Marine Air Temperatures in the Northeast Atlantic and UK waters have warmed rapidly over the last 25 years. The observed warming is greatest in the Southern North Sea with warming rates of over 0.6 °C decade-1. Similarly, sea-surface temperatures (SST) in UK coastal waters and in the north east Atlantic have risen by between 0.2 and 0.8˚C/decade since the 1980s.

The most rapid increases have been observed in the Eastern English Channel (Region 3) and Southern North Sea (Region 2) at a rate between 0.6 and 0.8˚C/decade

The temperature of the upper ocean (0-800m) to the west and north of the UK (Region 8) has been generally increasing since the 1970s. Superimposed on the underlying upward trend are decadal scale patterns of variability, fluctuating between relative maxima around 1960 and in the 2000s, with relative minima in the 1980s and 1990s.

The observed temperature variability has been attributed to a combination of anthropogenic global climate change and natural variability (i.e. 'internal' variability in the ocean atmosphere system). The Atlantic Multidecadal Oscillation (AMO) is a pattern of variability in North Atlantic SST that is thought to be representative of the internal variability, the decadal scale variations of temperature observed in UK waters are similar to those of the AMO. As a result of both of these 'drivers' (climate change and internal variability) a significant period of rapid warming occurred from 1995 to 2003.

West of the UK the water of the deep ocean (>1000 m) comes from the Labrador Sea and has cooled since 1975. North of the UK, the deep water (800 m) flows from the Nordic Seas and shows no long-term trend since 1950. The deep bottom waters of the Nordic Seas are known to be warming.

Despite the long term warming trend, in 2008 the temperatures observed in most areas were slightly lower than observed in 2003-2007. This is a reflection of the high inter-annual variability, particularly in coastal waters and is not indicative of a decreasing trend.

By the 2080s warming in the shelf seas around UK and the upper layers of the North Atlantic is predicted to continue, although perhaps at a lesser rate to that observed in the last 25 years. Natural variability, driven by atmospheric and oceanic processes introduces a level of uncertainty that makes it difficult to predict the direction of temperature change over the next decade.

Off the shelf in deeper waters, nearbed temperatures are influenced by deep ocean circulation and little or no warming is projected to occur by 2080. In the shorter term an observed warming trend in the Labrador Sea is likely to cause an increase in temperature in deep waters west of the UK. 

What is already happening: High

Air Temperature (in-situ observations): The number of VOS observations of the marine air temperature has declined in recent years. Additionally, an increasing number of observations are reported with a masked or missing call sign due to ship security and commercial concerns, preventing the association of the observations with the metadata required to height and bias adjust the observations. Both the reduction in number of observations and the loss of the ability to match the observations to metadata act to increase the uncertainty in the air temperature estimates. The highest confidence (lowest uncertainty) in our air temperatures can be found over the North Sea, English Channel and South West Approaches whilst the lowest confidence in the air temperature values can be found to the south and south west of Iceland (Figure 16a). There has been little change since 1970 in the uncertainty in the air temperature for the regions where we have highest confidence whilst the uncertainty in regions where we have low confidence has increased (Figure 16b).

Figure 16: a) Average uncertainty (°C) in the monthly mean air temperature averaged
over 2004 - 2008 and b) change in the uncertainty (°C) for the period 2004 - 2008
relative to 1970 - 1974.

Sea-Surface Temperature (in-situ and satellite observations) The gridded SST data presented here come from the HadISST1.1 dataset. The dataset uses a combination of in-situ and satellite observations, gridded and interpolated to create a complete dataset. Data coverage in the area of interest is generally good and a recent comparison of this dataset with independent in-situ data (Hughes et al., 2009) indicated that there was good agreement in the region of interest. Recent research (Kennedy et al.  2009; Reverdin et al.  2009) has shown that changes in the composition of the in situ SST observing system may cause systematic biases in the data. However the nature of these biases is unlikely to change the picture of rising temperatures and their distribution. This conclusion is reinforced by the agreement of SST changes with those in air temperature which is measured using different methods.

In-situ sea temperature: Measurements of temperature profiles at offshore sites are made 1-3 times per year, under-sampling the seasonal cycle which may alias the results. Shelf sea and coastal stations are sampled more frequently (up to daily), so the seasonal cycle is usually better resolved. Calibration is good (although data prior to 1970 are less reliable), so high confidence can be put on in-situ measurements.

Temperature profile information in the North Atlantic is now much better sampled than in the past due to the deployment of many Argo profiling floats. However questions of biases in recent batches of floats (Willis et al., 2007) and in homogeneity between Argo and eXpendable BathyThermographs (XBT's) data (Gouretski and Koltermann, 2007) mean the overall confidence probably remains moderate, pending further research.

Although there are gaps in the observational record of subsurface temperature and some areas are poorly observed, temperature is the most widely measured parameter and there is therefore a large amount of evidence. Although some of the observational records are shorter than others and have difference in sampling, they all offer a coherent picture of long term and shorter variability, giving rise to a higher level of confidence in the results.

What could happen: Medium

For shorter term predictions (i.e. decadal scale), natural internal variability cannot currently be predicted with any confidence and it is therefore difficult to determine if natural variability will continue to enhance or begin to oppose the long term warming trend over the next decade.

Confidence in the global increase in SST is high (e.g. IPCC, 2007) and there is high confidence in the long-term future warming trend. However our confidence in the exact rates of warming at regional scales is lower. With the UKCP09 scenarios, as the ocean model was run only once (medium emission scenario), there are no estimates of upper or lower bounds of change and consequently no confidence intervals. The effect of decadal uncertainties has been addressed by averaging model output over a 30year period.

The top priority knowledge gaps that need to be addressed in the short term to provide better advice to be given to policy makers are:

1. Satellite observations of SST have resulted in good data coverage in the surface waters around the UK, whilst data from below surface is still relatively sparse. Satellite SST also requires continuity of satellite missions and availability of adequate in-situ data for validation and bias adjustment. Further research is required to understand the impact of changes to the in-situ observing network for SST. The number of air temperature observations in UK coastal waters, and globally, have declined in recent years increasing the uncertainty of marine air temperature datasets.
2. It is of vital importance that the existing in-situ time series are maintained. For many time series there is a lack of funding security; most are maintained through a rolling programme of grants for a short number of years. Many time series face periodic funding shortages, especially during times when the political climate turns against ocean monitoring. Some series have suffered major gaps as a result, and some have reduced temporal resolution in recent years. At many stations the existing sampling is not sufficient for a full understanding of variability and hence reduces confidence in the representativeness of measurements made. The addition of more in-situ stations and improved sampling of the seasonal cycle is also therefore desirable.
3. The deep ocean (deeper than about 2 km) is poorly sampled. The Argo programme has addressed this to some extent for the upper 2 km of the deep ocean, but funding for this programme is also uncertain. For the surface to mid-depth ocean questions of the homogeneity of data from Argo floats and between Argo and other sampling technologies (e.g. XBTs) remain. Recent rapid changes in the in-situ observing system means that the homogeneity of the current observing system, and its consistency with earlier observations, needs urgent assessment.
4. Further research on ocean processes is necessary to help understand the inter-annual to decadal variability observed at regional and ocean scales and investigate the mechanisms that determine hydrographic properties and ocean transports.

The socio-economic impacts of changing marine temperatures will be through their role as the key underlying driver of climate impacts across all components of the ecosystem.

Hughes S.L., N.P. Holliday, J. Kennedy, D.I. Berry, E.C. Kent, T. Sherwin, S. Dye, M. Inall, T. Shammon, and T. Smyth (2010) Temperature (Air and Sea) in MCCIP Annual Report Card 2010-11, MCCIP Science Review, 16pp. www.mccip.org.uk/arc