Alpha-Klima takes a portfolio of physical assets and returns what climate change does to it in money. Hazard at each location, a vulnerability model for each asset class, impact as a distribution, then aggregation into annual average loss, value at risk and expected shortfall.
No result collapses into an unexplained score. Every figure records the dataset, vulnerability, and methodology that produced it, so it can be reviewed rather than trusted.
Free to enter with a Google account, including sample hazards and example analyses.
Percentile (%)
Financial Materiality
| Wildfire | Annual Wildfire Probability | Scored on modelled damage to this assetLosses modelled on structure and inventory given the probability of wildfire occurrence at the asset location. |
| Heat | Cooling Degree Days | Scored on losses associated to this assetEnergy consumption in kWh given the efficiency of the existing cooling solution for a Eurostat 21°C reference base. |
| Riverine Flood | Riverine Flood Depth | Scored on losses associated to this assetModelled on the frequency and intensity of flood events mapped to losses through structure, inventory and disruption vulnerability models. |
| Heat | Mean Work Loss | Scored on losses associated to this assetMean productivity loss due to heat stress for medium intensity labor. |
| Hail | Annual Large Hail Probability | Scored on modelled damage to this assetModelled on the frequency and intensity of hail events mapped to losses through structure vulnerability models. |
| Landslide | Landslide Susceptibility | Scored on modelled damage to this assetModelled on the frequency and intensity of landslide events mapped to losses through structure and inventory vulnerability models. |
| Wind | Maximum Gust Wind Speed | Scored on modelled damage to this assetModelled on the frequency and intensity of high wind speeds mapped to losses through vulnerability models differentiating structure and building characteristics. |
From model to decision
Hazard × exposure × vulnerability = financial materiality
Documentation opens after sign-in. Product and API guidance is available to any account, while the methodology sections are reserved for accounts holding platform credit.
Hazard data is increasingly a commodity. What decides whether a number holds up is the modelling on top of it and the record of how it was produced.
40+ indicators across 14 hazards covering: riverine and coastal flooding, windstorms, wildfires, landslides, subsidence, water stress, drought, precipitation, hail, temperature-related, earthquakes, snow and freezing rain. We use the latest GCM-based projections alongside satellite and observational datasets, with full support for RCP/SSP scenarios across time slices and regions. All inputs are consistent and versioned end-to-end to ensure transparency and auditability.
Vulnerability functions are selected for what an asset is, not only where it sits, covering real estate, industrial and manufacturing sites, power generation and grid infrastructure, transport, telecoms, utilities and agriculture. They separate physical damage to replaceable value from loss of function and operating costs, so a site that survives an event but stops producing is still counted. The relationships come from published academic and institutional research, which means an estimate can be traced back to the study behind it.
Impacts stay as distributions rather than point estimates, so tail behaviour survives the calculation instead of being averaged away at the first step. Aggregation to cluster and portfolio level uses a copula-based dependence structure, because nearby assets are hit by the same storm and hazards co-occur. From the aggregated distribution the engine derives Annual Average Loss, Value at Risk and Expected Shortfall, per hazard and combined.
Asset-level results are only the input. The portfolio view groups assets into spatial clusters, shows what each cluster contributes to total loss, and makes concentration visible as a map rather than as a table of stand-alone figures. It is where a portfolio stops being a list of addresses and starts being a risk position that can be compared, ranked and acted on.
Credit exposure is not physical, so the work starts by locating the assets behind each counterparty, with documented proxies where a location cannot be observed. The same pipeline then produces the EU CRFR Pillar 3 template, the ECB PEAR, NEAR and CEAR measures, and a counterparty overlay carrying damage and interruption through to PD, impairment and credit spreads.
Results are meant to leave the platform. A portfolio export carries asset-level risk scores and the portfolio's financial metrics as a spreadsheet, and every chart the interface draws can be saved as an image for a report or a slide. For anything larger the same results come back through the API, and a set of example notebooks works through the whole pipeline, from a first call to a finished figure.
Analyses belong to your organisation rather than to whoever ran them, so a portfolio one colleague uploads is available to the rest of the team with its results attached. Owners invite members and control access, and usage is metered against a shared allowance so consumption stays predictable. In practice that means a climate assessment stops being a spreadsheet on one person's laptop and becomes something the team can return to.
Everything the interface produces is reachable programmatically with a self-service API key, including the intermediate detail that makes a result reproducible: hazard exceedance distributions, vulnerability distributions and the explicit data paths behind them. Portfolios can be processed in batch, spatial aggregation works over arbitrary geometries such as NUTS regions, postal codes and custom polygons, and available hazards, indicators, scenarios and asset classes are discoverable through the API rather than hard-coded against a fixed list.
Try it
Free to enter with a Google account, including sample hazards and example analyses.
Auditability
A number that cannot be explained will not survive an auditor, supervisor or internal validation team. Every result carries its evidence with it: the indicator and its provider, model and version, the scenario and the averaging window behind the year, the asset class and any overrides, the vulnerability function selected, the aggregation and its dependence structure, and the units and thresholds applied.
That traceability lets one analytical foundation serve risk management, disclosure, assurance and regulatory workflows instead of being rebuilt for each.
One analysis, many frameworks
Where a programme needs more than the analysis itself, our team runs it: we research where your exposure physically sits, calibrate vulnerability to those assets, and carry the loss through to PD, valuation, insurance or the adaptation business case – inside the framework you already run, documented to survive review.
We contribute to OS-Climate's open-source physical risk stack, principally to physrisk, the calculation engine at its centre, and to the hazard layers it reads. Open foundations let a supervisor or an auditor inspect the method directly. The platform builds on them with our own financial engine, vulnerability inventory, portfolio pipelines and datasets.
