Two kinds of unconventional detection data from GPS and automatic weather stations and the role in weather forecast are analyzed. The result shows that the unconventional data can compensate the defects for conventional detection data in temporal and spatial scales. There is good corresponding relation between the time evolution characteristics of GPS/MET data and rainfall, and the distribution characteristics of precipitation are different for different weather systems, different types of precipitation and in different seasons, which can be used as important references for precipitation process analysis and short time forecasting and warning. Automatic weather station data are useful and important in analyzing the strength, scope, the maintain time, and evolution characteristics of meso and micro scale weather systems in local rainfall forecasting and warning.