Time series of average annual temperature at sea surface and bottom

Sub-challenge: Create time series of average annual temperature at sea surface and bottom.

This sub-challenge was interpreted as the average annual temperature of the entire Arctic area, as defined for this project (The Arctic Ocean as defined in the CIA factbook) and can be divided into two scenarios:

  1. Time series of average annual temperature at sea surface for the entire study area;
  2. Time series of average annual temperature at sea bottom for the entire study area.

No time span was indicated in the original sub-challenge (only ‘annual average’) and no spatial scale was indicated.

A time series of average annual temperature at sea surface (1) is available free of charge. For an interactive map with data on water temperature see http://www.emodnet-physics.eu/map/ and select “water temperature”. Here you can filter for temperature (but also salinity, conductivity etc.) and select individual locations which will open a new window showing the available measured data. This data is presented differently per location, depending on the data available. Figure 1 shows the interactive map with a selected location.

Interactive map with selected location
Figure 1: Interactive map with selected location

Figure 2 is a depiction of the sea surface temperature (SST) of the Northern Hemisphere, with the upper limits and lower limits for each year and the calculated sea surface temperature anomalies relative to the mean SST of 1961 until 1990. This data is downloadable free of charge at https://www.metoffice.gov.uk/hadobs/hadsst3/data/download.html. Unfortunately  this data shows the mean of the entire Northern Hemisphere, thus showing a very generalized image of the SST, which is not directly applicable to the Arctic Sea. More local data can be easily retrieved using the interactive map mentionedabove. However, the map does not show compared data, like SST anomalies.

Sea Surface Temperature
Figure 2: Median average Sea Surface Temperature anomaly for the Northern Hemisphere, datasource: UK Met Office).

The same data trend can be seen on the website of the National Oceanic and Atmospheric Administration (NOAA), showing the SST anomaly of 60 degrees North to 90 degrees North (Arctic circle), depicted in degrees Kelvin, Figure 3.
At the NOAA website you can also find daily updated maps with ocean temperatures: http://www.ospo.noaa.gov/Products/ocean/sst/contour/

Sea Surface Temperature
Figure 3: Sea Surface Temperature anomaly, datasource: NOAA.

Timmermans & Proshutinsky (2015) constructed the SST anomalies from three different Arctic seas, Figure 4. This is based on the same data available at the NOAA website. About the dataset:
“Sea surface temperatures are determined using the extended reconstructed sea surface temperature (ERSST) analysis. ERSST uses the most recently available International Comprehensive Ocean-Atmosphere Data Set (ICOADS) and statistical methods that allow stable reconstruction using sparse data. The monthly analysis begins January 1854, but due to very sparse data, no global averages are computed before 1880. With more observations after 1880, the signal is stronger and more consistent over time.” (www.ncdc.noaa.gov)
The graph shows considerable variation between the different seas, indicating that when using the mean SSTs for the entire Arctic, important information might be overlooked.

Time series of area-averaged SST anomalies
Figure 4: Time series of area-averaged SST anomalies [°C] for August of each year relative to the August mean for the period 1982-2010 for the Chukchi and Kara seas and eastern Baffin Bay. The dash-dotted black line shows the linear SST trend for the Chukchi Sea (the same warming trend as eastern Baffin Bay). Numbers in the legend correspond to linear trends (with 95% confidence intervals) in °C/year (source: Timmermans & Proshutinsky, 2015).

Information about the sea bottom temperature is scarce. Rabe et al. (2015) described the surface and bottom sea temperature from 86 locations which were visited consecutively in the year 2012 (Figure 5). The data can be downloaded free of charge from the Pangea website (www.pangaea.de).
This data treats all locations as 1 sampling site, meaning that local differences are not accounted for.

Sea Surface and Bottom Temperature
Figure 5: Sea Surface and Bottom Temperature in 2012, datasource: Rabe et al., 2015.

In the period from 2007 until 2014, Russian-German research programs operated a series of year-round oceanographic moorings in the Laptev Sea, which provide a nearly continuous hydrographic record from the central shelf (Janout et al. 2016). This data was analyzed by Janout et al. (2016), Figure 6 shows the graph from the paper in which the mid-water and bottom water temperatures are depicted.
The article can be found at Wiley online library and costs $6 for 48 hours of viewing, $15 for unlimited viewing and $48 for a pdf download. There is no option for downloading this dataset from the internet.
2007 - 2014 mooring record
Figure 6: The 2007–2014 mooring record: (a) near-bottom water (blue) and midwater (red) salinity and (b) temperature. The data were smoothed using a 2-day running mean to remove high-frequency variability. Areal-mean SSTs [Reynolds et al., 2002] are included for comparison (green). Red dots in the first half of the record indicate midwater values from CTD profiles. Note that the September 2013 to September 2014 deployment is ~146km north of the previous location; the black dashed line marks the transition time. Black bars at the top of Figure 1a indicate ice cover [Cavalieri et al., 1996]. (c) The white star marks the mooring location. Letters mark adjacent shelf seas: Barents (BS), Kara (KS), East Siberian (ESS), and Chukchi (CS) Seas. Light blue shading indicates regions

Temporal data of SST and bottom sea temperature,  can also be found in the sub-challenge: Change in average temperature at surface, 500 metre depth and bottom on a grid over the past 10 years and 50 years. Here data was discussed from the World Oceanic Atlas (WOA), which is available free of charge, concluding that the data requires careful analysis. Therefore, the analysis from a recently published paper (Seidov et al., 2015) was used (available at http://www.nodc.noaa.gov/OC5/regional_climate/arctic/).
Data are available from WOA for the periods 1955-64, 1965-74, 1975-84, 1985-94, 1995-2004 and 2005-2012. For the challenge the most recent data was used to calculate the 10 year change (Figure 1), for the 50 year period they used the difference between 2005-2012 and 1955-1965 (Figure 2).
Change in temperature 10 year period
Figure 1: Change in temperature (°C) calculated as difference between the average over 2005-2012 and average over 1995-2004 from surface (top), at 500 m (middle) and bottom (lower). Some artificial land occurs for bottom estimates due to interpolation challenges.
Change in temperature 50 year period
Figure 2: Change in temperature (°C) calculated as the mean of the 2005-2012 average and 1995-2004 average compared to the 1955-1965 average from surface (top panels), at 500 m (middle panels) and bottom (lower panels). Some artificial land occurs for bottom estimates due to interpolation challenges.
Data use, availability and gaps
There is sufficient data available on sea surface temperature, but less so  for bottom temperature. The available data is mostly point-data from measuring stations. As can be seen in the sub-challenge ‘temperature change on a grid’, different spatial areas can show different changes in temperature. When averaging the entire Arctic area, it would give an inaccurate impression of reality.
All but one dataset can be downloaded free of charge. However, the dataset with the most information regarding sea bottom temperatures is unavailable. The downloadable datasets can be found on the following websites:
Conclusion and lessons learned
This sub-challenge was divided into two scenarios:
  1. Time series of average annual temperature at sea surface for the entire study area;
  2. Time series of average annual temperature at sea bottom for the entire study area.
For the first scenario the information and data were available. Although a time series, with adjustable time periods, could not be found. Still, the data to construct such time series is present and downloadable free of charge.

For the second scenario there is less information available. Free available information was limited to research over the course of one year (2012) on different locations and comparative data from set time periods. A time series from sea bottom temperature between 2007 until 2014 was found in a research paper. This paper was not available free of charge and there was no option online for retrieving the dataset.  

A limitation for both scenarios is that when averaging the temperatures from the entire arctic into one mean temperature important local differences get overlooked.

The interactive map on the Emodnet website is already very informative. But it might hold more potential. The possibility to select multiple locations and compare them in some way could be an upgrade. Especially when available data could be combined and plotted to create comparative time series.
Another recommendation would be to retrieve the dataset from Janout et al. (2016) and make it publicly available, this dataset holds the most comparative data regarding sea bottom temperature and is a necessary addition to the compiled data.