[ADEOS Home] [Directory] [What's New of the ADEOS web pages] [EORC Home]

Documents & Paper
OCTS Global SST Val Report

Report of OCTS Ver. 3.1 (Ver. 4)
Global Sea Surface Temperature Data

Hiroshi Murakami, October 23, 1998

Ocean Color and Temperature Scanner (OCTS) Version 3.1 Level 3 Binned Map (L3BM) Daily Sea Surface Temperature (L3BMSST) data is validated by comparing it with Marine Environmental Data Service (MEDS) Drifter data.

L3BMSST data were matched with MEDS drifter data by the following criteria. 1) Their date is the same and the L3BMSST grid is nearest to drifter location. 2) When more than one drifter data are set in one date and one grid under the above criteria, a value with local time nearest to 10:30am (OCTS observation time) is selected. 3) L3BMSST data which vary by [Plus Minus]3 K from the drifter data are eliminated to exclude irregular values which may be caused by the sensor, processing or cloud masking errors. Figure 1 shows results of the "original data" comparison.
OCTS was in "Tilting" operation from 1 November to 19 December 1996, and from 19 March to 29 June 1997. The tilt angle was changed from 0 to -20, -20 to 20, and 20 to 0 degrees by 60, 0, and -60 degrees north from a latitude just under the sun, which change seasonally, to avoid sun glint in this period. The tilt changes seem to cause some SST bias in comparison with the Japan Meteorological Agency (JMA) Global SST (GLSST) data (Figure 2).
We can identify a gap of deviation at the tilt change latitude in the tilt months (Nov., Dec., Mar., Apr., May, Jun.) in Figure 2. These gaps will disturb some analysis of the global and tropical SST except for in mid-latitude of the Northern Hemisphere.
This erroneous bias can also be seen in comparisons with the MEDS drifter SST (Figure 3).
The deviations increase positively from the north part to the tilt change latitude (0 degrees), and suddenly become negative and increase negatively in the southern part. We are searching for the cause of the bias but have not been able to identify it yet (it may concern blackbody calibration).
For the present usage, we will try to revise the SST values using the relation between the bias and the angle from the tilting latitude. The relation is shown in the following formulas (least square regression which pass y=0 at x=60 and -60 degrees) and green lines in Figure 3.

SSTrevised=SSToriginal-(-0.0116948*(x-60)) (60>x>0: Northern part)
SSTrevised=SSToriginal-(-0.0172867*(x+60)) (-60<x<0: Southern part)
x=-23.4*cos(2*3.14159*(cdy+6)/365) (: Tilting Latitude [degree N])
cdy: Year-days from 1 January

The deviation in Figure 3 is improved as shown in Figure 4 by applying these formulas.
Scatter diagrams of Figure 1 are also improved (Figure 5).
The plot in Figure 2 is improved to that in Figure 6.
For example, we prepared an anomaly SST map based on JMA climatological GLSST. The quality of revised L3BM SST data seems adequate for analyzing anomaly SST fields.
Conclusion and Remarks
OCTS L3BM daily SST data is validated using MEDS drifter Data. The comparison between original L3BMSST and MEDS data reveals bias=0.01 and RMSE=0.82, and there is an erroneous bias which is caused by OCTS tilting. Applying regression formulas by degrees from the latitude of tilt switching, they are improved to bias=0.05 and RMSE=0.74. This accuracy is comparable to SST estimates by AVHRR and other thermal infrared sensors. However the remaining error is unsatisfactory and some part is assumed to be caused by the dependence on the scan angle of OCTS. Further investigation will be needed to improve the OCTS SST accuracy.

In this report, we used MEDS Drifter Data and JMA Global SST data for the comparisons. The MEDS drifter data were provided by MEDS to National Space Development Agency of Japan (NASDA) for OCTS Calibration and Validation activities and edited by Akira Mukaida in the Remote Sensing Technology Center of Japan.

Appendix ---------------------------------

Users frequently ask about grid coordination of L3BM and Intensive LAC (ILAC) data, and how to convert L3BM ILAC count values to physical values.

1. L3BM Data

 Size: 4096x2048, 1byte per pixel
 Area: 90N to -90N (North -> South), -20E to -20E (West -> East)
 Lon E  =160+360/4096*(n-2048.5), n=1....4096
 Lat N  =180/2048*(1024.5-m), m=1....2048

2. ILAC (L3M) Data

The cut-out data and Level 3Map data are displayed Mercator
projection (default). 
The pixel and line number of a location are calculated from
its latitude (Lat, degree) and longitude (Lon, degree) :
Lo0=Scene Upper Left Longitude   (degree) 
La0=Scene Upper Left Latitude    (degree)
(Lo0, La0) are correspond to (pixel No. 1, line No. 1)

3. Conversion methods

Count value (0 to 255) -> Chlorophyll-a concentration
 Chl [mg/m^3]=10**{0.015*Count-2.0}

 Count value (0 to 255) -> Sea Surface Temperature
 SST [degrees Celsius]=0.15*Count-2.0
 (SST [K] =0.15*data+271.15)

Link Button Click Here Figure 17 Images Page !!

- Download Figure -

************************************************************** Hiroshi Murakami E-Mail : National Space Development Agency of Japan (NASDA) Earth Observation Research Center (EORC) **************************************************************

[ADEOS Home] [Directory] [What's New of the ADEOS web pages] [EORC Home]

Earth Observation Research Center

Comments to:
Last Update: 27 October 1998