Products and Library
|Monthly Mean SST of MTSAT-1R|
Sea Surface Temperature (SST) is important information for meteorological
services and fishery since it represents ocean currents and ocean fronts,
and is closely related to the generation of clouds and sea fog in coastal
areas and over the ocean. Grid-point values of SST are calculated every
three hours from MTSAT-1R imagery in the area from 60° N to 60°
S and from 80° E to 160° W with a 0.25-degree square mesh using
the Multi-channel SST (MCSST) retrieval algorithm (McClain, E.P., 1981)
modified for MTSAT-1R.
The SST grid-point values are sent to JMA Headquarters, where they are operationally used for issuing information on oceanographic and climatological monitoring.
The terrestrial radiation emitted in the spectrum 10-12 µm is observed
by MTSAT-1R as infrared imagery, and provides information on the temperature
of the underlying surface and/or clouds. In this region, most of the radiation
originated from the Earth's surface and/or clouds toward satellites is
not greatly modified by the atmosphere. The region is known as the atmospheric window since the atmospheric absorption in it is small. The brightness temperature
in the spectral band of 10.3-12.5 microms is used for SST estimation, and
is slightly attenuated due to the absorption of atmospheric constituents
such as H2O, O3 and CO2. In this regard, it is necessary to correct atmospheric absorption to
estimate SST precisely.
The infrared band of MTSAT-1R's JAMI is divided into two channels, namely IR1 and IR2, which are known as split-window channels. The brightness temperatures in these channels are differently observed according to variations in absorption efficiency. The brightness temperature difference can be used for atmospheric correction since the attenuation of brightness temperature is almost linearly related to such variations. SST is calculated using the brightness temperature of IR1 and the difference in the brightness temperatures of IR1 and IR2 taking into account the satellite zenith angle using multiple linear regression.
SST is calculated using the MCSST algorithm, which involves three processes: (1) cloud filtering; (2) SST calculation; and (3) quality check and mapping charts.
In cloud-contaminated areas, SST is estimated less because clouds cover
the sea surface and the radiation from the colder tops of clouds is significant.
SST is therefore calculated in cloud-free areas where the sea surface is
visible, and cloud-free pixels are discriminated from MTSAT-1R imagery
with the original resolution using a cloud-filtering algorithm.
In this algorithm, the threshold levels for the brightness temperature of IR1 and the brightness temperature difference between IR1 and IR2 are established to enable discrimination of cloud-free pixels.
The visible channel of JAMI is also used for cloud filtering during the day.
SST is calculated from the brightness temperatures of IR1 and IR2 in cloud-free pixels using the following equation of the MCSST retrieval algorithm:
SST= a · T11 +b (T11-T12) +c (T11-T12) · (sec θ -1)+d
where T11 is the brightness temperature of IR1 at 10.3-11.3 µm, T12 is the brightness temperature of IR2 at 11.5-12.5 µm, θ is the satellite zenith angle, and a, b, c and d are coefficients of the multiple linear regression. These coefficients are determined twice a day from simulated brightness temperatures of IR1 and IR2, which are derived in advance by radiative transfer calculation (MODTRAN) using JMA Numerical Weather Prediction (NWP) atmospheric profiles, to get rid of the regional and seasonal biases. The coefficients are therefore changed every day.
In order to eliminate unreasonable values, the calculated SST is compared with climatological values and JMA's headquarters' analysis SST data as a quality check. The mean of the reasonable values is calculated every three hours for each 0.25-degree square in latitude and longitude, and is used to represent the SST within that area.
The satellite-derived SST is regularly compared with SST measured by buoys in the MTSAT-1R coverage area. The root-mean-square error of SST is between 0.8 and 1.2 K and the bias is within ±0.5 K.
At the MSC, the quality of meteorological products including MTSAT-1R's SST, is regularly checked.