11:00 AM - 1:00 PM
[AOS21-P09] Status and impact of abrupt salty drift in Argo salinity profile data and new challenges to quality control for Argo data
Keywords:Argo, quality control
Thanks to Argo, which is the international program for ocean observation, temperature and salinity in the ocean shallower than 2000dbar have been monitored and their profile data have been accumulated. More than 150,000 profile data have been obtained annually since 2013.
It has recently been reported that the Sea-Bird Scientific CTD sensors on the Argo floats with abrupt salty drift have been found. Serial number of CTD sensors with this issue are bigger than 6000. Most of floats equipped with the CTD sensors in the range of the serial numbers are now operating, and as a result, there is a salt error exceeding the target accuracy of 0.01 in the real-time data. In fact, it was reported that the rise in global mean sea level was underestimated due to this problem (Barnoud et al. (2021)).
Then, I reported the current status of this issue, the response of the Argo Data Management Team, and the continuity of this issue at the 2021 Autumn Meeting of the Oceanographic Society of Japan. Six months have passed since then, and the state of the Argo data quality after that is reported. The percentage of Argo profiles flagged as “bad data” in all layers of the salinity profile was as low as 5% in 2014, gradually increased from here to 2021, and reached 16% in 2021. This is the highest percentage since the beginning of Argo. In the presentation, I will report again on the continuity of this problem. In addition, we will introduce the quality control results using the new quality control method (Sugiura and Hosoda (2019)) using machine learning, which we have newly started to work on.
It has recently been reported that the Sea-Bird Scientific CTD sensors on the Argo floats with abrupt salty drift have been found. Serial number of CTD sensors with this issue are bigger than 6000. Most of floats equipped with the CTD sensors in the range of the serial numbers are now operating, and as a result, there is a salt error exceeding the target accuracy of 0.01 in the real-time data. In fact, it was reported that the rise in global mean sea level was underestimated due to this problem (Barnoud et al. (2021)).
Then, I reported the current status of this issue, the response of the Argo Data Management Team, and the continuity of this issue at the 2021 Autumn Meeting of the Oceanographic Society of Japan. Six months have passed since then, and the state of the Argo data quality after that is reported. The percentage of Argo profiles flagged as “bad data” in all layers of the salinity profile was as low as 5% in 2014, gradually increased from here to 2021, and reached 16% in 2021. This is the highest percentage since the beginning of Argo. In the presentation, I will report again on the continuity of this problem. In addition, we will introduce the quality control results using the new quality control method (Sugiura and Hosoda (2019)) using machine learning, which we have newly started to work on.