International trade has been considered a critical driving force of materials’ physical flows and their environmental pressures, which has been a global research hotspot. Various databases have been developed and applied to support the physical trade analysis, in which the United Nations Commodity Trade Statistics Database (UN Comtrade) is the original and probably the most widely-used data source. However, data quality issues in UN Comtrade were identified reportedly, which may lead to diametrically conflicted conclusions if not properly addressed. To advance analyses on physical trade flow based on UN Comtrade data, we established an improved methodology to greatly increase UN Comtrade data quality. We applied improved methodology to all available trade records from 1988-2019 (retrieved in 2020/10/4) in 6-digit HS0 codes of all reporters. Our work is designed in two parts, which are (1) detecting and handling outliers, and (2) estimating the missing values, because our previous attempts reveal that the bilateral asymmetry issue caused by misreporting has been highly resolved based on trade data with outliers eliminated and missing values estimated. Details and contributions of our work can be found in our publications.

Downloading the full dataset