Bias correction of the systematic observation error is vital for the development of longterm gridded sea surface temperature(SST)dataset since 1900.In this study,based on the optimized SR02bias correction method and the global hourly ocean surface observation dataset from National Meteorological Information Center,we have developed the monthly bias-corrected SSTA dataset over the Indian-Pacific Ocean from 1901to 2016,with a spatial resolution of 2°* 2°.The results show that the spatial-temporal distribution of the SST bias derived from our newly developed dataset is generally consistent with the history of SST observation techniques, and also indeed reflects the seasonal variation of SST systematic observation errors.As the threshold of the optimized method varies with the space sample sizes,the bias derived from it reflects more local characteristics,and changes more consistently with transformation of observation techniques as compared with the bias features of ERSST V4.Both the mean bias and root-mean-square error(RMSE)of the bias-corrected SSTA compared with ERSSTv5are smaller than the original one,with varying of reduced mean bias from 37.7%to 87.9%,and decreasing RMSE around 0.06℃.In addition,the comparisons with international products(i.e.,ERSST V5,HadSST3,HadISST1and COBE2)demonstrate that our newly developed bias-corrected SSTA dataset shares high correlations over 0.97with those,and comparable trend features. Except for the coastal region of the East Asia in the higher latitudes,the general differences between our newly developed bias-corrected SSTA dataset and the other international products are mainly between-0.2~0.2℃.