Studies On The Orbit/Clock Combination Of IGS Analysis Centers And Kinematic Precise Point Positioning Based On Local GNSS Network

IGS products can provide support for scientific activities such as improving andexpanding the International Terrestrial Reference Frame maintained by IERS, monitoring thedeformation of solid earth, changes of liquaid earth and the earth’s rotation, troposphere andionosphere monitoring. IGS products as free for users are achieved by combining severalanalysis centers’ products which are calculated independently. IGS orbit and clock productsare the basic data of precise point positioning, that are also the key products of maintainingthe coordinates and time reference frame of GNSS. Some issues on the orbit and clockcombination are studied in this thesis. Besides, the rapid development of CORS networkconstruction makes it possible that adopting the useful information provided by local networkand the integer constraint of the double-difference ambiguities. The algorithm of kinematicprecise point positioning (PPP) based on GNSS local network are also studied in this paper.The accuracy and reliability of kinematic PPP are greatly improved. The main content of thisthesis is as follows:1. Several weighted methods on IGS orbit and clock combination are proposed.They areanalyzed and comparied through data processing.Combination orbit and clock are calculatedby weighted average.The prior weight of analysis center (AC) determines the result ofcombination directly.2. Two kinds of robust methods of parameter estimation are studied. One is L1-normestimation which is used by IGS. The other is RLS estimaiton, which is proposed by thispaper. We analysised the characteristics of the two methods and present their mathmaticalmodels and algorithms in details. The numerical example of the two methods is given andcomparied..3. In view of the outlier of AC clock and orbit data, we constructed robust estimationwith high-breakpoint of pollution rate based on the robustness of median and the idea ofdouble eliminated weight function. The caculation results of this algorithm show a higherrobustness and higher efficiency.4. In view of the clock jump of satellite broadcast clock, a linear regression model fittingthe difference value of before and after the epoch of satellite broadcast clock is proposed. Weget clean reference clock data though eliminating satellite clock correction with clock jump.5. Two methods of adaptive selection of reference analysis centerc for ultra-rapid clockcombination are proposed. One is to select the analysis center whose epoch number of themedian is the largest. The other one is to select the analysis center whose statistical accuracy of the present phase is the highest.6. The mathematical model and algorithm of kinematic PPP based on GNSS localnetwork is studied. The accuracy of kinematic PPP can reach2cm though adopting the usefulinformation provided by local network and the integer constraint of the double-differenceambiguities.