Recent improvements to the NESDIS SST Quality Monitor (SQUAM)

Satellite-based sea surface temperature (SST) products have been operationally generated at NESDIS since the early 1980s, using data from the Advanced Very High Resolution Radiometers (AVHRR) onboard NOAA and later on MetOp-A satellites. An SST Quality Monitor (SQUAM) was developed in the last years for ensuring quality, stability, and cross-platform consistency of these SST products. The SQUAM is designed to operate in near real-time (NRT) and provides diagnostics at a web-interface for the SST developers and the researchers at: http://www.star.nesdis.noaa.gov/sod/sst/squam/.

Initially, the SQUAM prototype was designed to monitor only the NESDIS SST products, namely the SSTs from the heritage Main Unit Task (MUT) system and the newly developed AVHRR Clear-Sky Processor over Oceans (ACSPO) currently operational in NESDIS. Diagnostics provided by SQUAM are based on statistical analyses of differences between satellite SST (T_S) and a number of global level-4 (L4) reference SST fields (T_R). The underlying assumption is that the probability density function of global SST differences (ΔT_S) is close to a Gaussian distribution (although the distributions of T_S and T_R are highly asymmetric). The initial SQUAM version (v.1) provided diagnostics for MUT and ACSPO SST products from NOAA 16, 17, 18, and MetOp-A from 2004 to the present by comparing against the following global analysis fields: weekly Reynolds-Smith OIv2 SST, two daily Reynolds OIv2 SSTs (AVHRR-based and AVHRR+AMSR-E based), RTG low and high resolution, OSTIA, ODYSEA, and Bauer-Robinson climatology. The analyses are based on ΔT_S histograms, time series of the statistical moments (conventional and robust), global maps and dependence plots (ΔT_S vs. observational and atmospheric parameters such as view zenith angle, water content etc.). Median and a robust standard deviation are used to identify and remove outliers for quality control.

Initial conception and design of the SQUAM prototype was presented before. This work is intended to report the recent developments and additions to it. Based on the initial encouraging results and valuable diagnostics, considerable effort was put to extend the capabilities of SQUAM and to include further a number of operational SST products, with the vision to have a collective diagnostics of various products at one place.

Regarding further data addition, NOAA-19 SSTs were added to SQUAM for both MUT and ACSPO and the temporal coverage was also extended back until 2001. Also, under collaboration with the NAVOCEANO, operational NAVO SSTs are now monitored (for NOAA-14 through NOAA-19): http://www.star.nesdis.noaa.gov/sod/sst/squam/NAVO/. Efforts are also underway to implement SQUAM for daily 1km full resolution area coverage (FRAC) data generated by NESDIS and Eumetsat Ocean & Sea Ice Satellite Application Facility (O&SI SAF), and perform their intercomparison within the SQUAM framework. The intention of adding data from various agencies is to provide a common framework and metrics for intercomparison purposes and eventually move towards reconciliation of all the products towards a high quality benchmark. Comparison of TS with a suite of L4 SSTs (T_R) also convincingly demonstrated significant differences between various L4 products, especially in the high latitudes. To understand these differences, a dedicated L4-SQUAM module was set up, in collaboration with the RTG SST developers at NCEP. Such intercomparison work has also been a main focus of the Group for High Resolution SST Pilot Project (GHRSST; http://www.ghrsst.org) and has been recognized in their webpage as a NESDIS L4-comparison element. The L4-diagnostics are available at: http://www.star.nesdis.noaa.gov/sod/sst/squam/L4/.

Besides ΔT_S histograms, time series of first four statistical moments, maps and dependence plots, further statistical techniques were recently included in the SQUAM. For example, a double differencing (DD) technique was implemented to quantitatively measure the “cross-platform” and “day minus night” consistencies. The DD methodology is based on subtracting two single differences where a more suitable “transfer standard” cancels out providing globally representative estimation of cross-platform consistency. Additionally, Hovmöller diagrams were added to provide a snapshot for time series of zonal differences. Also, interactive plots were added to the SQUAM, which enables the users to focus on satellites or L4 products, and temporal coverage of their choice.

These recent developments and additions will be briefed in this work. Products from the future platforms (MetOp-B, NPOESS, and GOES-R) will also be added to SQUAM.