.032.0009
[6]M.‐L.Clausen,“AchievingasustainablepeaceinYemen
lieswithlocallydriveninitiatives,”DanishInstitutefor
International Studies, 2019. [Online]. Available:
https://pure.diis.dk/ws/files/2971161/DIIS_Policy_brief_
Yemen.pdf
[7] P.W. De
Langen, “Port competition and selection in
contestable hinterlands: the case of Austria,” European
JournalofTransportandInfrastructureResearch,vol.7,
no. 1, pp. 1‐14, 2007.
https://doi.org/10.18757/ejtir.2007.7.1.3370
[8] R. Dekker, P. Voogd, E. van Asperen, “Advanced
methodsforcontainerstacking,”ORSpectrum,vol.28,
204
pp. 563‐586, 2006. https://doi.org/10.1007/s00291‐006‐
0038‐3
[9]A.Frémont,M. Soppé,“Transportmaritimeconteneurisé
etmondialisation,“(inFrench),AnnalesdeGéographie,
vol. 114, no. 642, pp. 187‐200, 2005.
https://doi.org/10.3406/geo.2005.21432
[10] A. Frémont, “Global maritime networks: the case of
Maersk,”JournalofTransportGeography,vol.15,no.6,
pp.
431‐442, 2007.
https://doi.org/10.1016/j.jtrangeo.2007.01.005
[11]GoogleEarthPro.(2021),Google.Accessed(June,2022).
Available:
https://www.google.com/earth/versions/#earth‐pro
[12] D. Guerrero and J.‐P. Rodrigue, ”The waves of
containerization: shifts in global maritime
transportation,” Journal of Transport Geography, vol.
34, pp. 31‐43, 2014.
https://doi.org/10.1016/j.jtrangeo.2013.12.003
[13] D. Guerrero, “Deep‐sea hinterlands:
some empirical
evidence of the spatial impact of containerization,”
JournalofTransportGeography,vol.35,pp.84‐94,2014.
https://doi.org/10.1016/j.jtrangeo.2014.01.010
[14] F. Guiziou,ʺLa piraterie somalienne : innovation,
masque du désordreet mondialisation,ʺ (in French), in
Lapiraterieau fil del’histoire:undéfipour l’État,M.
Battesti,Ed.,
Paris:PUPS,pp.197‐214,2014.
[15] Y. Guzanski and E. Shavit, “Yemen after Saleh:
Microcosm of a Regional Struggle,” Institute for
NationalSecurityStudies,998,2017.[Online].Available:
https://www.inss.org.il/publication/yemen‐saleh‐
microcosm‐regional‐struggle/
[16]Y.Guzansky,“Yemen:betweenIran,al‐Qaedaandthe
West,”InstituteforNationalSecurityStudies,350,
2012.
[Online]. Available:
https://www.inss.org.il/publication/yemen‐between‐
iran‐al‐qaeda‐and‐the‐west/
[17]E.Kam,“Iran:navalbasesinSyriaandinYemenKam,”
Institute for National Security Studies, 879, 2016.
[Online]. Available:
https://www.inss.org.il/publication/iran‐naval‐bases‐
syria‐yemen/
[18]S.M.Mousavi,M.Anjomshoa,“COVID‐19inYemen:a
crisis within crises.”,
International Journal of Equity
Health, vol. 19, no 1, 2020.
https://doi.org/10.1186/s12939‐020‐01231‐2
[19] “Shuttle Radar Topography Mission,” Distributed by
NASA, 2021. [Online]. Available:
https://www2.jpl.nasa.gov/srtm/
[20] “NGA GEOnet Names Server,” 2021. Distributed by
National Geospatial Agency, 2021. [Online]. Available:
https://geonames.nga.mil/gns/html/
[21]A.K.Y.Ngetal.,“Portgeographyatthecrossroads
with
humangeography: betweenflowsandspaces,”Journal
of Transport Geography, vol. 41, pp. 84‐96, 2014
https://doi.org/10.1016/j.jtrangeo.2014.08.012
[22] H. Grundberg, Stokholm Agreement, Office of the
SpecialEnvoyoftheSecretary‐GeneralforYemen,2018.
[Online]. Available: https://osesgy.unmissions.org/full‐
text‐stockholm‐agreement
[23] J.‐P.Rodrigue, C. Comtois, B. Slack.The geography
of
transportsystems.London,Routledge,2013.
[24] J.‐P. Rodrigue, “The geography of maritime ranges:
interfacing global maritime shipping networks with
hinterlands,” Geojournal, 87, pp. 1231‐1244, 2020.
https://doi.org/10.1007/s10708‐020‐10308‐y
[25] J. M. Sharp, 2020. “Yemen Civil War and regional
intervention,” Congressional Research Service,
Washington, R43960. [Online]. Available:
https://crsreports.congress.gov/product/pdf/R/R43960/38
[26] Yemen data, The Armed Conflict Location & Event
Data Project, 2021. [Online]. Available:
https://acleddata.com/
[27] Maritime transport, United Nations Conference on
TradeandDevelopmentdata,UNCTAD,2021.[Online].
Available:https://unctadstat.unctad.org/
[28]D.Varisco,“PumpingYemendry:ahistoryofYemen’s
watercrisis,”HumanEcology,vol.47,pp.317‐
329,2019.
https://doi.org/10.1007/s10745‐019‐0070‐y
[29] I. Williams, S. Shaikh, “The Missile War in Yemen,”
Center for Strategic and International Studies,
Washington, 2020. [Online]. Available:
https://www.csis.org/analysis/missile‐war‐yemen‐1
[30]WorldPop,DistributedbyUniversityof Southampton,
2021.[Online].Available:https://www.worldpop.org/