Associations of Regional-Background PWV Variability and the Nearest-Cluster Residual with Daily Rainfall in a Dense Mountain GNSS Network: A Chuxiong Case Study

Dense global navigation satellite system (GNSS) networks enable continuous retrieval of precipitable water vapor (PWV), but single-station analyses cannot separate regionally coherent moisture variability from a nearby station’s departure from that background. We developed a target-cluster-excluded decomposition to test whether the nearest GNSS cluster provides incremental rainfall information beyond the regional signal. In a 46-day record from mountainous Chuxiong, China, 46 of 53 stations formed 20 one-kilometer clusters. The median of 19 centered clusters excluding C06 defined the regional background; C06’s unexplained component was estimated without rainfall information. The first principal component explained 93.6% of intercluster PWV-anomaly variance and correlated at 0.952 with regional station-height PWV anomalies from ERA5. The background explained 80.9% of C06 daily PWV variance. Its correlation with RG01 daily rainfall was 0.479, compared with −0.109 for the unexplained C06 component; the difference was 0.588 (7-day moving-block bootstrap 95% confidence interval: 0.227–0.850; circular-shift p = 0.065). Conditional analysis and blocked cross-validation did not identify a stable incremental contribution from C06 within this sample. Across all clusters, the correlation difference did not vary monotonically with distance from RG01, and C06 was not atypical. Half-hourly analysis showed that daily aggregation smoothed within-event variability, but ten events did not support a universal pre-rainfall increase and post-rainfall decrease. These results suggest a stronger rainfall association for regional-background PWV variability, but the evidence remains tentative given the circular-shift p-value of 0.065 and the record length of only 46 days. The transferable framework separates network-common variability from target-site departures and supports scale coordination between dense GNSS networks and point rainfall observations.