Improvements in GPS precision: 10 Hz to one day

Seeking to understand Global Positioning System (GPS) measurements and the positioning solutions in various time intervals, this dissertation improves the consistency of pseudorange measurements from different receiver types, processes 30 s interval data with optimized filtering techniques, and analyzes very-high-rate data with short arc lengths and baseline noise.; The first project studies satellite-dependent biases between C/A and P1 codes. Calibrating these biases reduces the inconsistency of satellite clocks, improving the ambiguity resolution which allows for higher position precision. Receiver-dependent biases for two receivers are compared with the bias products of Center for Orbit Determination in Europe (CODE). Baseline lengths ranging up to ∼2,100km are tested with the receiver-specific biases; they resolve more phase ambiguity by 4.3% than using CODE's products.; The second project analyzes 1 s and 30 s interval GPS data of the 2003 Tokachi-Oki earthquake. For 1 Hz positioning, Iterative Tropospheric Estimation (ITE) method improves vertical precision. While equalized sidereal filtering reduces noise for multipath-dominant 30--300 s periods, it can cause long-term drifts in the timeseries. A study of postseismic deformation after the Tokachi-Oki earthquake uses 30 s interval position estimations to test multiple filtering strategies to maximize precision using lower-rate data. On top of the residual stacking, estimation of a random walk constraint of sigmaDelta = 1.80 cm/ hr shows maximum noise reduction capability while retaining the real deformation signal. These techniques enhance our grasp of fault response in the aftermath of great earthquakes.; The third project probes noise floor characteristics of very-high-rate (> 1 Hz) GPS data. A hybrid method, designed and tested to resolve phase biases, minimizes computational burdens while keeping the quality of ambiguity-fixed solutions. Noise characteristics are compared after an analysis of 5 and 10 Hz Ashtech MicroZ and ZFX as well as Trimble NetRS receivers. The Trimble NetRS receiver noise has a timeseries standard deviation double that of Ashtech MicroZ receivers. Also, the power spectral density function has a 0.1 Hz peak. Noise power shows white noise for the frequency range from 2 Hz and higher.; Each research project assesses the methods to reduce the noises and/or biases in various time intervals. Each method considered in this dissertation will fulfill the needs for scientific applications.