The historical data of meteorological paper forms have high scientific research value and important historical preservation significance. Digital archives are the collection, preservation and provision of various digital information resources of paper-based original materials through digital methods. This research proposes a rapid digitization method and system of “paper form scanning imaging”+“image fragmentation processing”+“crowdfunding recognition,” using meteorological big data resources, DBnet model, DSCC algorithm and other technologies to process the scanned images of paper weather forms for image fragmentation. “Crowdfunding Entry” is realized through user login behaviour, and the digital file work of weather paper table data is completed. It has been verified that the accuracy rate of one-recording of the fast digital system is about 99.7%, which is higher than the accuracy rate of one-recording of traditional digital manual input (95.6%); the input time efficiency is improved by 22.2% compared with traditional digitalization. The system ensures data accuracy, improves work efficiency, achieves the rapid formation of digital archives of meteorological paper forms and data, and provides new ideas for digital archives. The COVID19 epidemic has caused difficulties in traditional digital manual aggregation work models. Under the background of big risks, it has certain practical significance.