Product FAQ
Last updated
Last updated
Overall
Are there plans to develop risk analytics for other climate hazards other than the current hazards? Yes, AlphaGeo plans to enhance the Risk and Resilience Index by incorporating additional hazard types such as hail, landslides, cold storms, etc. and even non-climate hazards such as earthquakes.
Key links:
What are your primary data sources? Our Risk and Resilience Index integrates data from 85 unique sources, encompassing satellite observations, climate model projections, historical weather records, and socio-economic datasets.
Notable sources include CMIP6 climate projections, Aqueduct 4.0, Copernicus, IPCC reports, NOAA's IBTrACS, OpenStreetMap, WorldPop, and the World Bank.
More details at: Data Sources
Why do you choose to use open-source data? Utilizing open-source data ensures transparency, reproducibility, and broad accessibility. It allows AlphaGeo to build upon peer-reviewed and well-documented datasets, fostering trust and enabling users to understand and verify the underlying data and methodologies.
What is the value you provide, if underlying data is open-source? AlphaGeo adds value through rigorous data curation, advanced geospatial data processing and engineering, and the calculation of resilience-adjusted risk scores for actionable insights that go beyond raw data.
Learn more about how we transform raw adaptation data into standardized, actionable insight here: Methodology (2/3): Global Adaptation Layer
What is the global coverage of your data? AlphaGeo's datasets offer global coverage, enabling assessments of climate risk and resilience for any location worldwide.
What is the data resolution? The data resolution reaches up to 30 meters. For a detailed breakdown on data source resolution, please request a copy of our Data Dictionary.
What GCMs are used for future scenarios? How are they downscaled?
We use an ensemble of GCMs such as CanESM5, MRI-ESM1 and downscale them according to the methods described by
The native resolution of some data sources are far coarser than the building/parcel level resolution of your analytics. How have you downscaled this data?
We employ a combination of machine learning algorithms and spatial interpolation techniques. These methods enable us to estimate finer resolution data by identifying patterns and distributing data values appropriately across smaller units. Additionally, we validate the downscaled data by comparing it with high-resolution reference datasets to ensure accuracy and reliability.
Is the resolution of your features uniform across all global locations? Specifically, how granular is your coverage for developing countries or rural zones? Does this differ depending on risk versus adaptation features?
All features share the same resolution globally.
How frequent are data updates? Quarterly.
Do you have historical or current risk data?
Yes, we have historic, current and future climate data from 1975 to 2100.
Does your data incorporate historical weather events, i.e., incidents? Does it include the severity of the weather event and how is that measured?
We have access to publicly available historical data on events such as hurricanes, wildfires, and major floods around the world. However, we do not incorporate these events into our scoring or forecasts due to the limited number of observed occurrences. Instead, we rely on simulated events, such as hurricane tracks and flood inundation levels, to provide a more comprehensive assessment of risk across all areas, rather than focusing solely on locations with a history of hazardous events. Historical events are utilized for accuracy benchmarking rather than directly influencing projections.
How is AlphaGeo's data different from what FEMA provides?
We have more granular data, and cover multiple SSP scenarios and time periods:
AlphaGeo's data is more granular. Our data is provided at the building/parcel-level, compared to FEMA's neighbourhood or block-level data.
AlphaGeo covers 3 emissions scenarios (SSP245, SSP370, SSP585) and 4 time periods (2025, 2035, 2050, 2100). FEMA focuses on current and historical flood risk based on observed data and past events. They do not project future flood risk under climate scenarios, and are less suitable for scenario analysis or forecasting.
How do flood models account for precipitation?
Our scores are based on rainfall intensity and flood inundation levels. For example, compared to Singapore, Bangkok receives approximately half the amount of precipitation and has fewer days with extreme precipitation, leading to lower levels of inundation due to rainfall.
Key links:
What framework or methodology do you use to select and group risk features for each hazard?
Features are grouped based on their correlation to key components of climate risk:
Exposure: The degree to which a location is exposed to climate-related hazards.
Frequency: How often a location might experience these hazards.
Intensity: The severity of the impacts when these hazards occur.
What is your hazard forecasting methodology?
Drawing from risk engineering practices for each risk category, we assign a damage function that maps the intensity (I) of the exposure to the mean damage ratio (MDR). The resulting MDR calculated from the intensity (I) represents the estimated damage on an exposed asset or a location given a certain climate hazard. As intensity of the hazard changes over time under different climate change scenarios for the target location, the MDR of a location change over time as well. Refer to Methodology (1/3): Resilience-adjusted Risk Framework
How do you define, or where do you obtain, the damage functions for each hazard, and its change over time/scenarios?
We define the range, threshold, and type of damage functions for each type of hazard through existing scientific literature and domain experience.
Is the change in MDR modelled at the hazard (e.g., coastal flooding) or feature level (e.g., mean sea level rise)?
Feature level. Each feature has a corresponding damage function.
How are the resilience-adjusted risk scores calculated? We apply an adaptation offset coefficient to the intensity of each hazard, based on data on the local adaptation measures in place.
How do you determine the degree of adaptation offsets? What thresholds are used, and is this based on any established method?
The adaptation offset is based on an aggregated Resilience Score for each location. Each hazard has its own corresponding resilience sub-index, which is comprised of its underlying adaptation features. For example, the aggregate Resilience Score for Hurricane Resilience is computed from the sum of Building Strength, Flood Defenses, and all features under General Societal Resilience. This produces a "resilience aggregate" of each adaptation measure.
The Resilience-adjusted Risk score is derived by subtracting this aggregated "Resilience Score" from the Physical Climate Risk score. This adjustment is weighted, with 70% assigned to the physical risk and 30% to the resilience measures (15% for local adaptation measures, 15% for societal resilience), ensuring that both aspects are appropriately considered in the final assessment.
Are your methods peer-reviewed? How can we trust your methodology?
AlphaGeo's analytics are built upon the evolving peer-reviewed data sources and literature. With the open access nature of our methodology and documentation, we invite the scrutiny and feedback from all experts in the field to continuously enhance our methods.
What are the biases and limitations of your methodology?
Data Limitations: Our methodology relies on the quality and quantity of available data. Insufficient or biased data can impact the results.
Assumptions: The analysis is based on certain assumptions that may not hold true in every scenario.
Model Bias: The algorithms used may have inherent biases, affecting the objectivity of predictions or outputs.
Static Framework: The methodology may not account for immediate changes in the environment or context.
Understanding these limitations helps in interpreting results more accurately and applying improvements to future iterations.
Key links: Methodology (3/3): Indexing
What do your scores represent, and how are they calculated?
Our hazard-specific risk scores (e.g., Hurricane Risk) are absolute risk scores, while the overall scores (e.g., Overall Physical Climate Risk) are relative scores.
Hazard risk scores: Absolute risk profiles for any one type of risk for any given scenario and time period, on a scale of 0-100. Each score is assigned based on thresholds determined for each risk feature. The risk category thresholds are based on the Mean Damage Ratio (MDR) for the respective features.
Overall scores: Global percentile scores that measure overall risk profile of a location relative to other locations, allowing users to compare across space.
Details are available at our scorecard here: Methodology (3/3): Indexing
Do the overall risk scores change if a different benchmark is selected? Or is it always relative to a global distribution?
No, the overall risk score are percentile scores on a global distribution. The benchmark helps the user to compare the performance of the selected location to a narrower geographic region (i.e. country or city).
What are the implications you're stating in your Risk and Resilience Index Scorecard based on? The implications are based on the expected damages caused by the hazard for each risk category. This is based on industry standards and the thresholds in the damage functions
How are features benchmarked?
All features are benchmarked globally, i.e., the indicators are scored based on datasets that are globally available. For example, FEMA Flood Zone data is not used in scoring Singapore or the US. We continue to include these datasets, however, for users who wish to pull these engineered risk features to create their own models.
How are "regional averages" calculated?
We calculate the regional average by aggregating data from the sub-provincial level. In most countries, this means the data is aggregated at the city/county/municipality level. In the case of Singapore, for example, since it's a city, the sub-provincial level will be the five regions. When you access data for a property in Singapore, regional-level data aggregates data in those regions. If the access point is Asia Square, then the sub-provincial level is the average of the Central Region of Singapore.
Which scenarios do you support, and why?
We support three emissions scenarios: SSP2-4.5, SSP3-7.0, SSP5-8.5. These are the CMIP6/IPCC AR6 scenarios.
Key links: AlphaGeo for Regulatory Disclosures
Are your hazards aligned to the EU taxonomy? Yes, AlphaGeo's assessments align with the EU Taxonomy Regulation by identifying material physical climate risks and evaluating adaptation measures that substantially reduce these risks. The methodology incorporates best practices and guidance from the IPCC and other scientific bodies.
How does your data help with regulatory disclosures? AlphaGeo's Climate Risk and Resilience Index supports compliance with various regulatory frameworks, including the Task Force on Climate-Related Financial Disclosures (TCFD), International Sustainability Standards Board (ISSB) IFRS S2, EU Taxonomy Regulation, and the EU Corporate Sustainability Reporting Directive (CSRD). The platform provides standardized risk scores, adaptation assessments, and financial impact metrics, facilitating comprehensive and transparent climate-related disclosures.
If you have further questions, please feel free to reach out to us at info@alphageo.ai.