GPEX · A practical introduction
How to use GPEX
Explore the data, compare models, build custom scenarios and understand sensitivity.
Explore the evidence, make your assumptions explicit, and see how they shape the map. Follow these four steps, even if geophysics is new to you.
Start with ExploreBegin with the existing results
Explore the evidence
Start with a prepared scenario so you can understand the map before changing any assumptions.
- Open Explore and select a scenario. Use Selected scenario in the Layers panel. Start with Baseline geological interpretation, then read its Scenario details and weights.
- Inspect an evidence layer. Expand a theme under Evidence layers and switch on a layer. For a clear view of one input, hide other evidence layers and untick Show prospectivity. These are processed favourability layers on a 0–1 scale, rather than raw survey measurements.
- Click a valid map location. The Location inspector shows the selected scenario’s score and the contributions from its themes and evidence layers. An evidence contribution combines that layer’s favourability with its effective weight.
- Adjust the view. Use Fit to extent to return to the study grid. Choose Navy, Warm, Cool, or Purple in Palette. Opacity and palette change how the map looks, while the underlying values stay the same.
| Theme | Layers you can explore | What they represent in this model |
|---|---|---|
| Magnetics | Structural favourability; Source contrast | Interpreted structure-related and magnetic source-contrast favourability. |
| Radiometrics | Potassium (K); K/Th; K/U | Potassium and ratio-based evidence used to represent alteration-related favourability. |
| Gravity | Density-boundary proximity; Domain architecture | Density-boundary and broader architectural evidence, including an exploratory domain criterion. |
New to geophysics? What do these themes mean?
Magnetics records variations in Earth’s magnetic field related to rocks’ magnetic properties. Radiometrics uses natural gamma radiation to estimate near-surface potassium (K), thorium (Th), and uranium (U). Gravity measures subtle variations in the gravity field associated with rock density.
GPEX uses evidence derived from these measurements and transformed into favourability layers. These model inputs are not raw instrument readings or mineral assays.
Learn more: Geoscience Australia: geophysics (opens in a new tab) · radiometrics (opens in a new tab).
Model results use only cells where all seven required evidence layers are valid. Areas outside this common footprint are NoData. The dashed grid rectangle also includes NoData; it is not the survey boundary.
Compare two prepared interpretations
See what changes
Compare shows two of the eight curated scenarios. Use it to connect a change in assumptions to a change in the result.
- Choose Scenario A and Scenario B. A useful first comparison is the baseline against Magnetic-theme emphasis. Read the descriptions and theme-weight changes alongside the map.
- Use Swipe. Drag the swipe control to reveal A on the left and B on the right. Both maps share the same position, palette, and fixed 0–1 scale.
- Switch to Difference. The map shows B − A. Positive values mean B assigns a higher score; negative values mean B assigns a lower score. Zero means no score change at that cell.
- Read the comparison metrics and inspect a location. Look at score changes and overlap of the highest-ranked areas, then click a cell to examine how its contributions differ.
The difference map keeps a diverging colour scale centred on zero. Its range is specific to the selected pair, so read the legend when you change scenarios. The Palette selector applies to Swipe view.
Compare uses the curated set. Create and save your own scenarios in Model Builder, then select them together under Custom Sensitivity.
Make a custom interpretation
Create, generate, and save
A scenario is one set of choices about the relative influence of the evidence. Model Builder translates your pairwise judgements into weights, then calculates a prospectivity surface from the same seven inputs.
- Name the scenario. In Custom Scenario, enter a clear, unique name and an optional note explaining your reasoning. For example, Stronger structural preference says more than Test 2.
- Set the pairwise judgements. Work through the main themes, then the comparisons within each theme. Choose which of the two inputs should have more influence, or choose Equal. Use the strength control to say how strongly you favour it. A larger preference is an assumption about importance, not better-quality data.
- Read the weight preview and checks. The bars show the resulting influence. If a group says Review, revisit the comparisons in that group. The applicable consistency checks must satisfy CR ≤ 0.10 before a model can be generated.
- Click Generate prospectivity. Wait for the new map and statistics. Until generation finishes, any previous map remains a reference. Changing a judgement does not automatically update the displayed prospectivity surface.
- Inspect the result, then click Save scenario. Review the statistics and click locations to understand their contributions. Saving adds the generated scenario to your Saved scenarios list and makes it available for Custom Sensitivity.
- Build another interpretation. Change at least one judgement, give the scenario a different name, generate it, and save it. You need at least two distinct saved scenarios for the next step.
First generate and save the baseline judgements with a clear name, such as My baseline. Then change the preference for Magnetics over Radiometrics from 2 (slight) to 3 (moderate), name it More magnetic influence, and generate and save again. These two different judgement sets give you a small, traceable custom comparison.
Generating creates a result for the current page. It does not automatically add it to Saved scenarios. After changing judgements, the name, or notes, generate again before saving or exporting the updated scenario.
What does a passing consistency check mean?
CR is the consistency ratio. It checks how well the pairwise preferences agree with one another. A passing value does not establish that the geological assumptions are correct. Groups containing only one comparison have no independent three-way consistency check; their zero CR is expected by construction.
Use the checks to identify conflicting preferences, then use the evidence and research rationale to justify your interpretation.
Where are saved scenarios kept?
Saved scenarios are stored in this browser for this website. Use the same browser profile and site address to find them again. They are not sent to other visitors when you share the website link, and they do not automatically appear on another device.
Clearing site data or using a private browsing session can remove or isolate these saved entries. Export a JSON record if you want a separate copy of a scenario’s settings. GPEX currently has no JSON import control.
Saving the same set of judgements again replaces the existing entry for that model, even if you give it a new name. To keep separate alternatives for sensitivity analysis, use different judgement sets and unique scenario names.
Examine dependence on assumptions
Which priorities persist?
Sensitivity asks how often a location remains among the highest-ranked cells when the weighting assumptions change.
| Mode | Scenario set | Question it addresses |
|---|---|---|
| Curated Agreement | Eight prepared geological scenarios | Which cells remain highly ranked across these selected interpretations? |
| Weight Robustness | A fixed, precomputed 900-scenario weighting ensemble | Which cells remain highly ranked across a broader tested set of weight variations? |
| Custom Sensitivity | Two to twelve valid scenarios saved in your browser | Which cells remain highly ranked across the alternatives you created? |
The first two modes display prepared results. Selecting Weight Robustness does not start a new random simulation.
Run your own comparison
- Open Sensitivity and select Custom Sensitivity. Your saved scenarios appear in Saved custom scenarios.
- Tick two to twelve scenarios. Choose the interpretations you want to test together. Their names must be unique, and all required consistency checks must pass.
- Click Run custom sensitivity. Wait for the agreement surface and review the Included custom scenarios list to confirm which alternatives were used.
- Explore Agreement surface and Stable zones. The agreement surface shows the fraction of selected scenarios in which each cell is highly ranked. Stable zones highlight cells meeting your chosen Persistence threshold.
- Click a valid cell. Inspect its persistence and scenario membership. If you change the selected scenarios, run the analysis again before interpreting or exporting the new set.
Reading one cell
Three of four scenarios
If a cell is highly ranked in three of four selected scenarios, its agreement is 3 ÷ 4 = 0.75. It meets a threshold of at least three of four, but not a four-of-four threshold.
This is a 75% frequency within that tested scenario set. It is not a 75% chance of mineral occurrence.
What counts as “highest-ranked 20%”?
Each scenario has its own 80th-percentile score cutoff, calculated over valid model cells. Cells at or above that cutoff are included. Ties at the cutoff can make the selected share larger than 20%.
The cutoff is a ranking rule, not a universal prospectivity score of 0.8. A score threshold and a persistence threshold answer different questions: the first selects high-ranking cells within one scenario; the second selects cells that remain high-ranking across scenarios.
High persistence indicates stable relative priorities within the chosen set of assumptions. Similar scenarios can agree strongly. Neither agreement nor the number of tested scenarios provides independent confirmation of mineralisation.
Keep the settings and the spatial result
Save your outputs
| Output | Where to get it | What it contains |
|---|---|---|
| Scenario JSON | Model Builder → Generate prospectivity → Export scenario JSON | A record of the scenario settings and reported results. It contains no map geometry or raster cells, and cannot by itself become a map in GIS. |
| Prospectivity GeoTIFF | Model Builder → Generate prospectivity → Export GeoTIFF | The generated custom model’s numeric favourability surface. |
| Agreement GeoTIFF | Sensitivity → run Custom Sensitivity → Agreement surface → Export GeoTIFF | Numeric agreement fractions from 0 to 1 for the selected scenario set. |
| Stable-zones GeoTIFF | Sensitivity → run Custom Sensitivity → Stable zones → choose a threshold → Export GeoTIFF | 1 for cells meeting that threshold and 0 for other valid cells. |
The GeoTIFFs retain the original 20 m analysis grid and UTM zone 35S coordinates (EPSG:32735). Cells outside the common valid footprint are marked NoData, with a stored value of −9999. A valid zero in a stable-zones file is different from NoData.
In ArcGIS Pro, add the downloaded .tif as a raster using Map → Add Data. Apply your preferred symbology there. The website’s palette is a display choice; the exported TIFF stores numeric values rather than a coloured screenshot. ArcGIS Pro help (opens in a new tab)
Keep a note of the scenario set and persistence threshold used for each sensitivity export so another reader can interpret the file correctly.
A few useful terms
Read the labels with confidence
- Prospectivity
- The model’s relative favourability for exploration. It ranks locations under stated evidence and assumptions.
- Evidence layer
- One processed map used as a model input. GPEX brings seven such layers together.
- Scenario
- One explicit interpretation expressed through weighting or evidence-inclusion assumptions.
- Weight
- The influence assigned to an input. An effective weight combines its theme weight and its within-theme weight.
- Pairwise judgement
- A choice about which of two inputs should have more influence, and by how much.
- Consistency ratio (CR)
- A check on agreement among pairwise preferences. It does not measure geological correctness.
- Persistence / agreement
- The fraction of tested scenarios in which a cell meets that scenario’s high-ranking criterion.
- NoData
- A cell without a valid model result. It should not be interpreted as a score of zero.
Find your next step
If something is unclear
I generated a model, but Custom Sensitivity has no saved scenarios.
Return to Model Builder and click Save scenario after generation. Create and save a second scenario with different judgements and a unique name. Then reopen Custom Sensitivity using the same browser profile and website address.
Two names for the same judgement set do not create two separate saved models: saving the same model replaces its earlier entry.
Generate prospectivity is disabled.
Wait for the weight preview to finish. If a group is marked Review, inspect its comparisons and adjust conflicting preferences. If the check failed to load, use the displayed retry control. Generation also stays disabled while another generation, reset, or export is in progress.
The map still looks like my previous result.
Changes to judgements first update the weight preview. Click Generate prospectivity to apply them to the map, then save again if you want the updated scenario in Custom Sensitivity. Follow the map notice while a new result is being prepared.
Export GeoTIFF is disabled or asks me to run the analysis again.
In Model Builder, generate a current accepted model after your last edit. In Sensitivity, choose Custom Sensitivity, select the saved scenarios, and run the analysis. The export button becomes available for that completed result. If the page was refreshed or the selected set changed, run it again.
The first calculation is taking a while.
The application needs to load its analytical data into your browser. Keep the tab open and wait for the status message to finish. A new browser session may need to load this data again. An internet connection is also needed for the online basemap.
I changed the colours, but the result did not change.
That is expected. The Palette control affects map colours only. To test a different interpretation, change the judgements in Model Builder and generate a new model.
Ready to begin?
Start with one scenario and one location.
Read the evidence first. Then change an assumption and investigate the difference.