An Extended Triple Collocation Method With Maximized Correlation for Near Global-Land Precipitation Fusion

An Extended Triple Collocation for maximized Correlation (ETCC) method was proposed with a unique correlation function, the purpose of which is to maximize the correlation between the merged product and unknown truth. The method was tested over quasi-global land by combining three independent precipitation products. The performance of the ETCC-merged product was then evaluated against three reference data sets and compared with the existing Triple Collocation (TC) merging. The merged product was found to be generally superior to each contributor. Moreover, the ETCC method is better able to improve the correlation of merged product compared with the TC approach. Other improvements are also shown in the absolute difference of the ETCC-merged product, such as regional validation for central North America and mainland China. These demonstrate the effectiveness of the ETCC method, and accordingly, it can provide a promising solution for maximizing the correlation of merged product without the truth. Precipitation plays a significant role in global atmospheric and hydrological modeling systems. Owing to uncertainties in precipitation estimates from a single satellite or reanalysis product, it is necessary and challenging to blend multi-source precipitation information to improve the quality of precipitation estimates. The present study proposed a new merging method, called Extended Triple Collocation for maximized Correlation (ETCC), which aims at maximizing the correlation of the merged product against unknown truth. The ETCC-merged product combined three independent precipitation information derived from the satellite and reanalysis products, and then, was evaluated using three reference data sets. Meanwhile, we compared ETCC's performance with that of the original Triple Collocation (TC) merging. Results show that ETCC method not only performs better than each contributor, but also is more effective in improving precipitation accuracy relative to the TC approach especially in central North America and mainland China. A new method is proposed for maximizing the correlation between the merged product and unknown truthThe merged product is more accurate than each contributorThe new method performs better than the original triple collocation

Automated summary

An Extended Triple Collocation for maximized Correlation (ETCC) method is proposed to maximize the correlation between the merged product and unknown truth. The method combines three independent precipitation products and outperforms the existing Triple Collocation (TC) merging. Regional validations in central North America and mainland China demonstrate the effectiveness of the ETCC method.