| In order to guarantee the observational data quality of Automatic Weather Station (AWS), a method for calibrating data-acquisition unit temperature channel is studied and the corrected temperature data uncertainty is evaluated in this paper.According to the meteorological industry standards and technical specifications of AWS, data-acquisition unit temperature channel performance is analyzed briefly. A method of utilizing active resistance network to calibrate the temperature channel is proposed and a calibration experimental scheme is designed.In order to correct the temperature channel data of data-acquisition unit, a fitting regression algorithm base on improved least squares support vector machines (LSSVM) is presented to obtain the changing regular pattern of system errors. In this algorithm, adaptive genetic algorithm is used to look for optimal parameters, establishing optimization model. With the UCI standard data simulation, the modeling effect of this improved algorithm is tested. The simulation result shows that the modeling accuracy using improved LSSVM is higher than traditional LSSVM. Based on this algorithm, fitting model of systematic errors is established to correct the verified data of temperature channel. And the result shows, the errors of correction temperature data are controlled in0.03℃, far less than the initial errors.Finally, the uncertainty of temperature correction data is evaluated to character credibility and dispersion of measurement value. The mathematical model of measurement results is established and the uncertainty components introduced by influence variables are assessed. Then the combined standard uncertainty and expanded uncertainty are calculated.In practical applications, calibrate data-acquisition unit temperature channel regularly utilizing the calibration method designed in this paper and correct the measurement data in data-acquisition unit internal procedure or automatic weather station data processing software base on the pattern of systematic errors can improve the temperature observational data accuracy of AWS. |
