Abstract: To enhance the application value of ground meteorological observation data in radish production in Dongying, this study focused on the meteorological sensitive indicators during the growth period of Dongying green radish. A meteorological data quality control technology system and a three-level quality control process adapted to radish production requirements were constructed, and an INLM-PSO missing value repair method was proposed. The results showed that the error rejection rate of the RF anomaly identification method ranged from 1.37% to 1.62% across different growth periods. The INLM-PSO missing value repair method achieved repair completeness rates over 98% at daily, weekly, and ten-day-period scales. In the application evaluation of radish production, the emergence rate during the sowing and emergence period remained above 86%, the rotten seed rate was controlled within 3% to 8%, and the emergence uniformity ranged from 80% to 95%, indicating better overall performance than the KNN neighborhood discrimination method and the improved 3σ threshold method. This meteorological data quality control system could reduce the false rejection and missed detection rates of anomalous data, improve the repair completeness rates of missing data, and provide reliable data support for radish production and field management decisions in Dongying.

Key words: ground meteorological observation, data quality control, random forest, radish production, Dongying