GPS Time Series Output Description

A description of ground deformation and quality control parameters shown on GPS time series plots.

Ground Deformation Parameters

east, north, up

These three plots show the east, north and up components of GPS station position, evaluated once per day.

The use of these plots is to show the variation in station position with time.

In the absence of nearby earthquakes or slow-deformation events, these should be nearly straight lines. Variations about a straight line show the uncertainty (or noise) in the measurement of station position from a single day of GPS data. Large departures from a straight line usually indicate some problem in the data processing; these problems are likely to be fixed if you view the plots again in a few days or weeks.

The values in the plots are in millimetres. They are in the IG00b version of the ITRF2000 reference frame, but with a constant value subtracted from each time series so that its mean value is zero. This means that the absolute values in the time series do not relate directly to a position on Earth, but that the slopes of the lines do give velocities relative to ITRF2000.

The plots are updated with new information after each day’s daily processing of the GPS network is complete, which is presently about 20 hours after the end of each UT day.

The values in the “raw data” plots come directly from the output of the daily processing, which includes orientation of each daily solution into the IG00b reference frame. Some “outlier” data points, which are far different to the neighbouring data points, are removed from the plots using an automated procedure. In the event that a real ground-deformation episode is currently underway, this sometimes means that the data points at the end of the time series, which should show the deformation, have been deleted as outliers.

The values in the “regionally-filtered” plots have been processed to remove a “noise” signal that is more-or-less common to all GPS sites in New Zealand. This “common-mode” signal may be due to the use of slightly erroneous satellite orbits, to regional-scale or global-scale mass redistributions (in the ocean, atmosphere or groundwater), to the use of non-optimal models in the daily processing, and perhaps to other causes. Removing the common-mode signal makes it much easier to see signals due to real deformation of the ground. For example, the slow-slip events at sites like GISB are much clearer in the regionally-filtered time series.

It is possible to show a mixture of raw and regionally-filtered data on a single graph, or to show either one or the other. Note that regionally-filtered time series are only available for sites within the New Zealand North and South Islands.

Data Tables

The values in the associated data tables (press the Data button) are better defined:

• For stations in the LINZ PositioNZ network, the values are millimetres relative to the official NZGD2000 position of the GPS station;
• For other New Zealand stations, the values may either be relative to the NZGD2000 position of the time series, or relative to the first point in the time series (a long-term aim is to have all values relative to the NZGD2000 position);
• For stations outside New Zealand, which do not have NZGD2000 coordinates, the values are relative to the ITRF2000 position evaluated at 00:00 UT 1 January 2000.

Click here for more complete information on the format of the data tables.

Quality Control Parameters

Epoch-Fraction

These plots show the fraction of data received compared to the amount expected. The epoch fraction is calculated over a whole day for daily RINEX files, and over the session for near real time (nrt) RINEX files.

For 30-second data, there are expected to be 2880 epochs of data each day. If, for example, one hourly data file is missing from the 24 expected files, the epoch fraction will show as 0.958.

Note that the epoch fraction is a measure of the number of epochs of data collected, not the total amount of data. Whether one satellite or twelve were recorded at a particular time, this still counts as data being recorded.

avgMP1 and avgMP2

MP1 and MP2 are linear combinations of the code (pseudorange) and phase measurements made by the GPS receiver. Unusual variations in MP1 and MP2, particularly sudden jumps or trends, may indicate data quality problems at the GPS station.

For more detail, read on...

MP1 is a linear combination of P1 (or C/A pseudorange), L1, and L2. MP2 is a linear combination of P2, L1, and L2.

MP1 and MP2 can be calculated throughout the GPS session. In this case, the non-constant part of both combinations is dominated by the respective pseudorange multipath, with very small contributions from the phase multipath.

The values plotted by GeoNet are not the record of MP1 and MP2 throughout the GPS session, but are their RMS values averaged over the length of the session (usually 1 hour for near real time (nrt) data, or 24 hours for daily data). Hence the names avgMP1 and avgMP2.

The values of avgMP1 and avgMP2 are typically a few tenths of a metre. The values are generally different for different receiver types, and may be different for different firmware versions in the same receiver.

The use of avgMP1 and avgMP2 is as a quality control tool to assess whether anything has changed in the environment of the GPS site. If the receiver, firmware and antenna have remained the same, avgMP1 and avgMP2 should remain fairly constant. If the values change it is likely that something has changed at the site. Perhaps the antenna has become partially covered in snow? Perhaps there has been a partial failure of the receiver or antenna? Perhaps a tree has grown up, or a building erected, in the vicinity of the antenna?

The derivations of MP1 and MP2 are summarised in a 1999 article: "TEQC: The Multi-Purpose Toolkit for GPS/GLONASS Data" in GPS Solutions, Vol. 3, No. 1, pp. 42-49, 1999 (published by John Wiley & Sons, Inc.).

chisqd

chisqd is an output from the daily GPS processing. It is an output from a network solution so it only depends on the day, and not on individual stations.

It is used as a quality control parameter to tell if the processing has behaved normally for that day.

Normally, chisqd varies between about 0.3 and 0.6. If it is much higher than this it means there has been a processing problem on that day.

chisqd is the value of the chi-squared statistic per degree of freedom in the least-squares solution for station positions and other parameters.