The principle
Every Marenn Risk report rests on one rule. If we can get data at the property pin, we score at the pin. If the underlying data is genuinely at a coarser resolution, we say so.
Four perils are scored at the precise property location. Three are scored at the 1 km block that contains the property, because the underlying data is native to that scale (a radar pixel, a soil-class polygon, a cyclone wind field). We label those scores honestly and we never claim sub-block precision for data that doesn't have it.
This page documents the seven perils, the sources behind each one, the refresh cadence, the banding thresholds we use and how to request a correction.
The Marenn Score
Every report leads with a single composite number from 0 to 100. Higher means more climate-risk exposure. The composite combines seven peril scores using a weighted power mean.
Composite bands
| Score | Band | What it means |
|---|---|---|
| 0 to 31 | Low | No peril dominates. Standard hazard exposure for a well-sited property. |
| 32 to 47 | Moderate | One or two perils show meaningful hazard. Worth understanding before buying or renewing. |
| 48 to 59 | Elevated | A peril is materially loaded. Mitigation options and building defences are worth considering. |
| 60 to 71 | High | A peril is severely loaded, or several perils are stacked. A Marenn inspection helps quantify defences already in place. |
| 72 and above | Severe | Multi-peril extreme exposure. The hazard signal sits at the top of the modelled distribution. |
Per-peril bands
The seven individual peril scores use slightly wider bands than the composite, because a single peril uses the full 0 to 100 range more aggressively than the cross-peril composite.
| Score | Band |
|---|---|
| 0 to 24 | Low |
| 25 to 44 | Moderate |
| 45 to 64 | Elevated |
| 65 to 79 | High |
| 80 and above | Severe |
Why a power mean, not a simple average
Damage to a property is driven by the worst exposure, not the average across every peril. A simple average lets six benign perils mathematically cancel out one severe one, that's the opposite of how property risk actually works in real events. The composite uses a power mean (the formula is described in the appendix; the practical effect is that the dominant peril drives the score). A property with one Severe peril and six Low perils reads as elevated, not low. That's correct.
Composite weights
How much weight each peril carries in the composite for the South-East Queensland launch market:
| Peril | Weight | Rationale |
|---|---|---|
| Storm & hail | 28% | Above national 27%. Brisbane and SE QLD sit in one of Australia's hail belts. |
| Flood | 20% | Above national share. The February-March 2022 SE QLD floods are the single most expensive natural disaster in Australian history. |
| Bushfire | 15% | Around national share. 2019 Peregian Beach event factored in. SE QLD is less catastrophic than Victorian Black Saturday-class fire grounds. |
| Cyclone | 10% | Well below national 29%. SE QLD sits south of the typical tropical-cyclone track; ex-TC Alfred reaching Brisbane in March 2025 was the first significant SE QLD impact in many years. |
| Subsidence | 10% | Real for reactive-clay sites across the launch market; slow-burn but accumulates over ownership. |
| Storm vegetation | 9% | Secondary mechanism to storms; tree-fall is a major contributor to storm-event damage in coastal QLD. |
| Erosion & inundation | 8% | Slow-burn but material for coastal-strip properties in the launch market. |
How we set these weights. The hierarchy is anchored to the canonical Australian peril-loss research:
- McAneney, J. et al. (2019). "Normalised insurance losses from Australian natural disasters: 1966-2017." Environmental Hazards, 18(5), 414-433. Peer-reviewed, open-access. The paper normalises the Insurance Council of Australia's Disaster List across 52 years to a common dollar base. Their Table 3 breaks the national totals down by peril:
- Tropical cyclone: 29% of normalised losses (dominated by North Queensland strikes, Tracy 1974, Larry 2006, Yasi 2011)
- Hail / severe storm: 27% (Sydney 1999 is the most expensive single event in the record at AUD 5.6 billion)
- Flood + bushfire + non-hail storm: ~38% combined, split roughly evenly
- Earthquake: 5% (not in the Marenn model, weight redistributed across the perils we do score)
- SE QLD adjustments away from the national figure. The launch market sits in a different peril mix than the national average. Cyclone shifts down because SE QLD is south of the typical tropical-cyclone strike zone. Storm/hail shifts up because Brisbane and the Sunshine Coast are hail-belt territory. Flood shifts up because the 2022 SE QLD floods reset the regional baseline. Bushfire shifts down because SE QLD doesn't carry the Victorian fire-ground catastrophic risk profile.
- Slow-burn perils, judgement. Coastal erosion, storm vegetation and subsidence aren't event-based perils in the McAneney sense. They accumulate over ownership tenure rather than concentrating in catastrophe events, so they're not well-captured in event-list datasets. The 8 / 9 / 10 percent weights are our own judgement on how much of total ownership-tenure damage exposure each accounts for in the launch market.
What the weights are not. They are not a calibration against Marenn's own claims data, we don't hold any. They are not an attempt to replicate any insurer's pricing recipe. They are not a forward projection of climate-change-driven loss shifts. They are the percentages we use today, anchored to the public record of what has actually cost Australian property owners money for the last fifty years, with SE-QLD-specific adjustments noted.
Regional weight variants are on the roadmap. Tropical Queensland properties will shift cyclone up. Granite Belt properties will shift bushfire up. The current set applies to the Noosa-Coolum coastal launch market and South-East Queensland generally.
Two readings of the same property
Where the report has information about the building itself, the roof material you entered on the /risks form, walls material classified from the aerial and Street View imagery or notes from a Marenn Inspection, the composite is shown two ways.
The first is the hazard exposure: what the site itself faces, ignoring the building. The second is the property-adjusted reading: the hazard exposure refined by the building's materials and condition. Colorbond steel roofs handle hail better than aging tiles; rendered block walls handle ember attack better than weatherboard. Where these differ materially, the report shows both. The cohort comparison and the suburb peer tables always use the hazard-exposure number, so two neighbours are compared on the site they share, not on the building each has chosen.
The seven perils
For each peril we list the resolution (pin or 1 km block), the data sources, what we measure and how often the underlying data refreshes. The same scoring code runs daily across every property in the cohort; the score on your report uses the latest run.
- Sources
- Queensland's Bushfire Prone Area (BPA) overlay, the official state map of land at heightened bushfire risk, plus the Vegetation Hazard Class (VHC) map of fuel load by polygon. North Australia Fire Information (NAFI) for the history of actual fire scars in the area. NASA FIRMS for satellite-detected hotspots over the last 15 years.
- What we measure
- Five signals, in order of weight in the score: (1) the worst BPA-rated polygon within 50 m of the property (25 percent of the score); (2) recent fire-scar overlap with the property or within 200 m (20 percent); (3) the BPA class the property itself sits in (15 percent); (4) the fuel-load score at the property and within 50 m, whichever is higher (10 percent); (5) the worst direction for recent satellite-detected fires within 5 km (10 percent). Weights sum to 80 percent and redistribute across whichever signals have evidence for the address.
- How signals (1) and (3) overlap
- Signal (1) is the worst BPA polygon found anywhere within 50 m of the pin, and a property that sits inside a BPA polygon is, by construction, within 50 m of one. So when the at-pin polygon is also the worst within 50 m, the same polygon's tier carries both weights for an effective 40 percent contribution. That is intentional: a property that sits inside a BPA polygon faces both at-parcel and adjacent fuel exposure and the score reflects that compounding. Signal (1) acts as a pure adjacency signal only when a worse polygon sits nearby off-parcel, in which case (1) and (3) read different tiers.
- In plain language
- The score answers four questions: how much vegetation is on or beside this property, has fire reached this stretch in living memory, where is the wind likely to bring fire from and is the council's own bushfire mapping flagging the site.
- Refresh
- BPA quarterly from QSpatial. Fire scars after each fire season. Satellite hotspots multiple times a day.
- Sources
- Bureau of Meteorology AURA radar archive (Australian Unified Radar Archive). Three radars are stitched into a single mosaic for the Noosa region: Gympie (75 km north-west), Marburg (150 km south-west) and Mount Stapylton (145 km south). For each 1 km square at each moment in time we pick the closest radar whose beam reaches low enough to see what's actually happening.
- What we measure
- Three counts per square: how often a severe thunderstorm passed over, how often the radar saw a hail-strong return and how often the return was strong enough to be large hail. About 26 years of history. The square's score is compared against the median across all Marenn-cohort squares so a "high" reads as high relative to actual neighbours, not absolute.
- In plain language
- If your address is in a stretch the radar has seen storms cross frequently for two decades, the score reflects that. If it's in a quieter pocket, the score reflects that too. Storms hit areas, not parcels, the same storm cell drops hail on you and on your neighbour 200 metres away.
- Refresh
- Daily, with a typical few-day publishing lag from the Bureau.
- Sources
- Bureau of Meteorology's tropical cyclone database and the international IBTrACS archive, together a record of every cyclone track that has crossed the Australian region since the late 1800s.
- What we measure
- For every cyclone that passed within 100 km of the 1 km square at the property, we ask two questions: how close did it pass and how strong was the wind at closest approach. Closer counts more, a cyclone that passed 25 km away contributes roughly 40 percent of the weight of a direct hit; one 100 km away contributes only a few percent. We sum those contributions across every cyclone in the archive. Higher score means more, closer and stronger cyclones have crossed near here historically.
- In plain language
- Cyclones don't damage homes in proportion to how many have brushed by; they damage in proportion to how close and how strong. The score reflects that. Counts of direct hits and the most intense systems are also shown on the report, alongside the score, so you can see what's behind it.
- Refresh
- Annually, after the official cyclone season closes.
- Sources
- Geoscience Australia's Water Observations from Space, an archive of every Landsat satellite image of Australia going back to 1987, with each 30-metre pixel classified as wet or dry. We use two versions: the all-time record (how often this pixel has been wet across the whole archive) and the per-year record (which lights up specific flood years like 2022). The council flood-overlay maps (Noosa Plan and Sunshine Coast Council Planning Scheme) are layered on as the regulatory line. For properties outside both the satellite cohort and council mapping, regional rainfall extremes from the long-term SILO archive provide a backstop.
- What we measure
- How often satellites have seen this 30-metre patch of ground wet across 39 years of imagery, and the worst single year on the record. A property classed as wet on 40 percent of clear satellite passes during 2022 carries direct evidence of having flooded that year, regardless of its long-run average. A property mapped inside a council flood zone always reads at least at the council-mapped severity, even if the satellite record is sparse. Where both signals are absent, the rainfall backstop applies, capped at the Moderate band because rainfall alone can't distinguish a flood-prone parcel from its drier neighbour 500 metres away.
- Honest limitation
- Satellites pass over every 8 days, and event-day cloud often hides the worst of a flash flood. Short floods that drain in 12 to 24 hours can happen entirely between passes and leave no satellite record. The satellite signal undercounts flash floods by design; the council floor and the rainfall backstop partly compensate, but a property in a known flash-flood corridor may carry more historical exposure than the satellite number alone suggests.
- Refresh
- Quarterly, when Geoscience Australia republishes the satellite archive. Council overlays refresh as they are reissued, typically every one to three years. Rainfall extremes refresh after each annual archive release.
Flood and erosion-and-inundation sound similar but score different physical mechanisms. Flood covers acute event water: heavy rain that can't drain, council-modelled flood extent, an acute storm-tide pushing sea level up during a single storm. Erosion and inundation covers the chronic, decadal trajectory of the coastal land itself: shoreline retreat, dune recession and the sea-level-rise allowance that projects where mean tide will reach in coming decades. A property near a tidal estuary can score on both and that's correct rather than double-counted. They are different time horizons of different processes.
- Sources
- Geoscience Australia's Digital Earth Australia Coastlines, the official record of where the Australian shoreline has been every year since 1988, derived from satellite imagery, with a calculated rate-of-change in metres per year at points along the whole coast. Queensland's Coastal Management dataset adds three regulatory overlays: the highest-tide buffer, the calculated 75-year erosion zone and the long-horizon sea-level-rise allowance.
- What we measure
- The measured rate of shoreline change at the property in metres per year, a negative number means the shoreline is retreating toward the property, a positive number means it's building seaward. We band this into score tiers from stable-or-growing-seaward (low) through to catastrophic retreat of more than one metre a year (severe). The measured-rate signal fades with distance: pins within 100 metres of the coast feel the full effect; pins 500 metres inland feel half; beyond 500 metres the measured-rate signal doesn't apply. Separately, properties mapped inside any of the three regulatory overlays read at least at that overlay's severity, even if no measured rate is available within tolerance.
- Honest limitation
- The satellite shoreline record is most reliable for open-coast pins along clean sand or rock interfaces. Tidal-estuary and lake-edge pins (Noosa River, Lake Weyba, Doonella, Cootharaba) can carry lower-certainty rates because seasonal water-level changes confuse the shoreline-detection algorithm. The confidence label on each erosion score reflects the underlying data's confidence. Pins more than 500 metres from any coast read zero on the measured-rate signal regardless of inland inundation history (which the regulatory overlays capture instead).
- Refresh
- DEA Coastlines republishes annually with each new Landsat year. State coastal overlays refresh as they are republished, typically annually.
- Sources
- Queensland Fire Department's Vegetation Hazard Class map, the official state record of how much fuel (vegetation) sits on every polygon of land. This gives us the fuel-load at the property and within 50 m and 200 m of it. Plus five years of Sentinel-2 satellite imagery (a free European satellite that passes every five days) processed to a greenness index. Aerial imagery analysed by Marenn's Vision AI is used for the narrative description of the canopy on the report but does not feed the composite score.
- What we measure
- Two things. First: how much vegetation is near the structure, at the property itself, within 50 m, within 200 m. The 200 m band matters because mature trees can drop limbs that travel a long way in a wind event. Second: whether the satellite record shows that vegetation is healthy or in decline, we compare the median greenness at the property over the past 12 months against the two-to-five-year baseline. A meaningful drop in greenness indicates stressed or dying vegetation, which snaps in lower winds than healthy vegetation.
- How decline applies
- The decline signal lifts the score only where there's nearby vegetation to begin with. A dropping greenness reading on a property with no tall trees nearby has no physical mechanism to cause storm damage and contributes nothing to the score. Where the property has at least moderate vegetation near it AND we see meaningful decline in the satellite record, the decline adds points to the score in three bands: 10 points for 5-to-15-percent decline, 20 for 15-to-30, 30 for above 30. A dense canopy in measurable decline reads as a stronger storm-vegetation risk than the same dense canopy in good health.
- Honest limitation
- Sentinel-2 needs clear sky to see the ground. Properties that have been cloud-obscured during the baseline or recent windows (we require at least 5 valid observations in each) fall back to the vegetation-amount signal alone, and the confidence label is set to Medium or Low accordingly.
- Refresh
- Vegetation Hazard Class map quarterly when the Queensland Fire Department republishes. Sentinel-2 every five days subject to cloud; the decline baseline updates as new observations land.
- Sources
- The CSIRO Soil and Landscape Grid of Australia, the national map of soil reactivity (how much the soil swells when wet and shrinks when dry) and clay content, sampled per 1 km square.
- What we measure
- The soil reactivity class for the 1 km square the property sits in. Class 1 is the most stable soil (less than 10 percent clay, score 8 out of 100). Class 5 is the most reactive (more than 40 percent clay, score 90 out of 100). The score jumps non-linearly across the classes because most foundation damage, slab heave, cracking walls, doors that stop closing, concentrates on the higher-class sites. Class 1 and 2 sites rarely show movement damage on a well-built slab; class 4 and 5 sites need a specifically-designed slab to avoid it.
- Build-era adjustment
- If you tell us when the home was built, the score adjusts downward. The Australian standard for slabs in reactive-clay soils (known as AS 2870) tightened in 1986 and again in 1996; homes built to the modern standard on the same soil show materially less foundation movement than older builds. Post-1996 builds typically get a 45 percent reduction; post-2010 builds a 60 percent reduction. The reduction is shown on the report alongside the raw soil-reactivity number.
- Coverage in the launch area
- Our launch market (the Noosa-Coolum coast) covers 124 inhabited 1 km squares. Of those, 16 sit predominantly over water at the centre (Noosa Spit, Noosa River banks, the strip just offshore at Sunshine Beach) and the national soil grid can't return a direct reading there. For those squares we inherit the soil-reactivity class from the nearest neighbouring square, usually one kilometre away, that does have a reading. Every grid row records how its reading was sourced (direct sample, fallback search or nearest-neighbour inheritance), so the inheritance is auditable.
- Honest limitation
- The national soil grid is an interpolation from roughly 100,000 soil-survey points across the country, which works out to about one survey point per 30 square kilometres on average. The grid that comes out is modelled, not directly measured under your house. The 1 km square resolution we publish is honest about what the model can actually deliver; reading at finer scale would just return the same interpolated number. The build-era adjustment carries most of the per-property signal in the final score.
- Refresh
- When CSIRO republishes the national soil grid, which is infrequent. The build-era adjustment applies immediately when a build year is added to the report.
Pin or 1 km block
The four pin-resolution perils (bushfire, flood, erosion and inundation, storm vegetation) score at the precise property location. A property backing onto a national park reads differently from a property on the same street 200 m to the east, because the underlying polygon data supports that distinction.
The three block-resolution perils (storm and hail, cyclone, subsidence) score at the 1 km grid cell containing the property. Two properties in the same cell receive the same score for these three perils. This is not a Marenn limitation. It reflects the native resolution of the underlying physical data:
- Radar pixels are 1 km native. A storm cell that drops hail on one property almost always drops hail on its neighbour 200 m away.
- CSIRO soil-class polygons rarely change at sub-kilometre scale in South-East Queensland. The clay content under your house is the same as the clay content under the house three doors down.
- Cyclone wind exposure varies at storm scale, not parcel scale.
Every block-resolution score on a report carries a "1 km block" footnote so you can see at a glance which perils could vary from a neighbour and which won't.
Refresh cadence
Different sources update at different speeds. The Marenn Risk pipeline reflects that, recomputing the per-property pin scores daily because BigQuery spatial joins are cheap. The sources behind those joins update at their own cadence:
- Daily ingest: BoM AURA radar (with a typical two-to-three-day publishing lag), NASA FIRMS hotspot scrape. The FIRMS scrape runs daily on schedule; the underlying public archive for the Noosa-coolum AOI has had no hotspot detections since the 2019 Peregian Beach fires, so a "daily refresh" here means a fresh pull of an archive that has not moved.
- Weekly (in rollout): Sentinel-2 NDVI at pin (every five days when not cloud-obscured). The per-pin NDVI backfill is currently rolling out across the cohort; until completion the storm-vegetation dieback signal falls back to the vegetation-structure read alone.
- Quarterly: QSpatial Bushfire Prone Area (latest available vintage July 2017; we re-ingest whenever the publisher releases a new version), QFD Vegetation Hazard Class (QSpatial; the fuel-load reading used in the bushfire and storm-vegetation scores).
- Annually: BoM tropical-cyclone track archive (we re-pull daily during cyclone season for late-arriving track points and after each season's official release), state coastal-management overlays.
- As published: council flood studies (typically every one to three years), CSIRO Soil Landscape Grid (infrequent).
- Pin-score recomputation: daily at 05:30 UTC via the
marenn-pin-scores-dailyCloud Run Job, refreshing all four pin-resolution perils (bushfire, flood, erosion and inundation, storm vegetation) against the latest state of every underlying source. The BigQuery spatial joins complete in about two minutes for the full 38,946-pin Noosa-coolum-coastal cohort.
The date stamp at the bottom of every report identifies the methodology version and the most recent refresh of each underlying source.
Confidence
Each peril score carries a confidence label that reflects how complete the input data is, not how high or low the score itself is. A missing source lowers confidence; it doesn't deflate the score.
- High: 90% or more of the expected inputs are populated.
- Medium: 60% to 89% of inputs are populated. A meaningful signal is missing.
- Low: fewer than 60% of inputs are populated. Use the score as a directional indicator only.
What the score does not include
The Marenn Score measures how much hazard sits around the property. It does not predict whether the next event will damage your home, and several things that materially affect what would actually happen are deliberately outside the score:
- Defensive features beyond roof and walls material. Roof and walls material flow into the property-adjusted composite (see "Two readings of the same property" above) when supplied via the
/risksform, classified from the aerial and Street View imagery by Marenn's Vision AI or recorded during a Marenn Inspection. But cyclone-rated tie-downs, gutter guards, defendable-space clearing, raised floors, engineered foundations and other defensive features that reduce real-world damage are recorded during the Marenn Inspection only. - Climate-change trajectory. The score reflects today's hazard distribution from the historical record. Forward projections are part of the Marenn Holistic Risk TimeFrame product, not the Marenn Score itself.
- Anything we can't see in public data. The score is built from public datasets: government overlays, satellite archives, radar history, soil maps. It can't see what's hidden, recent maintenance, defects behind walls, the actual depth of a slab. A Marenn Inspection is where those questions get answered.
Honest limitations
We surface what we know and label what we don't. Some specifics worth being clear about:
- Pin scoring uses the address geocode supplied by the Queensland address register. For most homes this lands on the building footprint. For large rural parcels, the pin may land in the middle of the lot rather than at the dwelling. Lot-level scoring is a v2 build.
- The radar mosaic has a few-day publishing lag, so very recent storms (the last week or so) may not yet appear in the storm and hail score. The score reflects the multi-decade record, not the last fortnight.
- The Sunshine Coast Council planning-scheme flood layer represents the council-regulated flood-and-inundation extent. It is not always identical to the modelled 1% AEP extent. We document which layer is being read on each report.
- Some council studies that are publicly viewable are only available as raster images, not vector polygons. Where this is the case, we say so and exclude them from the score until vectorisation is complete.
- Outside South-East Queensland, source coverage varies. The methodology is national in design; the data backfill is rolling out region by region.
Request a correction
If a Marenn Risk score for your property looks wrong, tell us. Council overlays carry errors, our spatial joins occasionally misclassify edge cases and your direct knowledge of your property is sometimes the most reliable signal in the system.
Email hello@marenn.com with the property address and what you believe is wrong. Attach any supporting documents (council flood certificates, surveys, BPA exemption records, recent photos of vegetation cleared after the council's last assessment).
We acknowledge requests within two business days and investigate within ten. If a correction lands, the underlying score recomputes on the next daily refresh, usually within twenty-four hours.
Versioning
Methodology versions are documented at the bottom of every Marenn Risk report so the score you read today is reproducible in three years' time. This page captures the current version. Material changes (a new peril, a new source, a recalibrated band) trigger a version bump and a dated entry below.
- v1.10 · 26 May 2026 · Pre-launch honesty pass. Bushfire card discloses that signals (1) "worst BPA polygon within 50 m" and (3) "BPA class at pin" overlap when the at-pin polygon is also the worst within 50 m, that polygon's tier contributes to both signals (intentional, reflecting both at-parcel and adjacency exposure). No scoring change; methodology language now matches the live + batch implementation.
- v1.9 · 25 May 2026 · Composite weights re-anchored to McAneney et al. (2019) "Normalised insurance losses from Australian natural disasters: 1966-2017" (Environmental Hazards, peer-reviewed, open-access). New set: storm and hail 28, flood 20, bushfire 15, cyclone 10, subsidence 10, storm vegetation 9, erosion and inundation 8. Replaces the previous 22 / 22 / 13 / 13 / 10 / 10 / 10 Marenn-judgement set with weights anchored to the published national peril-loss breakdown plus documented SE-QLD adjustments. Composite weights section now carries the derivation table and the source citation.
- v1.8 · 25 May 2026 · Plain-English rewrite across the page. Acronyms spelled out on first use, per-peril cards reframed so a non-specialist reader can follow what each peril measures and why. Body copy reframed as educational climate-risk information based on public data and Marenn's own modelling judgement.
- v1.7 · 25 May 2026 · Scoring alignment pass. Cyclone, subsidence and bushfire formulas reconciled across the live consumer profile and the batch pipelines so all surfaces show the same number for the same property. Subsidence coverage closed to 100 percent of inhabited cells in the launch scope. Daily pin-score recompute scheduler shipped.
- v1.6 · 23 May 2026 · Storm-vegetation card sources corrected (the right state-government vegetation map, not the federal one). Card body updated to describe the actual scoring inputs.
- v1.5 · 23 May 2026 · Erosion and inundation card rewritten to lead with the satellite-measured shoreline change rate rather than overlay membership. The council overlays now act as a floor on the score.
- v1.4 · 23 May 2026 · Honesty pass on data freshness and limitations. Flood card gained an "Honest limitation" row acknowledging that satellite-derived flood detection undercounts short flash-flood events. Subsidence card gained the same row acknowledging the national soil grid is a modelled interpolation, not a direct measurement.
- v1.3 · 23 May 2026 · Flood card updated to document the regional rainfall-extremes fallback signal. Bushfire scoring narrowed to the documented set of signals.
- v1.2 · 23 May 2026 · Composite band descriptions revised to hazard-language only. Earlier wording made claims beyond the scope of a climate-risk report; those have been removed. Band thresholds and tier names unchanged.
- v1.1 · 23 May 2026 · Satellite-derived vegetation-decline signal added to the storm-vegetation peril. Storm-and-hail peril card relabelled for parity with the report rendering.
- v1.0 · 21 May 2026 · Initial publication. Seven perils, pin-versus-block resolution, weighted composite, South-East Queensland weight set.
Sources and credits
Marenn Risk reports synthesise public-sector and open data. We acknowledge the agencies that publish the underlying records:
- Bureau of Meteorology (AURA radar archive across the Gympie, Marburg and Stapylton radars; tropical cyclone database).
- Queensland Department of Resources (QSpatial Bushfire Prone Area and Vegetation Hazard Class, the QFD VHC fuel-load reading).
- Queensland Department of Environment, Science and Innovation (Coastal Management dataset).
- Geoscience Australia, Digital Earth Australia (Water Observations from Space and DEA Coastlines, both under CC BY 4.0).
- Sunshine Coast Council and Noosa Council (planning-scheme overlays, flood studies).
- CSIRO (Soil Landscape Grid of Australia).
- NASA (FIRMS hotspot detections, Sentinel-2 imagery via the Element 84 EarthSearch archive).
- North Australia Fire Information (NAFI) for fire-scar polygons.
Marenn Risk is not endorsed by, sponsored by or affiliated with any of the agencies above. The score and its presentation are Marenn's own analysis of public data.
Contact
Questions about a specific score, a correction request or the methodology itself:
Marenn Australia Pty Ltd
hello@marenn.com
Peregian Beach, Queensland, Australia