Examples

PPPx ships with ready-to-run examples under the example/ directory of the repository. They are the fastest way to confirm your installation works and to see realistic configurations.

This example processes the ZIM2 station (Zimmerwald, Switzerland) with several configurations. By default it uses GPS + Galileo (RTK uses GPS only). Run the whole set with:

cd example/pppx
./run.sh

run.sh unzips the bundled products and runs each *.ini, writing the .pos, .log, and .stat files (and a plot) into a folder per case.

Two configurations compare the ionosphere-free combination against a single-frequency solution that fixes the ionosphere with a GIM. Both use the LSQ solver and broadcast ephemeris.

pppx -c 00_spp_if.ini rinex/ZIM200CHE_R_20221000000_01D_30S_MO.rnx   # IF combination
pppx -c 01_spp_sf.ini rinex/ZIM200CHE_R_20221000000_01D_30S_MO.rnx   # single-frequency + GIM

00_spp_if.ini (IF)	01_spp_sf.ini (SF + GIM)

At mid-latitude the single-frequency solution is actually the better of the two: the GIM corrects the ionosphere well, while the IF combination amplifies pseudorange noise (~3×).

The same data processed with the EKF (forward-only, simulating real-time) and the FGO (batch post-processing) solvers, using precise products.

pppx -c 02_ppp_ekf.ini rinex/ZIM200CHE_R_20221000000_01D_30S_MO.rnx   # EKF
pppx -c 03_ppp_fgo.ini rinex/ZIM200CHE_R_20221000000_01D_30S_MO.rnx   # FGO

02_ppp_ekf.ini (EKF)	03_ppp_fgo.ini (FGO)

The EKF shows the familiar ~15-minute convergence period because it solves epoch-by-epoch; the FGO forms one large least-squares problem over the whole session, so there is no convergence ramp.

A short baseline between ZIM2 (rover) and ZIMM (base), EKF solver. Because the baseline is short, [model] trop and iono are set to none.

pppx -c 06_rtk.ini rinex/ZIM200CHE_R_20221000000_01D_30S_MO.rnx rinex/ZIMM00CHE_R_20221000000_01D_30S_MO.rnx

With the ambiguities resolved, the precision is typical for a short baseline: about 0.24 cm (east), 0.28 cm (north), and 0.48 cm (up).

TDP targets rapid displacement monitoring (e.g. earthquakes): it eliminates the ambiguities by differencing adjacent epochs, estimates the displacement between them, and accumulates it onto the initial position. It is run on the high-rate (1 Hz) ALGO data with precise products (to minimise drift from error accumulation).

pppx -c 07_tdp.ini rinex/ALGO00CAN_R_20221000000_15M_01S_MO.rnx

For the static ALGO station (zero true displacement), the precision over 15 minutes is about 0.28 cm (east), 0.59 cm (north), and 0.93 cm (up).

This example reproduces the high-precision ZTD estimation from Android smartphone GNSS published in Atmospheric Measurement Techniques. Three devices recorded 24 h of data: a geodetic station (ETH2), a u-blox receiver (UBLX), and a Google Pixel 4XL (PIXL).

Download the dataset from the ETH Research Collection, unzip it into the example folder, convert the Hatanaka .23d files to .23o, and run:

unzip dataset_2023-05-18.zip
cd dataset_2023-05-18/
for f in *.23d; do crx2rnx $f; done

./run.sh
python plot_ztd.py

The ZTD estimates appear in the last three columns of each .pos file (ZTD = zhd + zwd + dzwd).