Diagnosing epoch-wise GNSS PPP-AR outliers using principal component analysis and Hotelling’s T 2 monitoring

Abstract Global navigation satellite system (GNSS) precise point positioning with ambiguity resolution (PPP-AR) can exhibit intermittent epoch-wise degradation that is difficult to interpret when measurement-quality indicators vary jointly. This study combines principal component analysis (PCA) and Hotelling’s T 2 monitoring with standardized top-three contributor attribution to characterize statistically unusual measurement conditions. Three representative International GNSS Service stations (KIR8, ASCG, and CHPG) were selected from a week-long assessment of 29 stations and analysed on days of year 015 and 016, 2024. Seventeen indicators comprised elevation, signal-to-noise ratio (SNR), code-minus-carrier (CMC), and code and carrier-phase prefit residuals. The 95 % cumulative explained-variance criterion retained 12–14 components, and Hotelling’s T 2 flagged 0.45–1.48 % of observations. KIR8 showed prominent SNR and elevation signatures, ASCG mixed SNR, CMC, and residual contributions, and CHPG persistent CMC-dominated signatures. Compared with Bonferroni-adjusted univariate screening, Hotelling’s T 2 identified 5,045 points, including 2,056 (40.8 %) not detected by individual-indicator thresholds; univariate screening also yielded exclusive detections. These results demonstrate the value of joint detection and contributor attribution for interpreting station-dependent measurement-quality patterns beyond position-domain screening.