Downloads

Everything here is free, generated locally, and documented in the sample dataset guide. The three sample files are the same ones every tool loads with its demo button — download them to rehearse the workflow in your own time, or to use as format references for your own exports.

Sample datasets (synthetic, for training)

The dataset is a synthetic teaching construction — see the guide's important statement about how it may and may not be used.

Engineering templates

• InSAR screening work plan (work plan template) — coming with the first template release.
• InSAR–GNSS validation log spreadsheet — coming with the first template release.
• Subsidence hazard register — coming with the first template release.
• Coherence & decorrelation field checklist — coming with the first template release.

The free templates and the locally-adapted premium packs are being prepared; the pre-check and report tools already produce Markdown deliverables in the meantime.

How the sample files map to the workflow

The three files are not random extracts: they form a single, connected exercise that walks the same chain you would run on a real project. Start with the InSAR point table, which is the output format of a satellite processing stack (point id, longitude/latitude in WGS84, height, line-of-sight velocity in mm/yr, incidence angle and coherence). That table feeds the LOS Converter, where you project velocities from the satellite line of sight onto the vertical or the slope direction — the geometric step that turns a satellite measurement into a quantity you can compare with ground instruments. Next, the GNSS benchmark file holds 12 stations with vertical velocities and their declared accuracy; the GNSS Validation tool compares the satellite-derived rates against those benchmarks station by station and reports the bias and RMSE of the comparison. Finally, the time series file is in long format — one row per point per epoch — with 6 points and 12 monthly epochs of cumulative displacement; the Time-series Calculator plots the series, fits the linear trend and flags points whose residual behaviour departs from the model. Running that chain end to end with the samples takes about ten minutes and reproduces the exact numbers in the sample dataset guide.

Column formats and units

Before you point a tool at your own data, check the columns against these references — the converters are deliberately strict so that a wrongly labelled column fails loudly instead of producing a plausible wrong result. The InSAR point table uses: id, lon and lat (decimal degrees, WGS84), height_m, vel_los_mm_yr (velocity along the line of sight, mm/yr), inc_deg (incidence angle at the point, degrees from vertical), coherence (dimensionless, 0–1) and regime (a label identifying which of the four deformation regimes the point belongs to in the synthetic dataset). The GNSS file uses station, lon, lat, vel_v_mm_yr (vertical velocity, mm/yr) and accuracy_mm_yr (declared 1-sigma accuracy of the vertical rate). The time series file uses point_id, epoch (YYYY-MM), and cumulative_mm (cumulative displacement since the first epoch, in mm). If your own export uses different column names, rename them before loading — every tool shows the expected columns on its page and refuses a file that does not match, which is a feature, not a limitation.

Engineering templates — what they will contain

The template pack is being prepared to match the workflow this site documents, so each template answers one step of the chain rather than repeating generic boilerplate. The InSAR screening work plan will give the scoping document of a screening study: objectives, area, stack selection, point density targets, validation plan and deliverables. The InSAR–GNSS validation log will structure the station-by-station comparison (satellite rate, ground rate, difference, bias and RMSE per network) so the acceptance test of a provider or the internal QC of a project is auditable. The subsidence hazard register will turn the velocity map into a risk register: zone, mean velocity, acceleration flag, affected infrastructure, priority and action. The coherence and decorrelation field checklist will cover the site conditions that decorrelate the signal (vegetation growth, ploughing, construction change, snow, surface moisture) so that an empty or noisy area is explained in the report rather than interpreted. Until the pack ships, the screening and validation tools already produce their outputs as Markdown, which you can adapt into your own register format.

Using the samples in a training session

The samples are sized for a short supervised exercise: three files, one page each, ten minutes end to end. A trainer walks the chain — load the point table, convert the LOS rates, validate against the GNSS file, plot the series — while the trainees follow on their own machines; the guide's expected numbers let everyone confirm that their chain is configured correctly before moving to real data. For self-study, the same sequence is the fastest way to learn the column conventions and the sign conventions of each tool without risking a real dataset. Beyond training, the files serve as format references: when you prepare your own export, keep one sample file open as the template for column names and units so that the tool's strict loader accepts your data on the first attempt — the discipline of checking the format before loading is the one that prevents the silent failures that cost survey projects the most time. A team that rehearses with the samples once also removes the most common support questions about column order and sign convention before the first real campaign, which is why the exercise is recommended at project onboarding rather than under deadline pressure. Finally, the samples double as acceptance test data: if a team's local copy of a tool produces the guide's numbers, the installation and the data pipeline are healthy — a ten-minute regression test that costs nothing to run and that every project site can repeat before each new campaign starts. The same check doubles as the handover test when a new engineer joins the team, and as the first step of any on-site workshop or training session with the client's field staff.

Reproducibility and license of the samples

The sample files are generated by a fixed-seed script in the project repository, so every copy is identical and the expected results in the guide hold for every download. If a file you downloaded reproduces different numbers, you have found a bug — and we would genuinely like to hear about it. You may use the samples freely for training, evaluation and internal procedure development; the only restriction is that they are synthetic and must not be cited as survey evidence, and the disclaimer below applies to every file regardless of how it is combined with your own data.