We know the ocean absorbs circa 90% of the Earth’s heat budget and 30% of the human–induced carbon emissions, but the full comprehension of all the processes related to the ocean’s role in climate change still requires important scientific advances. For instance, knowledge gaps persist regarding the Ocean’s tipping points that might lead to irreversible changes. On the other hand, global warming consequences are already detectable in several regions, particularly those related to extreme seawater temperature events, with corresponding ecosystem’s health loss and economic impacts.
It is under this scope that Marine Heat Waves (MHW) have been the topic of recent discussions. Qualitatively, MHW are localised, persistent, and anomalously warm seawater temperature events, usually detected at the sea surface level. Quantitatively, several definitions have been put forth to detect sea surface temperature (SST) anomalies and quantify the occurrence and intensity of MHW. Defining objective thresholds and metrics upon which MHWs are detected and measured is indeed a complex challenge, particularly due to the long–term positive SST trends, even though with contrasting global patterns. MHW, like atmospheric heatwaves, have nevertheless also been increasing globally in terms of frequency, intensity, and duration.
The consolidation of Earth observation (EO)–based SST data has allowed the scientific community to characterise global and regional SST patterns over the years. The availability of such extended time series, together with statistical distribution methods, namely percentile–base thresholds, are key to detecting MHW. The more broadly adopted method – also used in CAREHeat – has been developed by Hobday et al. (2016) and is based on the methods established for atmospheric heat waves.
Based on that method, a MHW corresponds to a period of (at least) five days of abnormally warm SST surpassing a predetermined threshold value. Such threshold – for example, the 90th percentile – on a given day of the year, is calculated based on a multi–decadal climatological reference, such as a 30–year period. A time series of MHW can be temporally aggregated by year, season, or event to convey key mensurable aspects, such as MHW frequency, duration, intensity, rate of evolution and spatial extent.
To compare various MHW on a global scale, Hobday et al. (2018) also developed a categorisation scheme to distinguish the severity of each event, comprising four levels: moderate, strong, severe and extreme. These categories translate the event’s maximum intensity – the SST anomaly – into discrete levels depicting the interval between the climatological mean and the reference threshold used. This rationale follows previous practices implemented for atmospheric heat waves, where severity categories usually correspond to expected levels of sectoral impacts, namely human mortality.
The detection of MHW is far from a simple algorithmic operation, including with the Hobday statistical–based approach. Several reservations should be mentioned, mostly because these MHW are the product of both long–term and short–lived phenomena, with corresponding repercussions.
The drivers triggering these extreme events can be related to either atmospheric or oceanic processes – such as air–sea heat flux and ocean advection – or complex combinations of the two. While stochastic atmospheric forcing may affect SST anomalies over middle and high latitudes, they can also occur due to oceanic flows or abnormal deep water ocean state re–emergences. These processes are controlled by large–scale climate modes, which may span from interannual to multidecadal durations and repeat cycles, such as the North Atlantic Oscillation (NAO) and the El Niño–Southern Oscillation (ENSO).
Moreover, long-term SST trends are quite heterogeneous across the world – in some cases, the warming trend is so strong that using the earliest (and colder) 30-year climate reference results in a quasi-permanent seasonal MHW over a region (for example, in the Mediterranean basin during summer). Establishing statistical metrics regardless of these underlying climate signals provides an incomplete picture on how to remove long–term changes from transient MHW events.
Understanding how global warming affect the ocean below the surface poses yet another challenge since deep ocean data profiles are much more limited. Detecting MHW only at the surface level does not provide sufficient insights on how they propagate towards the ocean bottom. A good example deals with the uncertainties about increased stratification and reduced vertical mixing due to the deep ocean slower circulation and thermal mixing.
CAREHeat will start by revising and improving MHW detection algorithms. This will provide metrics to better analyse the impact of MHW on marine ecosystems. The project will then combine state-of-the-art data science techniques (such as machine learning and artificial intelligence) and numerical models (such as atmospheric–oceanic processes coupling) to develop new approaches for the reconstruction of the physical processes related to air-sea interactions. Various methodologies commonly used to separate the various drivers of MHW events will also be revised. CAREHeat will provide new insights over the relative weights of long–term warming trends and climate modes, highlighting the differences between stable and transient phenomena. Novel technologies and EO–based information will be employed to reconstruct specific MHW events, describing leading processes and their spatio–temporal propagation patterns.