| The Global Ocean Data Assimilation Experiment (GODAE) requires the sea surface temperature (SST) data to be accurate, be available in near real time and have high-resolution spatial (at least 10km) and temporal (6-12 hours) resolution. At present, merged SST products are obtained by optimum interpolation which have higher coverage than individual satellite SST data. However, the merged data are smoothed and hence the fine oceanic features are lost due to the interpolation. In order to meet the need of high resolution coastal SST data, AVHRR and AMSR-E SST data are merged by wavelet transform in this study.Firstly, AVHRR and AMSR-E SST data in the China Seas and Northwest Pacific during the period of 2005 are compared with in situ SST data respectively. The bias and standard deviation of the difference between in situ data and AVHRR SST data in day and night time are 0.18+0.77°C and 0.3+0.81°C respectively; those of in-suit data and AMSR-E SST data in day and night time are 0.1+0.56°C and 0.09+0.61°C respectively. The relatively large error of AVHRR SST data are mainly caused by the error of cloud detection.Secondly, AMSR-E and AVHRR regional SST data in the area between 10°N-50°N and 105°E-145°E are merged by wavelet transform and the spatial resolution of results is about 4km. The oceanic coverage of the merged SST data is more than the coverage of observations from single sensor. The annual mean coverage values of the AMSR-E, AVHRR and merged SST data are 40%, 34% and 55% respectively. The comparisons of entropy, variance and definition of the images show that merged SST data well preserve the fine oceanic features appeared in the AVHRR SST images. The bias and standard deviation of difference between in situ and merged SST data in day and night time are 0.07+0.92°C and 0.2+0.92°C respectively.
|
PDF Full Text Request