306
Integrated cloud mask and quality control for GOES-R ABI SST: prototyping with MSG/SEVIRI

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 20 January 2010
Nikolay Shabanov, NOAA/NESDIS, IMSG Inc, Camp Springs, MD; and A. Ignatov, B. Petrenko, Y. Kihai, and A. Heidinger

Handout (2.1 MB)

Geostationary Operational Environmental Satellite-R Series (GOES-R) is the key future NOAA geostationary platform to carry Advanced Baseline Imager (ABI) onboard. A suite of algorithms for retrievals of environmental products is currently being developed by the ABI Algorithm Working Group (AWG), and individual products are being integrated into the AWG framework, which coordinates end-to-end product generation. The ABI Sea Surface Temperature (SST) product is being developed by the SST team at NOAA/NESDIS using the Meteosat Second Generation (MSG) Spinning Enhanced Visible and IR Imager (SEVIRI). The Advanced Clear-Sky Processor for Oceans (ACSPO) was initially developed for the Advanced Very High Resolution Radiometer (AVHRR) and tested with data from multiple polar satellites (NOAA-16 through 19 and MetOp). ACSPO retrieves SST from clear-sky radiances in the thermal infrared bands. This presentation is focused on one key issue in the current stage of ABI SST product development: integration of the upstream cloud mask (CM) module implemented by the AWG Cloud Application Team with the SST Quality Control (QC) module implemented internally by the AWG SST Application Team. The presentation will discuss synergism and integration of the CM and QC, using two months of MSG/SEVIRI data in July 2008 and January 2009.

Reliable cloud identification is critically important for accurate retrievals of clear-sky SST product. ABI CM module is built upon heritage systems (AVHRR, MODIS, and others) and is mainly based on general knowledge of cloud emission and reflection properties. CM algorithm detection comprises a system of 30+ tests and detects clouds both over land or ocean. The system output contains results of the individual cloud tests, which are further combined into a separate cloud flag with four states: (1) confidently clear, (2) probably clear, (3) probably cloudy, and (4) confidently cloudy.

Product-specific optimization of the CM and additional QC may potentially enhance product performance. By design, CM is relatively liberal to avoid over-screening or false alarms, while QC is more conservative to avoid cloud leakage in the product. While the CM module is executed prior to the SST module, the QC module is implemented downstream the SST module. The objective of the SST QC is to assess the SST retrieval's accuracy, degraded by various environmental factors (possible residual cloud and aerosols contamination, sun glint, radiometric noise, extreme observation geometry, proximity to coast, etc.). Thus, the implementation of the QC module is different from CM. It significantly relies on using prior SST information, including Reynolds SST (daily or weekly) and GFS upper atmosphere fields.

Two categories of tests are currently implemented in the QC module. The first category, “Point Tests,” includes an SST test (comparison of the retrieved and reference SST), BT test (comparison of sensor-measured BT with BT predicted by the radiative transfer model), and optical depth test (comparison of retrieved optical depth with that predicted by the radiative transfer model). The second category, “Texture Tests,” includes an SST spatial uniformity test (comparison of spatial variability in retrieved SST with that expected around the pixel) and ambient cloud test (comparison of the amount of nearby cloudy pixels with the expected amount). The system output contains results of the individual QC tests, which are further aggregated into separate QC with several states, from best to worst accuracy (“excellent,” “acceptable,” “suspect,” and “bad”).

This presentation provides details of the analysis of the CM and QC modules and addresses three main questions: (1) how both modules perform individually, (2) how the CM module can be improved for SST application, and (3) how the QC module complements and enhances the CM module. Special emphasis is given to sensitivity studies and ranking of individual CM and QC tests performed over the diurnal cycle, as well as defining the interface between the CM and QC for the ultimate integration of the SST sub-system into the AWG framework.