Estimating Annual Global Upper-Ocean Heat Content Anomalies despite Irregular In Situ Ocean Sampling

The effects of irregular in situ ocena sampling on estimates of annual globally integrated uppr ocean heat content anomalies (OHCA) are investigated for sampling patterns from 1955 to 2006. An analytical method is presented for computing the effective area covered by an objective map for any given in situ sampling distribution. To evaluate the method, appropriately scaled sea surface height (SSH) anomaly maps from Archiving, Validation, and Interpretation of Satellite Oceanographic data (AVISO) are used as a proxy for OHCA from 1993 to 2006. Use of these proxy data demonstrates that the simple area integral (SI) of such an objective map for sparse datasets does not agree as well with the actual integral as the weighted integral (WI), defined as the simple integral weighted by the ratio of the total area over the "observed" area. From 1955 to 1966, in situ ocean sampling is inadequate to estimated accurately annual global integrals of the proxy upper OHCA. During this period, the SI for the sampling pattern of any given year underestimates the 13-yr trend in proxy OHCA from 1993 to 2006 by around 70%, and confidence limits for the WI are often very large. From 1967 to 2003 there appear to be sufficient data to estimate annual global integrals. Limited by the constraints of this analysis, the SI for any given years's sampling pattern still underestimates the 1993-2006 13-yr trend in the proxy by around 30%, but the WI matches the trend well with small confidence limits. For 2004 through 2006 in situ sampling, with near-global in situ Argo data coverage, the 1993-2006 13-yr trend in the proxy is equally well represented by the SI or WI.