Retrieval Of Wind And Temperature For Middle And Upper Atmosphere Based On Full-closed And Non-full Circular Fringes Of Fabry-Perot Interferometer(FPI)

Fabry-Perot Interferometer(FPI)is widely used for the wind observation of the middle and upper atmosphere.The wind and temperature retrieval algorithm of FPI has been studied internationally.To date,in China,a few studies of FPI wind retrieval based on simulation data have been carried out.In this paper,methods of wind and temperature retrieval are carried out based on full-closed and non-full circular fringes which are both observed from FPIs at Xilong and Kelan sites.The contents and innovations are as follows:(1)Wind velocity retrieval of the middle and upper atmosphere was made using ?data processing? method based on the ground-based FPIs(Xinglong FPI: 40.40 °N,117.59 °E;Kelan FPI: 38.71 °N,111.58 °E),which included the pre-processing,the fringe center determination,the fringe radius calculation and the wind velocity retrieval.For validation,the wind results of 8 days of Xinglong FPI(May 6~13,2010)retrieved from observation data were compared with the FPI wind products with an average deviation of 7.7 m/s(892.0 nm airglow),5.5 m/s(630.0 nm airglow)and 2.7 m/s(557.7 nm airglow)respectively.The average deviations of wind velocities of 28 days of Kelan FPI(9th April ~6th May,2013)for the three airglow emissions are 5.7 m/s(892.0 nm airglow),6.18 m/s(630.0 nm airglow),and 3.66 m/s(557.7 nm airglow),respectively.Furthermore,the detailed analysis of the influencing factors mentioned above was also carried out including airglow intensity,center determination,and focal length.The results demonstrate that the influence of intensity is the greatest among them.Then,a new method is proposed for error calculation considering the influence of fringe radius fitting error and mean standard deviation of several fringes.The radius fitting error includes random fitting error and errors from signal-to-noise(airglow intensity and CCD noise).The relationship between errors derived from the two retrieval methods and airglow intensity are compared and analyzed.It is found that the variation of errors is inversely correlated with the variation of airglow intensity.(2)Wind and temperature retrieval are carried out based on the Airy function.Firstly,the FPI instrumental function is derived from simulation of lidar data using Airy function.Then,the forward model is founded by using convolution of instrumental function and airglow radiance function.Finally,the wind velocity and temperature are derived from the simulation of airglow measurements by use of forward model.In this paper,we rerieved wind velocity firstly,then airglow intensity and noise are calculated,the temperature is retrieved finally.For validation,the retrieval results are compared with FPI products.The averaged deviations of wind velocities for the three airglow emissions are 10.8 m/s(892.0 nm airglow),5.89 m/s(630.0 nm airglow),and 1.44 m/s(557.7 nm airglow),respectively.The relative deviations of temperature between them are 8.43%,6.57%,and 7.45%,respectively.(3)Wind retrieval is studied based on the non-full circular fringe.Firstly,the center of non-full circular fringe is determined based on Taubin method.Then,the fringe is annular summed to improve signal-to-noise,and wind velocity is retieved finally.This method can improve the accuracy of wind retrieval when the airglow is weak.It can supply much more area to image airglow from several directions onto the same CCD simultaneously,which can improve FPI temporal resolution greatly(4-5 times).For validation,the wind results of 28 days of Kelan FPI are compared with FPI wind products which are derived from full-closed circular fringe.The averaged wind deviation between them demonstrates reasonable differences with 5.38 m/s for 892.0 nm airglow emission,5.81 m/s for 630.0 nm emission,and 3.03 m/s for 557.7 nm emission.Mean squared deviations shows a small discrete distribution of retrieval difference with 8.15 m/s for 892.0 nm emission,8.56 m/s for 630.0 nm emission,and 4.58 m/s for 557.7 nm emission.Besides,wind results of Xinglong FPI are compared roughly with measurements of meteor radar which is deployed in Ming Tombs of Beijing(40.3°N,116.2°E).Good agreement demonstrates that this method is robust enough for FPI wind measurements of mesosphere and thermosphere.