Quality Control, Error Estimation And Application Research Of GPS Occultation Data

Global Positioning System(GPS)radio occultation(RO)is a kind of active remote sensing satellite data,which provides atmospheric observations with global coverage,high-precision,high vertical resolution,all-weather availability and long-term stability through limb detection.Due to its unique characteristics,many studies have proven its great value in researches on ionosphere,climate,and meteorology,as well as in operational weather forecasting.However,the application of the RO data in lower troposphere is still a challenge.At early stages,only RO data above 4-km altitude were used in operational weather forecasting,and until now most reanalysis datasets only assimilates RO data above the trapping layer(ducting,where strong negative refractivity gradients occur).Therefore,how to properly process and maximize the usage of RO observations,especially the data in lower troposphere,always attracts great attention in RO research.The RO observations processed by the Radio Occultation Meteorology Satellite Application Facility(ROM)and COSMIC Data Analysis and Archive Center(CDAAC)are quality controlled by a strict and targeted procedure,and the similarities and differences of the RO data from the two center after quality control(QC)are objectively and quantitatively compared.Then the“three-cornered hat”(3CH)method is used to estimate the observation error of both RO bending angle and refractivity observations on a global scale and the three-dimensional distribution of RO observation errors is given for the first time.Subsequently,the specific factors affecting the RO detection depth are studied using the COSMIC-2 RO observations,and finally the vertical thermal structure of tropical cyclones(TC)is explored using the quality-controlled COSMIC-2 RO observations.The main contents and conclusions of this paper are as follows:(1)Strict sequential QC procedure are conducted on the RO observations from ROM and CDAAC,including:range check,biweight check of refractivity observations,biweight check of deviations between RO observation and European reanalysis and biweight check of ROM bending angle vertical sampling interval.The last QC step is proposed according to the height-dependent sampling interval of ROM bending angle data.The QC procedure effectively eliminates the outliers of RO observations,and the mean fractional difference between ROM-and CDAAC-processed bending angle is reduced from±0.4%to±0.2%after QC.For the researches on RO application,we recommend using data processed by CDAAC,which provides RO data from most international missions;For the researches related to RO processing,e.g.error tracing and data retrieval,the RO data from ROM will be more suitable since ROM also provides the RO processing package for the public.(2)The 3CH method is used to estimate the observation errors of RO bending angle and refractivity observations near 48 radiosonde stations in China and 6°×6°grid on the adjacent ocean,and the variation of RO observation errors on land and ocean and their differences are analyzed.Results show that the RO observation errors near the 48 radiosonde stations show significant latitude dependence,that is,greater at lower latitudes.The RO bending angle observation errors distribute consistently with the bending angle uncertainty provided by CDAAC.Then the 3CH method is applied to 521 radiosonde stations around the world,revealing for the first time the global three-dimensional distribution characteristics of RO observation errors for both bending angle and refractivity.The observation error distributions of bending angle and refractivity are similar,roughly symmetrical between the northern and southern hemispheres.Specifically,the observation error is the largest near 15°N/S below 700h Pa,largest at the equator above 700 h Pa,decreasing as the altitude increases at all latitudes.(3)Using COSMIC-2 RO data in two areas on the ocean,we examine the RO observed lowest altitude and its possible relationship to refractivity gradients and planetary boundary layer(PBL)heights.Results show that the strong vertical gradient of N-N(r)(<-65N-unit km^-1),where N(r)is the vertical profile of the spherically symmetric refractivity,at the PBL top is the main reason preventing the GPS RO signal to propagate at lower altitudes.When the vertical gradient is weak,the strong tangential horizontal gradient of refractivity and the absorption and scattering effects of clouds will also affect the detection depth of RO rays.Fortunately,there are only about 0.6%COSMIC-2 RO profiles whose lowest observing altitudes do not reach below 2-km altitude.(4)The COSMIC-2 RO data collocated with the best track of all TCs in 2020 are used to statistically analyze the vertical structures at different intensities and in different radius ranges from the TC center.Results show that the mean heights of boundary layer and cloud top of tropical depressions are about 2.5 km and 16 km,respectively,while the boundary layer heights and cloud top heights of tropical storm and hurricane categories is about 0.5 km and 1km higher,respectively.In addition,the RO-retrieved temperature is used to characterize the vertical temperature structure evolution of Hurricane Laura(2020).It reveals good correspondence between the variation of the warm core and the intensification of Laura.