Research On Retrieval Methods And Application From Microwave Radiometer In Semi-arid Area

Ground-based microwave radiometer is a new means, it not merely supply a gap of sounding data but also offer monitoring capability of artificial precipitation. But microwave radiometer observed accuracy of atmospheric temperature and humidity and water vapor content and cloud liquid water content is a difficult problem.Using TP/WVP-3000 microwave radiometer of semi-arid climate observatory and laboratory of Lanzhou University and sounding data and ground-based observation data of Yuzhong station in 2009, comparative analyses is made to the meteorological elements retrived from sounding data and microwave radiometer data, evaluate of microwave radiometer inversion method, and basing on back propagation neural network and regression algorithm separately calculate atmospheric temperature and humidity and water vapor content and cloud liquid water content get to preferable inversion method of semi-arid region and analysis their characteristic and primary study the capacity of precipitation forecasts of microwave radiometer in semi-arid region. The main research contents of this article are as follows:(1) Using sounding data analysis quality of atmospheric temperature and humidity and water vapor content and cloud liquid water content of microwave radiometer autologous inversion, comparative analyses sounding data and microwave radiometer data. Results are as follows:correlation coefficient of atmospheric temperature of microwave radiometer inversion and sounding is between-0.41 and 0.96, MAE of temperature is between-12.15℃ and 13.48℃, RMSE of temperature is between 1.07℃ and 18.01℃; humidity is between-0.06 and 0.68, MAE of humidity is between-23.24% and 46.97%, RMSE of humidity is between 11.11% and 55.84%; correlation coefficient of the water vapor content is 0.9348; correlation coefficient of the liquid water content is0.2387 and correlation of winter is worst.(2) Basing on back propagation neural network inversion atmospheric temperature and humidity profile and this back propagation neural network better suits semi-arid region. Results are as follows:Back propagation neural network inversion accuracy is superior to microwave radiometer autologous inversion accuracy, correlation coefficient of back propagation neural network and sounding calculate atmospheric temperature is between 0.57 and 0.96, MAE of temperature is between-1.77℃ and 2.40℃, RMSE of temperature is between 0.78℃ and 11.73℃; humidity is between 0.39 and 0.91, MAE of humidity is between-7.91% and 11.74%, RMSE of humidity is between 7.06% and 27.82%.(3) The characteristics of atmospheric temperature and humidity were analyzed. Results are as follows:Temperature and humidity have obvious diurnal variation, Diurnal variation of temperature weaken over height, winter’s probability of appearing temperature inversion is more during the 0 to 11 hour; humidity reached maximum during the 8 to 9 hour and minimum during 19 to 20 hour; fog and haze weaher appear temperature inversion, fog and haze can increase the intensity of inversion close to ground, the haze wear off when the temperature inversion decreases and the humidity increases above ground, the fog wear off when the intensity of inversion decreases and the humidity increases.(4) basing on regression algorithm calculate water vapor content and cloud liquid water content and comparative analyses the water vapor content and cloud liquid water content from sounding and microwave radiometer. Results are as follows:correlation coefficient of atmospheric water vapor content of microwave radiometer inversion and sounding is 0.9589, and cloud liquid water is 0.3637 and different seasonal correlation is defferent. The correlation of winter’s water vapor content was worst and value is 0.5504 and the correlation of summer’s cloud liquid water content was worst and value is 0.1780.(5) The characteristics of atmospheric water vapor content and cloud liquid water content were analyzed. Results are as follows:the water vapor content and cloud liquid water content have weak diurnal variation and had apparent seasonal. The maximum value of diurnal variation of water vapor content is 0.25 cm and that of cloud liquid water content is 0.14 mm. The mimum value of water vapor content and cloud liquid water content is in January and the maximum value is from june to September.(6) The value of water vapor content reach 2.20 cm can be used as a threshold of precipitation forecast. Results are as follows:when the value of water vapor content is more than 2.20 cm within 24 hours before precipitation and its variability reach 0.16 cm/h or the variability of the water vapor content exceeds to 0.19 cm/h, the precipitation occurs; The cloud liquid water content attain 0.20 mm can be used as a threshold of precipitation forecast, the turning-point of cloud liquid water content will appear before precipitation as time changes and the variability begin to increase, the precipitation will come up for 1-hour forecasting in the future.