Stability Assessment of the (A)ATSR Sea Surface Temperature Climate Dataset from the European Space Agency Climate Change Initiative

Abstract

Sea surface temperature is a key component of the climate record, with multiple independent records giving confidence in observed changes. As part of the European Space Agencies (ESA) Climate Change Initiative (CCI) the satellite archives have been reprocessed with the aim of creating a new dataset that is independent of the in situ observations, and stable with no artificial drift (<0.1 K decade −1 globally) or step changes. We present a method to assess the satellite sea surface temperature (SST) record for step changes using the Penalized Maximal t Test (PMT) applied to aggregate time series. We demonstrated the application of the method using data from version EXP1.8 of the ESA SST CCI dataset averaged on a 7 km grid and in situ observations from moored buoys, drifting buoys and Argo floats. The CCI dataset was shown to be stable after~1994, with minimal divergence (~0.01 K decade −1 ) between the CCI data and in situ observations. Two steps were identified due to the failure of a gyroscope on the ERS-2 satellite, and subsequent correction mechanisms applied. These had minimal impact on the stability due to having equal magnitudes but opposite signs. The statistical power and false alarm rate of the method were assessed.

An Ensemble Data Set of Sea Surface Temperature Change From 1850: The Met Office Hadley Centre HadSST.4.0.0.0 Data Set

“…On the other hand, there is much to be gained from using the Argo measurements as independent validation. The latter approach is common in the satellite SST community (see, e.g., Merchant et al, , Berry et al, ), where drifting buoy data are often used for calibration and thus cannot be used for validation. We adopt the same approach and reserve Argo for validation of the final product.…”

Observing Requirements for Long-Term Climate Records at the Ocean Surface

“…These measurements were commonly used to estimate ocean heat content (e.g., Levitus et al, 2009), but XBT are problematic for SST analysis due to biases caused by the shock of entry into the water. Recent studies have however (e.g., Gouretski et al, 2012;Hausfather et al, 2017;Berry et al, 2018;Huang et al, 2018) have used these measurements to evaluate SST data products.…”

“…Increasingly CDRs will be constructed from satellite observations without blending with in situ observations, and the same considerations for stability and uncertainty will apply. In situ data are also required for calibration and validation of satellite climate records (e.g., Belmonte Rivas et al, 2007;Berry et al, 2018). As with in situ networks, satellite missions have not always been designed with climate applications foremost, so typically substantial work has to go into constructing stable records (Hollmann et al, 2013;Verhoef et al, 2017).…”

“…The satellite community are important users of in situmeasured surface ECVs (e.g., Belmonte Rivas et al, 2007;Stoffelen et al, 2015;Jackson and Wick, 2016;Kinzel et al, 2016;Berry et al, 2018;Liman et al, 2018;Thorne et al, 2018), and recommendations that aim to improve the quantity, quality, and consistency of in situ data will be of huge benefit for satellite calibration and evaluation.…”

“…High quality observations are important for the detection and evaluation of biases (Hausfather et al, 2017;Berry et al, 2018). Specially designed datasets are also useful, often containing enhanced metadata (Kent et al, 1993;Berry and Kent, 2005), and/or co-located observations from different methods (James and Fox, 1972;Berry et al, 2004).…”