Data Quality

Numerous factors go into position data quality before processing. Environmental objects such as vegetation, signage, fencing, and transient reflectors, or radio frequency interference can not only degrade the robustness of the data and increase errors but can actually induce an artificial shift in the positions. Hardware issues/failures such as improperly terminated/corroded antenna cable connections, receiver glitches, antenna LNA failure, and others can also create frustrating impacts to data that are at times hard to troubleshoot.

Quality control and troubleshooting are two of the primary services where VerQuin excels.

Everyone wants reliable, clean, and precise data with low latency. Yet Quality Control and State of Health monitoring are often overlooked, underappreciated, or done incorrectly. Commonly, it isn’t until someone relies heavily on the data that an undiagnosed, enigmatic problem has increased errors or caused what appears to be a real signal. Tracking positions, multipath, signal to noise, cycle slips, ping times, voltages, and temperature requires an integrative approach and sound QC analysis architecture. At VerQuin, we enjoy looking for transients, potential impacts on data, mitigating radio frequency interference, and updating receiver firmware.

For real-time networks, we monitor data communication metrics such as RSSI, RSRQ, RSRP, SINR, uptime, latency, and others to resolve any performances issues remotely or on the next O&M visit. In addition, communications firmware is tracked and upgraded as appropriate. Over the span of our experience, we have resolved thousands of QC and SOH issues. Let us handle the network monitoring so you can rely on the data or know when there is a problem, so you aren’t surprised.

Below are some examples of such issues and their respective impacts on data quality or positions.  These stations are operated by UNAVCO through the GAGE Facility as part of the Network of the Americas.

P800 Position Timeseries:

The daily average detrended position timeseries of P800 shows the detrimental effects of vegetation growing over the radome. The station was installed in 2011 and had a clean, low error timeseries until 2017 when a bush began to grow over the dome. Note the dramatic effect in the vertical component until the bush was cleared. Vegetation is the enemy of quality GNSS data and there are numerous examples of its impact. Scheduled O&M obviates this and other problems that lead to data loss. Subsequently in July 2019, the M7.1 Ridgecrest earthquake is documented in the N component. Less obvious in the position timeseries, the impact on data quality from hardware and firmware issues are shown below.

P800 QC Parameters:

Three principal QC parameters (slips, multipath L1/L2, SNR) document the vegetation growth over the P800 radome as well as the impact of faulty firmware. It can be difficult to discern if a position transient is real (tectonic/other) but when coupled with proper QC analysis it can often be determined whether it is vegetation or hardware/software related. Note the increase in slips and multipath due to faulty firmware installed on Jan 30, 2020, followed by a remedy on May 13, 2020, below.

P471

Radio frequency interference (RFI) is vexing and often leads to increased position errors, particularly for real-time data. Modern, sensitive GNSS hardware can be susceptible to previously unrecognized RFI that was transparent to GPS-only equipment. Following a GNSS hardware upgrade at P471, RFI degraded the slips, multipath, and signal to noise. Therefore, it is beneficial to have a GNSS receiver that allows auto and notch filter settings to mitigate RFI, such as the Septentrio PolaRx5 installed on April 11, 2017.

NDAP

If you aren’t looking at your data on a fixed schedule and with an experienced eye, it is possible to collect years of compromised positions. Installed in  2001, NDAP had a faulty LNA within the GPS antenna that was causing the transients in all components until the equipment was replaced in July 2006.

P474 Position Timeseries:

This is what a good, clean timeseries ought to look like; low average daily position errors (1 sigma <2mm in the horizontal) with small seasonal/annual variations and a small post-seismic transient. The deep-drilled braced monument is installed in weathered granite, with good sky view, and free of near-source reflectors or objects to cause slips.