Monthly Report on Climate System Separated Volume No.12

--- Explanatory Note ---

Characteristics of the Global Sea Surface Temperature Data (COBE-SST)

1. Summary

    The Sea Surface Temperature (SST) analysis for climate monitoring is updated to a new analysis (COBE1-SST, Ishii et al., 2005) in March 2006 at Climate Prediction Division in Japan Meteorological Agency (JMA).
    In this paper, we show various figures and charts which describe the fundamental characteristics of the SST analysis. They should be expected to support understanding climate contents shown in the Monthly Reports on Climate System.
    The SST analysis is prepared over 100 years aiming at climate research and analysis. There are some pioneer works in this field such as HadISST (Rayner et al., 2003) prepared at Hadley center of Meteorological Office in United Kingdom and ERSST.v2 (Smith and Reynolds, 2004) prepared at the U.S. National Climate Data Center of National Oceanic and Atmospheric Administration (NOAA). These works are compared with COBE-SST in this paper.
    It is our great pleasure that this work would be utilized as a reference of the Monthly Reports on Climate System and other researches.
  1) Centennial in situ Observation-Based Estimates of the variability of SSTs and marine meteorological variables

1-1. Marine meteorological observation data

    The outline of the marine meteorological data used in the SST analysis is described here.
    In recent years, in accordance with the spread of an awareness of the issues to global warming, collection and maintenance of historical marine meteorological data have been performed on the basis of international cooperation. Comprehensive Ocean Atmosphere Data Set (COADS) is known as pioneering achievements about maintenance of the historical data set, and it has been also used in the SST analysis in JMA. It is noted that about 2.7 million reports of Kobe Collection data from 1933 to 1960 which digitized in collaboration with NOAA in 1961 are included in COADS. Kobe Collection data contains about 6.8 million reports of marine meteorological data from 1890 to 1960 which had been preserved at the Kobe Marine Observatory.
    The Japan Meteorological Agency and Japan Weather Association performed works of digitizing Kobe Collection data from 1890 to 1932 with an assistance of Nippon Foundation during 1995-2003 (Manabe, 1999). In this work, about 3.1 million reports were newly digitized and about 1 million reports of them are already included in the upgraded International Comprehensive Ocean Atmosphere Data Set (ICOADS release 2.1, Worley et al., 2005). The rest of them will be included in ICOADS near future. The newly digitized Kobe Collection data in 1890-1932 contributes to fill the data in the data sparse region, especially in the North Pacific around World War I (Manabe, 1999). One of characteristics of the new SST analysis is utilization of newly digitized Kobe Collection prior to its international use.
    In the analysis after 1998 and in the present and future operational analysis, the marine meteorological data collected through the GTS are utilized. In addition, we also used the drifting buoy data edited by the Canadian Marine Environmental Data Service (MEDS, 2000 edition).
    In the figures of ‘Marine meteorological observation data’, the outline of the fundamental observation database utilized for SST analysis is shown by diagrams. According to the figure of the historical number of reports, the rapid increase in the number of reports is seen in 1950s and afterwards. Moreover, after 1850, the number of reports shows the exponential increase of reports, and the decrease to about 1/10 during the two World Wars. The ratio of the areas having observation reports is investigated based on the number of reports in 1x1 degree boxes. The ratio of areas having observation indicates about 50% after 1950. The accuracy of the SST analysis can be roughly estimated with the geographical distribution of the number of observation reports accumulated every ten years.

1-2. Outline of the Sea Surface Temperature (SST) analysis

    Details of COBE-SST should be referred to Ishii et al. (2005). In this section, the outline of the new SST analysis is briefly described.
    The SST analysis has a resolution of 1 degree latitude and 1 degree longitude. The east-west grid point starts eastward from 0.5 degree east longitude to 0.5 degree west, and the north-south grid point starts northward from 89.5 degree south latitude to 89.5 degree north. The analysis uses optimum interpolation method. The deviation from normal of the previous day is multiplied by 0.95 and is used as first guess. The analysis is performed daily and uses the marine meteorological data for 7 days, namely for the day of interest before and after three days. Moreover, the observed data in the same day are averaged in 1.5x1.5 degree box before analyzing data by optimum interpolation method. This procedure saves the processing time of analysis by a computer.
    In using the past SST observation reports, bias corrections were performed by the way of Folland and Parker (1995).
    The quality control of observed data is performed by checking ship tracks, dates and positions of reports, and then erroneous data are automatically corrected in compiling marine meteorological data in JMA. Moreover, using observed data deviations from climate values during period between one month before and one month after the day of interest, biases of the data are estimated for the same ship call signs, and biased data are listed in a blacklist in the daily analysis.
    These quality controls are not applied to the data of the drifting buoys which are increasing in recent years. For the operational use, some modifications of quality control for drifting buoys are performed.
    The information of ice concentration is used in estimating SSTs in the Arctic and Antarctic Ocean. The figures concerning ice concentration are also shown here for monthly normal and standard deviation.

1-3. Characteristics of monthly mean SSTs analyzed

    In order to investigate and show the fundamental characteristics of the COBE-SST analysis, monthly SSTs are shown by averaging daily analyses. The normal of every 30 years (30-year mean of each month) and the standard deviation in every 30 years are shown. The maps of the standard deviation of monthly SST show the variations in respective ages and areas. The differences in standard deviation maps indicate not only natural variability but also analysis accuracy for the data sparse area shown in the observed data distribution maps.
    The maps of SST anomalies which are deviations from 1971-2000 normal, indicate long term variation in the monthly 30-year mean or annual 10-year mean.
    In the time series of area-averaged SSTs, monthly averaged and 13-month moving averaged SSTs and SST anomalies (1971-2000 normal) are shown. The 5 regions in the tropical Pacific ocean for monitoring ENSO, and regions of 5 main oceans and global ocean are shown. The figures are prepared to show the intra-seasonal and inter-annual variations in each age.
    SSTs and SST anomalies in the time-longitude section of each latitude band show inter-annual variations in the ocean basin scale. The latitude bands in the tropics, Northern and Southern Hemisphere are selected to show phenomena with several years to decadal time scale.

1-4. Comparison with the Sea Surface Temperature analyses of other climate centers.

    Comparison with other centers’ SST analyses, which are produced for climate study and monitoring climate system such as global warming, is performed. The SST analyses utilized for the comparison with HadISST and ERSST.v2. Both SST analyses utilize COADS in processing.
    The 30-year mean monthly normal and standard deviation in every 30 years, and 10-year mean annual SSTs in every ten years, are compared and the differences between HadISST and COBE-SST, ERSST.v2 and COBE-SST are shown. The difference maps are arranged in the same form as in the characteristics of COBE-SST in order to compare easily.
    In the comparison of the time series, area-averaged SSTs and SST deviations from its own 1971-2000 normal, are plotted over the same graphs for COBE-SST, HadISST, and ERSST.v2. The area-averaged SST differences and anomaly differences from COBE-SST are also shown.

1-5.SST Trends in the global Oceans

    In this section, long term variation of the SST prepared for monitoring the global warming are shown for the period of 1891-2004. The plotted areas are global, Northern and Southern Hemisphere oceans. SST trends in the 5 main oceans, namely the North Pacific Ocean, the South Pacific Ocean, North Atlantic Ocean, South Atlantic Ocean, and the Indian Ocean are also plotted. The trend and long term variation in each ocean shows tendency of warming for 100 years.
    The rate of temperature rise over the last 100 years (1891-2003) is shown for each grid of the global ocean. Moreover, trends estimated by HadISST and ERSST.v2 are also shown respectively. In comparing figures, the differences among these horizontal trend patterns of the three SSTs, are considered to be arisen from the methods of SST analyses.


Folland, C. K. and D. E. Parker 1995: Correction of instrumental biases in historical sea surface temperature data. Q. J. R. Meteorol. Soc. 121, 319–367.
Ishii, M., A. Shouji, S. Sugimoto and T. Matsumoto, 2005: Objective Analyses of Sea-Surface Temperature and Marine Meteorological Variables for the 20th  Century using ICOADS and the Kobe Collection. Int. J. Climatol., 25, 865-879.
Manabe, T. 1999: The digitized Kobe Collection, Phase I: Historical surface marine meteoroical observations in the archive of the Japan Meteorological Agency. Bull. Amer. Meteor. Soc., 80, 2703-2715.
Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell, E. C. Kent and A. Kaplan, 2003: Global analyses of sea temperature, sea ice, and night marine air temperature since the late nineteenth century. J.Geophys. Res., 108(D14), 4407, doi:10.1029/2002JD002670.
Smith, T., M., and R. W. Reynolds, 2004: Improved Extended Reconstruction of SST (1854-1997). J. Climate., 17, 2466-2477.
Worley, S., S. D. Woodruff, R. W. Reynolds, S. J. Lubker and N. Lott, 2005: ICOADS Rlease 2.1 Data and Products. Int. J. Climatol., 25, 823-842.

2. About the Data

    In this site, colored and b/w figures are available in PDF format. The files having color image in PNG format are also available.
    Binary format data of analyzed monthly mean Sea Surface Temperature (for the period of 1891-2005) are contained in this website.
    The binary data of monthly mean SST are recorded as yearly files under the directory "/sstdata/hist/". The names of yearly files are "sstyyyy" (yyyy expresses year).
    In the directory "/sstdata/norm/", monthly normal based on 1971-2000, and monthly standard deviation in the period of 1971-2000 are stored in one file named sst-clm7100.grb.
    The format of binary files are based on WMO binary code FM92 GRIB.

The outline of data is as follows.
Data format WMO binary code FM92 GRIB
Unit degree centigrade
Grid Interval longitudinal 1.0 x latitudinal 1.0 degree in the global (360x180),
Data Range data are recorded in order of the following successively,
  1. start point is at 89.5 degree of South latitude, 0.5 degree of East longitude,
  2. having 360 points from west to east,
  3. having 180 points from south to north,
  4. from January to December.

    Please refer to "How to use GRIB format files" about the easy usage of the GRIB format file.