Abstract:

This study utilized paper−based annual reports (1981−2012) and electronic annual reports (2013−2022) from 653 national agro−meteorological stations across China. Data from 618 stations with continuous yield observations for seven major crops−wheat, rice, maize, oil−seed rape, cotton, soybean and peanut were selected. A comprehensive processing workflow−including standardization of crop names and observation items, imputation of missing values, data series alignment, unit and metadata harmonization and rigorous quality control and validation was applied to develop a high−quality dataset of major crop yields in China for 1981−2022, supporting agricultural climate change research. Results showed that the data availability rate for actual yield observations exceeded 91.0% across all crops, with the accuracy rate >97.0%. Among the 20 yield factor variables, all except winter wheat overwintering mortality rate had availability rates >78.8%, and all exhibited accuracy rate >97.0%. For the 60 yield structure variables, availability rate was >88.6% and accuracy rate exceeded 90.2%. Targeted data correction significantly improved data reporting rates: winter wheat overwintering mortality increased from 12.2% to 61.7%, and maize double−ear rate rose from 52.8% to 95.9%. This dataset provides standardized, high−quality baseline data for agro−ecological studies, climate change research, disaster risk reduction, and agricultural climatic zoning. The comprehensive quality control framework established in this study offers valuable insights for enhancing the quality of agrometeorological datasets.

Key words: Crop dataset, Actual yield, Yield structure, Yield factor, Quality control