Abstract:This paper selects the automatic observation data from precipitation instruments of 48 national meteorological stations in Shanxi as well as manual observation records for the time period from December 2017 to November 2018,and makes them analyzed and evaluated from the perspectives of data accuracy and consistency.It can be concluded that the capture rate and consistency rate of various precipitation phenomena show an increasing trend along with the increase of rainfall intensity and the rate of missing report shows a decreasing trend along with the increase of rainfall intensity.The capture rate of rain and snow phenomena is relatively higher and the underreporting rate of drizzle is highest.The rain is prone to be wrongly judged to be drizzle and more wrong judgment will occur between sleet and snow.The precipitation phenomenon recognition algorithm needs to be optimized continuously and a comprehensive judgment is suggested to be done by integrating temperature,humidity,visibility and other relevant factors.