Climate-risk screening is often described in a single number: the resolution of the grid behind it. A 0.25° global grid is a sensible, widely used backbone for climate data, and it is the common reference our models share. But it is a backbone, not the whole model. Credible multi-hazard analysis supplements that common grid with hazard-specific datasets at different, and often finer, resolutions. Understanding why is the key to reading a result with the right expectations.
What a 0.25° grid does
A 0.25° global grid divides the world into cells and assigns climate variables to each one. Used as a backbone, it provides:
- Consistent global coverage.
- A common climate-data reference across every hazard.
- Comparable treatment across countries.
- Efficient processing for large portfolios.
- Coverage of remote locations and supply-chain sites that local datasets often miss.
One technical point is worth stating plainly: the physical size of a 0.25° cell varies with latitude. It should not be described as a fixed number of kilometres everywhere. The grid is a consistent coordinate system for climate variables, not a guarantee of a uniform footprint on the ground.
Why a climate grid is not a site survey
A global climate grid, however well constructed, does not capture the local features that often decide whether an asset is actually affected. It does not see:
- Drainage capacity.
- Flood defences.
- Building elevation.
- Shading and ventilation.
- Slope-stability interventions.
- Local water storage.
- Construction standards.
- Site-specific emergency arrangements.
This is precisely why results are described as screening rather than as engineering conclusions. The grid tells you about the climate an asset sits in. It does not tell you how that asset has been built or defended.
Different hazards require different spatial structures
The most important reason not to force every hazard onto one grid is that hazards behave differently in space. A few examples show why.
Extreme heat and extreme cold
Temperature hazards rely substantially on climate variables and behave relatively smoothly across space. They are well suited to the common global grid, used with several temperature-related indicators rather than a single threshold.
River flooding
Fluvial flooding requires hydrological or hydrodynamic information describing river systems, flow and flood extent. The underlying flood data can be available at a finer resolution than the climate grid, because where water goes is governed by the river network and terrain, not by climate cells alone.
Pluvial flooding
Surface-water flooding is sensitive to local terrain, built-up surfaces, rainfall intensity and drainage. Screening can use higher-resolution terrain and urban information to sharpen the signal, but local drainage data remain important and are rarely available globally. The screen flags where surface water is likely to concentrate; it does not model a drainage network.
Coastal flooding
Coastal exposure depends on a site's relationship to the coastline, its elevation, sea-level rise and storm-related water levels. Point elevation and coastline geometry are far more relevant here than a broad climate-grid value assigned to a large cell.
Tropical cyclones
Cyclone screening depends on storm tracks, wind fields, historical occurrence and compound rainfall, rather than only on fixed grid cells. The relevant spatial structure is the track of the storm, not a static map of climate averages.
Landslides
Landslide susceptibility combines climate signals, such as extreme rainfall, with slope, terrain and land cover. The climate signal alone cannot identify an unstable slope.
Native resolution and model governance
It is tempting to assume that a finer pixel always means a better answer. It does not. Higher resolution does not automatically mean higher accuracy. A dataset may be:
- Spatially detailed but poorly calibrated.
- Based on uncertain assumptions.
- Inconsistent between countries.
- Unsuitable for the hazard being assessed.
- Too detailed for the quality of the asset coordinates supplied.
That last point matters more than it first appears. If an asset's location is only known to within a few hundred metres, a one-metre hazard map adds false precision, not insight. Good modelling means choosing datasets that are appropriate, traceable and consistent for the hazard and the data at hand, not simply selecting the smallest available pixel. Each hazard's data sources and resolution are set out in our model register.
Why this matters for global portfolios
Most organisations of any scale have a mix of asset types: offices, warehouses, dealerships, factories, distribution centres, supplier locations, ports and logistics nodes, and customer-facing sites. A consistent global backbone allows the whole portfolio, and much of the value chain, to be screened on the same basis. Hazard-specific datasets then add detail where it is useful and defensible. The two work together: breadth from the backbone, depth from the hazard-specific layers.
Appropriate language for reporting
Because resolution varies by hazard, disclosures should be specific rather than sweeping. A credible note identifies, for each result:
- The dataset used.
- The spatial resolution.
- The scenario.
- The time horizon.
- The baseline.
- The model's limitations.
- The intended use.
Stating these avoids implying a false precision that the underlying data cannot support, and it lets a reader judge how much weight a particular result can bear.
Conclusion
The strength of a multi-hazard model is not that every hazard is forced into the same spatial structure. It is that the analysis uses a consistent climate backbone for comparability while retaining the datasets and methods appropriate to each hazard. A 0.25° grid is where the analysis starts, not where it ends. The methodology sets out how the backbone and the hazard-specific layers fit together.