Abstract
A time series dataset is important to understand trends such as SST and NDVI fluctuation in
global changes and satellite observation data can be used for those purposes. However, how to
generate a time series of cloud free dataset with satellite observation data is a big problem. A
time composite method, typically 10-day Maximum Value Composite method (lOdayMVC) is
used commonly for the solution but it does not guarantee to complete it.
A NRF (Noise Reduction Filter) was developed by authors to implement the lOdayMVC
dataset and is adapted for a SST-lOdayMVC dataset that was generated from a time series of
Defense Meteorological Satellite Program (DMSP) / Operational Line Scan System (OLS)-
Thermal Channel (TIR) data. A cloud free SST dataset was generated by NRF and was evalu
ated by using the lOdayMean SST of the Japan Meteorological Agency. Significant reduction of
cloud influences was confirmed of the dataset.
In this study, thermal data from the TIR were converted to SST by an algorithm using regres
sion analysis method, and the multi-channel SST (MCSST) derived from the Advanced Very
High Resolution Radiometer (AVHRR) carried on the National Oceanic and Atmospheric
Administration polar orbiter series of satellite (NOAA-14) was used as standard data for this re
gression analysis.
