Climate risk vendors increasingly compete on meters. But a smaller grid does not necessarily mean more accurate climate risk data. Here’s what spatial resolution actually means, how climate risk data is downscaled, and how to evaluate climate risk analytics.
The short answer
Climate hazard resolution is the spatial scale at which a physical climate hazard is represented in a dataset or model. Higher resolution does not automatically mean higher accuracy.
A climate risk dataset can be delivered at 10 or 30 meters while relying on climate inputs originally modeled at much coarser scales. The additional grid cells may provide useful spatial detail, but they do not necessarily contain additional climate information.
The right question is therefore not simply “How fine is the resolution?”
It is:
At what spatial scale does the data contain meaningful information, how was the data downscaled, and is that scale appropriate for the hazard and decision being analyzed?
This distinction matters because climate risk analytics are increasingly being used for asset-level decisions: underwriting, investment, insurance, site selection, portfolio screening, and physical risk assessment. Buyers need to know not just the resolution at which climate risk data is delivered, but what that resolution actually represents.
What is climate hazard resolution?
Climate hazard resolution refers to the spatial scale at which a climate hazard is represented by a model or dataset, usually expressed in meters or kilometers.
For example, a dataset may be described as having a 30-meter, 100-meter, or 1-kilometer resolution. A smaller number means that the output is divided into smaller spatial cells.
But three different concepts are often grouped together under the word resolution:
1. Native resolution
Native resolution is the spatial scale at which the underlying model, observation, or dataset was originally produced.
Global climate models generally operate at scales of roughly 100–250 km. By contrast, some geospatial inputs — such as elevation and land cover — are available at resolutions of tens of meters.
2. Delivered resolution
Delivered resolution is the spatial grid at which a provider delivers its climate risk data to customers.
This is partly a product-design decision. Data can be resampled or modeled onto a much finer grid than the native resolution of the underlying climate forcing.
3. Effective resolution
Effective resolution is the finest spatial scale at which an output still contains meaningful, defensible information.
This is arguably the most important number for a buyer of climate risk data — and one of the least commonly reported.
A provider can deliver data at a 30-meter grid without claiming that every 30-meter cell contains independently modeled climate information.
Why the distinction matters
Downscaling translates information from a coarser spatial scale to a finer one. But downscaling does not create new observations. Every downscaling method introduces assumptions about how large-scale climate conditions translate into local conditions.
For some hazards, there is an additional modeling step.
Flood, wildfire, landslide, and urban heat risk are influenced heavily by local terrain, land cover, drainage, infrastructure, and other physical characteristics. Climate projections can therefore provide boundary conditions — such as rainfall, river discharge, sea level, or fire weather — which are then passed into a geospatial hazard model.
This means a 30-meter flood map does not mean that climate projections exist at 30 meters. It means the hazard model is being resolved using geospatial inputs that support that level of spatial detail.
The grid is only one part of the model.
Why climate risk analytics providers compete on resolution
Physical climate risk data has moved considerably closer to the asset over the past decade.
Earlier generations of climate risk assessments often operated at country, regional, or relatively coarse grid-cell scales. That was of limited use when the decision concerned a particular warehouse, office building, data center, infrastructure asset, or property.
The move toward asset-level climate risk analytics is therefore an important development.
But spatial resolution has also become a convenient proxy for quality.
A provider can say its climate risk data is available at 1 km, 100 m, 30 m, or even finer resolution, and the numbers are easy to compare. Model accuracy, uncertainty, validation, and the quality of the underlying inputs are considerably harder to compare.
This creates an important distinction:
Resolution tells you how finely a model represents information. It does not, by itself, tell you how accurate that information is.
A 10-meter climate hazard map built from coarse climate inputs may contain less useful information than a 100-meter model built using better observations, more appropriate downscaling, and stronger hazard-specific modeling.
For buyers, the question should therefore be:
What information supports the quoted resolution?
How is climate risk data downscaled?
Downscaling is one of the most important processes behind high-resolution climate risk analytics.
Broadly, climate risk providers use several approaches, often in combination.
| Method | How it works | Strengths | Limitations |
|---|---|---|---|
| Statistical downscaling | Coarse model output is statistically related to observed historical conditions and transferred to a finer grid | Relatively efficient, globally scalable and well established | Does not add new physics; can struggle with extremes and complex terrain |
| Constructed analog methods | Future climate fields are constructed from combinations of historical weather patterns | Preserves realistic spatial relationships | Relies on the assumption that historical patterns remain useful analogues for future conditions |
| Machine-learning super-resolution | Models learn relationships between coarse climate fields and finer-scale spatial patterns | Can capture complex spatial relationships efficiently | Performance depends heavily on training data and may be difficult to interpret |
| Dynamical downscaling | A regional climate model is nested inside a global climate model | Physically consistent and useful for some regional phenomena | Computationally expensive and geographically limited |
| Geospatial hazard modeling | Climate variables are used as inputs to a hazard model incorporating terrain, land cover and infrastructure | Particularly useful for locally driven hazards such as flood, wildfire and landslide | Fine spatial inputs do not eliminate uncertainty in the climate forcing |
The important point is that two datasets can have the same nominal resolution while being generated through very different methods. Those methods can produce materially different results.
A 2025 study published in Communications Earth & Environment illustrates the issue: seven statistical downscaling methods were run at the same 0.25-degree resolution, using the same climate model, emissions scenario and damage function. The resulting estimates of projected climate damages nevertheless varied substantially between methods.
The method can matter as much as — or more than — the grid.
Does higher resolution mean higher accuracy?
No. Resolution and accuracy are different concepts.
A higher-resolution model can provide more useful information when the underlying physical drivers of a hazard vary at a similarly fine spatial scale.
But a finer grid does not automatically improve:
- the underlying climate projection;
- the quality of observations;
- the downscaling method;
- the hazard model;
- the damage function;
- the treatment of uncertainty; or
- the representation of adaptation and vulnerability.
In fact, a model can sometimes create an appearance of precision that exceeds the information contained in its inputs.
This is particularly important for procurement teams evaluating competing climate risk data providers.
Two providers might both advertise 30-meter data while using different climate models, different downscaling methods, different terrain datasets, different hazard models, and different assumptions about adaptation.
Same resolution does not mean same information.
What is the ideal resolution for different climate hazards?
There is no single “best” climate hazard resolution.
The appropriate spatial resolution depends on the physical processes driving the hazard.
Hazards strongly influenced by terrain and land surface characteristics — such as flooding, landslides and wildfire — can benefit from relatively fine spatial data because elevation, slope, drainage, vegetation and fuel conditions can change considerably over short distances.
Other hazards are governed by much larger atmospheric or hydrological systems. Making the grid smaller does not necessarily add meaningful information.
A useful way to think about the question is:
Model at the scale at which the physical drivers of the hazard actually vary.
| Hazard | Typical useful resolution* | What determines useful detail? |
|---|---|---|
| Inland flooding | ~30 m or finer | Elevation, drainage and local topography |
| Coastal flooding | ~30 m or finer | Elevation, surge, tides and sea-level boundary conditions |
| Landslide | ~30 m or finer | Slope, aspect, curvature and terrain |
| Wildfire | ~30–100 m | Fuel continuity, vegetation and wildland–urban interface |
| Heat stress | ~100 m–1 km | Building density, land cover, surface materials and vegetation |
| Earthquake | ~100 m–1 km | Ground motion, site conditions and geological characteristics |
| Hurricane wind | ~1–10 km | Storm footprint, terrain roughness and atmospheric dynamics |
| Drought / water stress | Catchment or ~1–10 km | Water balance, catchments, aquifers and supply systems |
| Hail | ~10 km or coarser | Storm-scale atmospheric dynamics and sparse observations |
These are indicative ranges rather than universal accuracy thresholds. The appropriate resolution depends on the underlying inputs, model methodology and use case.
For example, flooding is particularly sensitive to elevation. A difference of tens of centimeters can determine whether a building is inundated, making high-quality elevation data important.
By contrast, drought is fundamentally a water-balance phenomenon operating across catchments, aquifers and supply systems. Reporting drought risk at extremely fine spatial scales can therefore imply a degree of precision that the underlying hydrology does not support.
Five myths about high-resolution climate risk data
Myth 1: Finer resolution means more accurate data
Not necessarily.
A smaller grid provides more spatial detail, but accuracy depends on the underlying inputs, methodology and validation. Modeled precision is not evidence of real-world accuracy.
Myth 2: The quoted resolution is the resolution of the science
No.
A delivered grid can be substantially finer than the native resolution of its underlying climate inputs.
Buyers should ask providers to report the native resolution of the climate forcing, elevation data, land-cover data and other important inputs separately.
Myth 3: Resolution is the largest source of uncertainty
Usually not.
Other variables can materially influence a climate risk estimate, including the emissions scenario, time horizon, climate model, downscaling method, observational dataset, damage function and treatment of adaptation.
Spatial resolution is only one part of that uncertainty chain.
Myth 4: Two climate risk datasets at the same resolution are comparable
No.
Identical grids can produce different estimates of hazard intensity, damage and relative risk because the underlying models and assumptions differ.
Resolution is therefore not a standardized measure of data quality.
Myth 5: High-resolution climate data replaces engineering diligence
It does not.
A hazard model cannot know whether an individual building has a flood barrier, what its finished floor elevation is, whether critical equipment is located in a basement, or whether a roof has been reinforced.
High-resolution climate risk analytics can narrow the geographic and hazard context. They cannot replace asset-specific engineering information.
What does AlphaGeo mean by high-resolution climate risk data?
AlphaGeo’s approach is to match spatial detail to the physical characteristics of each hazard rather than treating resolution as a standalone measure of model quality.
AlphaGeo provides globally consistent climate risk and resilience data covering nine acute and chronic hazards, across multiple emissions scenarios and time periods. Its climate forcing draws on CMIP6 global climate models, while locally driven hazards incorporate high-resolution terrain, land-cover and built-environment data. Outputs are available at resolutions of up to 5-30 meters where the underlying data supports that level of granularity.
The platform combines 85 unique data sources spanning satellite observations, climate projections, historical records and socioeconomic data, with data refreshed quarterly.
But resolution is only one component of the approach.
Climate risk + adaptation
Two buildings within the same spatial cell can experience materially different outcomes from the same climate hazard.
The difference may come from drainage, flood barriers, fire response infrastructure, building characteristics or other adaptation measures.
This is why AlphaGeo’s Global Adaptation Layer incorporates information on local adaptation capacity alongside hazard exposure. For example, adaptation indicators can include drainage systems and flood barriers for inland flooding, coastal defenses and natural buffers for coastal flooding, and fire response and detection infrastructure for wildfire.
The result is intended to move beyond a simple question of:
How much hazard exposure exists?
toward:
How much of that hazard exposure is likely to translate into real-world impact, given the adaptation capacity of the location?
This distinction is particularly important for investors and asset owners using climate risk analytics to make capital allocation, underwriting, resilience and site-selection decisions.
Transparency matters
No climate risk model perfectly represents local reality.
AlphaGeo therefore publishes assumptions and limitations around downscaling, proxy measures, long-term projections and local variability, and positions its analytics as complementary to local expertise and engineering assessment rather than a replacement for them.
The goal is not to produce the smallest possible grid. It is to produce decision-useful climate risk data at a spatial scale that is appropriate to the hazard and supported by the available evidence.
Six questions to ask any climate risk data provider
If you’re evaluating climate risk analytics for an investment portfolio, insurance book, real estate portfolio or corporate footprint, don’t stop at the resolution number.
Ask:
1. What is the native resolution of each important input?
Ask separately about climate forcing, elevation, land cover, hazard observations and adaptation data.
2. What downscaling method do you use?
Understand whether the provider uses statistical downscaling, dynamical models, machine learning, geospatial hazard modeling, or a combination.
3. What is the effective resolution?
Ask:
At what scale does the model output still contain independently meaningful information?
This can be more informative than the headline delivered resolution.
4. Is the quoted resolution actually global?
Some providers may offer very fine data in selected developed markets while relying on substantially coarser inputs elsewhere.
A global climate risk dataset should be evaluated for consistency across geographies.
5. How has the model been validated?
Ask for validation against observed historical events and, where available, quantitative skill metrics — not just the resolution number.
6. What does the model know about adaptation?
A hazard score tells you about exposure to a hazard.
A more decision-useful climate risk assessment should also consider what prevents that hazard from becoming physical damage.
Ask what the model knows about adaptation — and what it assumes when that information is unavailable.
Frequently Asked Questions About Climate Risk Analytics and Hazard Resolution
What is climate hazard resolution?
Climate hazard resolution is the spatial scale at which a physical climate hazard is represented in a model or dataset, usually expressed in meters or kilometers. A finer resolution means smaller spatial cells, but does not necessarily mean greater accuracy.
What is high-resolution climate risk data?
High-resolution climate risk data represents physical climate hazards at relatively fine spatial scales, often down to hundreds or tens of meters. Its usefulness depends on the hazard, the underlying data and the modeling methodology.
Does higher resolution mean more accurate climate risk data?
No. Resolution and accuracy are different. A fine grid can provide useful spatial detail, but the accuracy of the result also depends on the underlying climate forcing, observations, downscaling method, hazard model, damage function and validation.
What is downscaling in climate risk analytics?
Downscaling is the process of translating information from a coarser climate model or dataset to a finer spatial scale. Statistical, dynamical, machine-learning and geospatial approaches can be used. Downscaling creates a finer modeled representation — it does not create new observations.
What is the best resolution for physical climate risk assessment?
There is no universal best resolution. Flooding, landslides and wildfire can benefit from relatively fine terrain and land-cover data, while hazards such as hurricane wind, drought and hail operate at larger spatial scales.
Why does climate risk data resolution matter?
Resolution matters because climate risk decisions are often made at the asset level. A portfolio manager may need to distinguish between properties, while an insurer may need to understand how hazards vary across an individual exposure. But the resolution must be scientifically appropriate to the hazard.
What is effective resolution?
Effective resolution is the finest spatial scale at which a model output contains meaningful, defensible information. It can be substantially different from the resolution at which a provider delivers its dataset.
What should I look for when comparing climate risk data providers?
Look beyond the headline resolution. Compare the native resolution of inputs, downscaling methodology, effective resolution, geographic coverage, validation, uncertainty, hazard-specific modeling and treatment of adaptation.
Final takeaways
Spatial resolution is an important specification when evaluating climate risk data. But it is not a measure of accuracy by itself.
A credible climate risk analytics provider should be able to explain:
- where its underlying data comes from;
- the native resolution of its inputs;
- how those inputs are downscaled;
- the effective resolution of each hazard;
- how the models are validated;
- how uncertainty is quantified; and
- how adaptation and vulnerability are represented.
The best climate risk data is therefore not necessarily the data with the smallest grid.
It is the data that provides the right level of spatial detail, scientific validity and transparency for the decision being made.
Sources and further reading
- Miller, S., Ormaza-Zulueta, N., Koppa, N. and Dancer, A. (2025). Statistical downscaling differences strongly alter projected climate damages. Communications Earth & Environment, 6, 145.
- Fiedler, T., Pitman, A.J., Mackenzie, K., Wood, N., Jakob, C. and Perkins-Kirkpatrick, S.E. (2021). Business risk and the emergence of climate analytics. Nature Climate Change, 11, 87–94.
- Chegwidden, O.S. et al. (2022). Open data and tools for multiple methods of global climate downscaling. CarbonPlan.
- Fathom. Precision versus accuracy: the effect of grid size on model accuracy.
- Climate X. Flood model resolution: does size matter?
