Quality control of satellite-retrieved sea surface temperature

Qianguang Tu,Delu Pan,Zengzhou Hao,Haiqing Huang,Fang Gong,DongSheng Shi

DOI: https://doi.org/10.1117/12.2022704

2013-01-01

Abstract:The blended sea surface temperature (SST), which generated from satellites-retrieved SST, is widely used in the fields of oceanic and atmospheric researches. Due to the quality of satellites-retrieved SST will affect the blended SST, the quality control (QC) is necessary and important. In general, the quality of data is controlled by the in situ observations. However, the in situ SST observations are sparse and not available in near real time over the globally ocean, especially for the China Seas and their adjacent seas. In this paper, a complementary quality control procedure, which use the Optimal Interpolation SST as a reference standard (T-R) to identify outliers in infrared SST (T-S) is proposed. The T-R is validated against in situ SST first. Then a time evolution check for T-S is employed. The T-S lies between the limit checks, which are defined relative to TR of the previous 10 days. Spatial-coherence analyses of the differences (Delta SST) between the T-S and the T-R is taken into account later on. Then, robust statistics is applied to flag the extreme residual outliers. After those QC procedures for MODIS-retrieved SST, most of the outliers are removed. The histogram of Delta SST is strong asymmetry and the minimum value reach similar to-35 degrees C mainly due to the cloud contamination before QC. The corresponding histogram after QC shows that the Delta SST are close to Gaussian and the min and max Delta SST reach similar to +/- 4 degrees C. The further validation for this method is performed using a total number of 506 matchups of buoy and MODIS. The bias is -0.458 and the standard deviation is 1.341. This QC procedure can effectively remove the outliers and the remaining observation errors are mainly due to diurnal variability, which should be focused on in the future study.