Core demands
Surface meteorological observation is one of the important means for meteorological departments to obtain meteorological information. Conventional meteorological elements (such as temperature, pressure, relative humidity, wind direction, wind speed, precipitation, etc.) have been automatically observed by instruments, but some visual observation items (such as cloud shape, cloud amount, icing, frosting, etc.) have always been identified by the human eye, and due to the large differences between individual observers, Subjective factors affect the objective facts of cloud shape, cloud cover and weather phenomena. The weather system is changing rapidly, and the continuous observation with high space-time density cannot be realized by manual observation. The traditional artificial meteorological observation has been unable to meet the demand for high-precision meteorological services in the fast-paced modern society.
Solution
1 - Hangzhou National Benchmark Climate Station regularly acquires high-definition images through high-definition all sky fish eye camera installed on the top of the pole and ground grass monitoring camera.
2 - Image data is called through API, and Baidu Brain EasyDL model interface is called in real time for intelligent recognition, and the recognition results of cloud amount, cloud shape, weather condition, frost and dew are returned. The result information will enter the meteorological information system for meteorological forecast and service.
After the system is put into business promotion, ordinary users may be able to query real-time all sky images and AI recognition results of any city through mobile weather APP software.
