Trends of monthly mean sea surface temperature
The sea surface temperature, SST is not a climate index in the classical sense, but still a good climate indicator. The SST has an effect for example on the evaporation rate over the oceans and therefore an indirect effect on precipitation, air temperature or available energy for the formation of hurricanes.
The sea surface temperature trends from monthly averaged data of two different datasets for different open ocean areas were considered. Coastal regions were excluded. For the study of selected regions with presentation of their time series, nine areas were selected as boxes that have a size of 20 x 25 deg, which is about 5.5 million km². This study case shows the differences between the oceans. For the global analysis of all open ocean areas, 10 x 20 degrees wide areas were selected, which corresponds to about 2,000,000 square kilometers.
A trendcalculation from pure satellite data can have large uncertainties, as the data quality can be very variable from pixel to pixel.
Firstly, the monthly mean Reynolds-SST data from ICDC have been used. These are based on the Pathfinder AVHRR data, from which a consitstent, global (including land) dataset is derived with optimum interpolation method. It was used for the time period November 1981 to April 2011.
For an overlapping period and the same areas were also calculated the SST trends from the AMSR-E SST data to compare whether the selected data set affects the SST trends. The AMSR-E data is available for the period 02/06/2002 to 31/12/2010 as a 3-day composite. For this comparison, monthly means have been calculated (including all available values for a grid point) from the 3-day composites.
Both data sets have a regular grid with spatial resolution of 0.25°x0.25°.
SST Trends from Reynolds (1981-2010)
The result of the calculation of the Reynolds SST trends from monthly averaged data for 9 selected areas for the period 1991 - 2010 is shown on the following illustration. The numbered boxes indicate the selected areas. The mean percentaged temperature change per year (%) and the absolute change in temperature (° C) per year is stated in the boxes. The colors in the background stand for the average SST over all months of the period (in °C). The course and trend of the SST for each of the areas are shown in the illustrations below. The period of nearly 30 years, considered here, corresponds to a typical climate normal period.
SST time series and trends for 9 areas from Reynolds
The figures show the time series of SST with the trend for each area shown above. In the title line of each figure, the corner points of the areas, and the calculated temperature trend in ° C per year and percentages per year are given. In the second time series below the number of pixels is given, which were used for the monthly calculation. In this dataset, the number of pixels is always constant, because it was interpolated optimal and thus no errors occur for pixel values.
Global SST trends from Reynolds data
The figure shows global SST trends, calculated from Reynolds data for the period 1981 - 2010. The boxes show the areas of about 2,000,000 km² (20 x 25 degrees), that have been averaged. They contain the SST trend in percent per year, and in °C per year. To highlight the trends, the colors show the temperature rise in °C.
At first sight, there is no clear global trend. Therefore, the second figure shows the distribution of temperature increases per year of all boxes: All the values that fall into a specific interval are counted. Here a tendency to a general temperature increase of about 0.01 to 0.02 °C becomes visible. The data with coordinates and temperature trends for download: trends100reynolds1981_2011.txt
SST trends from AMSR-E (2002-2010)
The result of the calculation of the AMSR-E SST trends from monthly averaged data for 9 selected areas for the period 2002 - 2010 is shown on the following figure. The numbered boxes show the selected areas. Stated in the box are the percentaged change in temperature per year (%) and the absolute change in temperature per year (°C). The colors in the background state the average over all months of the period of SST (in °C). The course and trend of the SST for each area is shown in the figures below.
SST time series and trends for 9 regions from AMSR-E
For each of the above marked regions, the figures show the SST time series with the calculated trend. The corner points of the areas, and the calculated temperature trend in ° C per year and percentages per year are given in the title line of each figure. In the second time series below the number of pixels is given, which were used for the monthly calculation. Due to missing or erroneous values, the data has gaps, which can be seen from the variable number of pixels used for the calculations.
Global SST trends from AMSR-E
The figure shows global SST trends, calculated from AMSR-E data for the period 06/2002 - 12/2010. The boxes show the about 2,000,000 km² large regions, that have been averaged. They contain the SST trend in percent per year, and in ° C per year. The colors show the temperature rise in °C in order to highlight the trends.
Comparison between Reynolds and AMSR-E SST trends
This figure shows for comparison with AMSR-E in the same presentation as above, the global SST trends, calculated from Reynolds data for the period 06/2002 - 12/2010. Already visually, the figures look quite similar. In numbers, the average difference between the annual temperature variations stated in the box is: mean (deltaT_amsre-deltaT_reynolds) = 0.0053°C with a standard deviation std = 0.016. So there is a very good agreement of the results.
The figures also show that this short period of about 10 years is in fact too short, since the trend variations of the SST are significantly larger, than in the calculation over the whole Reynolds SST period of about 30 years. The latter should be used for relevant trend analysis.
Name: Annika Jahnke-Bornemann
Institute: ICDC, KlimaCampus Universität Hamburg
Email: annika.bornemann@ zmaw.de