Abstract

Extending previous work, long-term temperature trends in the upper layers of the ocean are calculated from gridded observed data to explore spatial and temporal trend variability. Interdecadal and spatial trend variability is presented for various analysis region sizes. Depths between 50 m and 1000 m are examined.

Despite using larger grid boxes, most of the ocean does not have 50-year trends that are significant at the 90% confidence level. At 100 m, only about 35% of the ocean has such trends and the coverage declines dramatically with increasing depth. As noted before there is much spatial structure in 50-year trends, with areas of strong warming and strong cooling. There is also strong interdecadal variability shown in 20-year trends; almost every region studied shows both warming and cooling 20-year trends over a 50-year period. 50-year trend results are compared and contrasted with trends calculated from data interpolated to standard levels and from a dataset used in previous heat content trend studies.

These comparisons show that different data treatments can yield significantly different trends. Thus, the uncertainty in heat content integrals is large and such integrals may not be truly representative of the global ocean heat content. This situation prevails because of the presence of strong interdecadal and spatial trend variability in conjunction with strong temporal and spatial sampling issues. It is suggested that calculating a world ocean residual heat content trend is highly problematic over this period.