---

This research in this page focuses on using a well known global sst climatology to try and determine the characteristics of the cloud contamination present in sea surface temperature (SST) fields generated from the CLAVR-x clear ocean pixels. While comparison with buoy measurements of the SST can determine the quantitative level of cloud contamination for limited sampling of global SST, it is hoped that comparison to an sst climatology can provide at least a qualitative look at the cloud contamination of  SST's derived from CLAVR-x globally. The SST climatology chosen for this comparison is NCEP's Optimally Interpolated SST (OISST) weekly climatologies  This data-set blends SST fields in space and time to derive smooth climatology that should have minimal cloud contamination.  The CLAVR-x SST refers to the mean SST within a grid-cell averaged from the values derived from the clear pixels.  The CLAVR-x SST standard deviation refers to standard deviation of the pixel values within the grid-cell.  The SST equations used are the MCSST equations. Note, that this analysis is based on the characteristics of SST differences and is therefore not overly sensitive to the accuracy of the SST equations used.  

The analysis is based on showing the behavior of the difference in OISST - CLAVR-x SST  plotted against variables thought to be related to cloud contamination.  The assumptions are as follows.

• cloud contamination should manifest itself as large OISST  - CLAVR-x values
• Due to the uniformity of the ocean surface relative to cloudy scenes, cloud contamination should be related to standard deviation of the clear ch1 reflectance or the pixel level SST standard deviation.
  

The data used for this analysis comes from the week centered on July 20, 1986.  The CLAVR-x data is taken from the NOAA-9 AVHRR.  The OISST is a daily average and the same data is used in the ascending and descending comparisons.  The data resolution is 1 degree which provides approximately 40000 points for each analysis.

Results from Reynolds OISST weekly climatology centered on July 20, 1986 compared to CLAVR-x data from the same period using NOAA-9 AVHRR data.

	NOAA-9 Asc	NOAA-9 Des
Reynolds OISST	view	view
SST derived from CLAVR-x clear pixels	view	view
Difference  (OISST - CLAVR-x)	view	view
Std. Dev. in CLAVR-x SST	view	view
OISST - CLAVR-x vs. cloud amount	view	view
OISST - CLAVR-x vs. OISST	view	view
OISST - CLAVR-x vs. CLAVR-x SST	view	view
OISST - CLAVR-x vs. grid-cell SST std. dev.	view	view
OISST - CLAVR-x vs. grid-cell ch1 reflectance std. dev.	view	view*

* - less points due to fewer day-time grid-cells in descending orbit.


Conclusions from these Results

• The NOAA-9 SST equations used are not optimal, there is a 2K bias evident from the figures (the SST difference intercept with the y-axis in each figure)
  
• There is no obvious correlation with OISST - CLAVR-x SST with any of the variables plotted that behaves in a manner consistent with cloud contamination.
• There seem to be many positive and negative differences (OISST - CLAVR-x) - this is puzzling and can not explained  by CLAVR-x cloud contamination solely.
  

---

NESDIS/Satellite Applications and Research Home | CIMSS Home | SSEC Home