Abstract: During the development of convective storms, their geometric characteristics, such as shape and size, are prone to significant changes. Therefore, accurate recognition of storm contours is crucial for correctly assessing the relationships between adjacent convective cells. Based on the multi-layer reflectivity factor data from Chengdu New Generation Weather Radar during May to October 2021, the traditional algorithms for storm cell recognition, tracking, and extrapolation were improved. Key improvements were made to the storm cell contour recognition, tracking matching, contour extrapolation, and centroid extrapolation algorithms. The performance of the improved algorithms was evaluated. Results show that compared to traditional methods, the proposed algorithm can accurately identify the contour features of storm cells, providing richer and more detailed radar echo characteristics. Furthermore, by introducing a storm matching index, the new algorithm enhances the accuracy and objectivity of the storm tracking matching process. For radar echo extrapolation forecasting, the new algorithm integrates characteristics such as the profile, structure, and centroid position of storm cells, enabling more precise predictions of the moving path, shape characteristics, and future trends of the storm body. The average prediction error distance was reduced to 2.7~3.11 km within the 6 to 60-minute forecast timeframe.