Improvements of the Daily Optimum Interpolation Sea Surface Temperature (DOISST) Version 2.1

TLDR

DOISST v2.0, a blend of ship, buoy, and AVHRR satellite sea‑surface temperatures, suffered a cold bias in the Indian, South Pacific, and South Atlantic Oceans because the transition from TAC to BUFR omitted drifting‑buoy data. This study investigates the causes of that bias and presents an upgraded DOISST v2.1 that corrects it. Six experiments tested bias‑reduction strategies—adjusting ship SSTs with ICOADS R3.0.2 buoy data, incorporating Argo observations above 5 m, switching to MetOp‑B satellite data, and applying a freezing‑point correction for Arctic SSTs—to develop the new version. DOISST v2.1 reduces global and Indian Ocean biases to –0.07 °C and –0.14 °C against independent Argo, –0.04 °C and –0.08 °C against dependent Argo, and brings the GMPE difference down to –0.01 °C globally and –0.04 °C in the Indian Ocean.

Abstract

Abstract The NOAA/NESDIS/NCEI Daily Optimum Interpolation Sea Surface Temperature (SST), version 2.0, dataset (DOISST v2.0) is a blend of in situ ship and buoy SSTs with satellite SSTs derived from the Advanced Very High Resolution Radiometer (AVHRR). DOISST v2.0 exhibited a cold bias in the Indian, South Pacific, and South Atlantic Oceans that is due to a lack of ingested drifting-buoy SSTs in the system, which resulted from a gradual data format change from the traditional alphanumeric codes (TAC) to the binary universal form for the representation of meteorological data (BUFR). The cold bias against Argo was about −0.14°C on global average and −0.28°C in the Indian Ocean from January 2016 to August 2019. We explored the reasons for these cold biases through six progressive experiments. These experiments showed that the cold biases can be effectively reduced by adjusting ship SSTs with available buoy SSTs, using the latest available ICOADS R3.0.2 derived from merging BUFR and TAC, as well as by including Argo observations above 5-m depth. The impact of using the satellite MetOp-B instead of NOAA-19 was notable for high-latitude oceans but small on global average, since their biases are adjusted using in situ SSTs. In addition, the warm SSTs in the Arctic were improved by applying a freezing point instead of regressed ice-SST proxy. This paper describes an upgraded version, DOISST v2.1, which addresses biases in v2.0. Overall, by updating v2.0 to v2.1, the biases are reduced to −0.07° and −0.14°C in the global ocean and Indian Ocean, respectively, when compared with independent Argo observations and are reduced to −0.04° and −0.08°C in the global ocean and Indian Ocean, respectively, when compared with dependent Argo observations. The difference against the Group for High Resolution SST (GHRSST) Multiproduct Ensemble (GMPE) product is reduced from −0.09° to −0.01°C in the global oceans and from −0.20° to −0.04°C in the Indian Ocean.