Table 1.

Characteristic MHW drivers from literature assessment

		Timescale
		Synoptic		Seasonal to intraseasonal		Interannual		Decadal
Typology	Case study	Mode/ teleconnection	Local process	Mode/ teleconnection	Local process	Mode/ teleconnection	Local process	Mode/ teleconnection	Local process
EBC	Benguela			ABF, RWS, KWO, MJO1	ADV, ASHF1	RWS, KWO1	ADV, VP1
	Leeuwin			RASC, SLP(−), LWS1	ADV, EHF, ASHF, VP1	ENSO(−), SLP, RWS1	ADV, ASHF1	PDO(−), ENSO1	ASHF1
	Baja California			LWS2	ADV, ASHF, VP1	ENSO(+), RWS, SLP(−)2	ASHF, VP, ADV1
	Iberian/Canary	AB4	ASHF4	NAO(−), RASC, RWS4	ADV, ASHF4	JS4	ASHF4
	Humboldt/Peru			KWO, RWS2	VP, ADV, ASHF2	ENSO(+), RWS2	ADV, ASHF2
WBC	*Gulf Stream	JS5	ASHF, ADV, EHF5	JS, NAO(+)5	ASHF, ADV, EHF, VP5			AMO5	ADV, EHF5
	* Kuroshio	RASC5	ASHF5			ENSO, RWO5	ADV (fronts), EHF5	PDO, PNA, AL, RWA, RWS5	ASHF5
	* Brazil-Malvinas Confluence	RASC, RWS, SLP5	ASHF, ADV5	RASC, SLP5	ASHF, ADV, EHF5
	* Agulhas			RASC, SLP, RWS5	ADV, VP, EHF5	ENSO(−), IOD(−)5	ASHF, ADV, VP5	PDO(-)5	ASHF5
	Agulhas Retroflection			RWS, SLP5	ADV5
	* East Australian Current			RASC3	ADV, VP, EHF3	ENSO(+), CPEN3	ADV, EHF, VP3	PDO/IPO3	ASHF, ADV, EHF3
	East Australian Current Extension			BI5	ADV, EHF5	ENSO3	ADV, ASHF3
Tropics	Great Barrier Reef	AB, RASC, LWS3	ASHF, ADV3			ENSO(+), CPEN1	VP, ADV, ASHF1	PDO/IPO, RWS1	ASHF, EHF1
	Seychelles Is.			RASC, SLP, RWS3	ADV, VP1	IOD(+)2	ASHF, ADV, EHF1
	Galápagos Is.			RWS1	VP1	ENSO( + ), RWS1	VP, ADV, ASHF1	PDO(+)4	ASHF, TD2
	Bay of Bengal	RASC, SLP4	ASHF, TM4	RASC, ASM4	ASHF, ADV4	IOD4	ASHF4
	Caribbean Sea					ENSO(+)5	ASHF5
MHL	Mediterranean Sea	SLP(+), LWS, RWA1	ASHF, VP1					AMO1	ASHF1
	Bering Sea	SLP(+)1	ASHF1	AL, LWS1	ASHF, ADV1	ENSO(+), RWA, SLP(+)1	ASHF1	AL, PDO1	ASHF1
	Northwest Atlantic			JS, RASC, RWS, SLP1	ASHF1
	Northeast Pacific	SLP(+), LWS1	ASHF, EHF1	AL1	ASHF1	ENSO(+), RWA1	AL, SLP(+)1	NPO(+), RWS, NPGO(+)1	SLP(+)1
	South Central Pacific					ENSO(+), RWA5	ASHF, ADV5

Large-scale and regional climate modes		Teleconnection processes and climatological features		Local processes affecting the mixed layer temperature budget
ENSO(+/−) CPEN IPO PDO(+/−) IOD(+/−) MJO NAM NAO(+/−) NPGO(+/−) NPO AMO SAM ASM	El Niño-Southern Oscillation Central Pacific El Niño Interdecadal Pacific Oscillation Pacific Decadal Oscillation Indian Ocean Dipole Madden-Julian Oscillation Northern Annular Mode North Atlantic Oscillation North Pacific Gyre Oscillation North Pacific Oscillation Atlantic Multi-decadal Oscillation Southern Annular Mode Asian Summer Monsoon	AB AL SLP(+/−) JS PNA RWA ABF BI KWO RWO RWS RASC LWS	Atmospheric Blocking Aleutian Low Sea Level Pressure Jet Stream position Pacific North American Pattern Rossby Wave (Atmospheric) Angola-Benguela Front Baroclinic Instability Kelvin Wave (Oceanic) Rossby Wave (Oceanic) Regional wind stress change Regional air-sea coupling Local wind stress change	ADV EHF ASHF VP	Ocean Advection Eddy heat flux Air-sea heat flux Vertical Processes (entrainment, turbulent mixing, thermocline deepening)

Our literature assessment of MHW drivers characterizes the contributions from: large-scale or regional climate modes (e.g. ENSO), atmospheric or oceanic teleconnection processes and climatological features (e.g. Rossby waves, fronts) [Mode/Teleconnection columns]; and local processes (e.g. ocean advection) affecting the MHW heat budget [Local Process columns] across four time scales (synoptic, intraseasonal, interannual and decadal) and classified by typology, i.e. eastern boundary currents (EBC), western boundary currents and extensions (WBC), tropics, and middle and high latitudes (MHL). For regions where no drivers or processes could be identified from the literature, the box is left blank. For ENSO, PDO, IOD, NAO and SLP, individual studies may indicate whether MHWs are associated with positive (+) or negative (−) phase. Case studies with an asterisk (*) have no documented MHWs, but literature identifies processes and modes that cause changes in the mixed layer temperature budget. Numbers correspond to a qualitative confidence assessment for literature documented MHW mode/teleconnection and local processes respectively. Confidence ratings are explained in the Methods and Supplementary Fig. 2, and include very high (₁), high (₂), medium (₃), low (₄), and very low (₅) confidence ratings. Corresponding references are provided in Supplementary Table 1