How Geo Prithvi Report works
This page describes what a Geo Prithvi Report is built from, how the 0 to 100 risk score is worked out, and where the numbers are weak. Everything here matches the code that generates the reports today.
A screening tool, not a survey. The score and the findings are statistical, desk-based estimates from public datasets. They are not a legal certificate, a geotechnical investigation or engineering advice.
Data sources
| Source | What we take from it | Limits |
|---|---|---|
| Open-Meteo (ERA5 reanalysis) | Daily rainfall, wind, and maximum and minimum temperature for the last 10 full calendar years | Gridded data, so it describes the surrounding area (roughly 25 to 30 km), not one point. It does not capture river overflow from rain that fell upstream. |
| USGS earthquake catalogue | Earthquakes of magnitude 3.0 and above within 100 km over the last 10 years. The full list is used for the statistics (M4 and above only in extremely active areas, where the list would be too large); the report table shows the 50 strongest at most | Smaller events and some remote areas are under-recorded. |
| Open-Elevation (SRTM-based) | Height above sea level and a slope estimate | Typically good to a few metres. A backup service is tried if it is down; if neither answers, no report is produced. |
| OpenStreetMap (Nominatim) | Address label and country for the pin | Display only. The pin's coordinates drive the analysis. |
| OpenStreetMap (Overpass), Premium only | Major roads, power substations and lines, water works and water towers within 1.5 km | Edited by volunteers, so rural areas are often incomplete. A missing feature is not proof it does not exist. |
Geo Prithvi Report is independent of all of these providers and is not endorsed by any of them.
How the risk score is built
The score is the sum of six components, capped at 100. The thresholds and point values are our own judgement of how much each factor usually matters. They are not taken from any standard, and a different reasonable person might set them differently.
| Component | Max | How points are given |
|---|---|---|
| Flood probability | 40 | Rises with the total number of days above 80 mm in the 10-year record and with the number of years that had at least one such day (2 base points, plus about 1.2 per day and 1.5 per affected year, never above 40). |
| Elevation | 20 | Under 5 m: 20. 5 to 15 m: 17. 15 to 30 m: 12. 30 to 60 m: 7. Above 60 m: 2. |
| Terrain slope | 10 | Flat ground drains slowly: under 0.3 degrees: 10; 0.3 to 1: 7; 1 to 3: 4; 3 to 8: 1. Steep ground adds slope-stability risk: 8 to 15 degrees: 4; 15 to 25: 7; 25 or more: 10. |
| Seismic level | 15 | Very High (zone V): 15. High (IV): 10. Moderate (III): 5. Low (II): 2. |
| Recorded earthquakes | 10 | Strongest event in the 10-year record (outside India, after reducing magnitude with distance and depth). M6 or more: 10. M5 to 6: 7. M4 to 5: 4. M3 to 4: 2. None: 0. |
| Rainfall-moisture index | 5 | From the rainfall-based index described below. Above 0.65: 5. Above 0.45: 3. Otherwise 1. |
A score of 0 to 34 is shown as low, 35 to 64 as medium and 65 or above as high. The project type you choose does not change the score. It changes the recommendations and checklists.
Safety floors. A weighted sum can hide one severe hazard behind several mild ones, so some hazards set a minimum score. A Very High seismic level or a HIGH flood indicator sets at least 55 (medium). A High seismic level, terrain of 15 degrees or steeper, or elevation under 10 m, sets at least 40 (medium). A High or Very High seismic level together with an earthquake of M6 or more in the record sets at least 65 (high). The report lists the hazards that drove the verdict.
A low score means fewer risk indicators in the data we use. It does not mean the land is safe.
Flood and rainfall
We count the days in each of the last 10 full calendar years on which rainfall exceeded 80 mm. This is our own screening threshold, not an official category (the India Meteorological Department, for example, calls 64.5 to 115.5 mm "heavy" and 115.6 to 204.4 mm "very heavy"). The flood level is HIGH at 30 or more such days in the 10 years, MEDIUM at 8 or more such days or when at least 4 years had one, and LOW otherwise. The report also shows in how many of the 10 years such a day occurred.
ERA5 is a grid of about 25 km, so it smooths out local cloudbursts and flash floods: the count under-states local extremes, especially in hills and mountains. Snow-melt floods are not included.
This measures how heavy the rain is, not whether a particular plot will go under water. Drainage, nearby rivers and local embankments are not in the data.
Seismic zone and earthquakes
In India, the seismic zone is worked out from coordinate ranges that approximate the IS 1893:2016 zone map. Northeast India, Kutch, the Andaman and Nicobar Islands and the far north are treated as Zone V, and other regions are assigned Zone IV, III or II by similar ranges. This is an approximation of the official map, not a copy of it, and the zone can differ by one step at town level. The zone is never reported lower than the site's own recent seismicity supports. Where our coordinate ranges are least certain, for example the Bihar and eastern Uttar Pradesh plains, the report shows a range such as "Zone III–IV" instead of a single zone, and the higher value is used in the risk score. Always confirm the exact zone for the town with your structural engineer.
Outside India, no code zone or national hazard map is applied. The level (Low to Very High) comes from the 10-year USGS earthquake record within 100 km, using the complete list returned by USGS (not a capped list). Each event's magnitude is reduced with distance beyond 20 km (about 0.8 magnitude units per tenfold distance) and for depth (0.4 units for events 100 to 300 km deep, 1.2 for deeper ones, which are also left out of the sequence count), and aftershocks (M4+ within 30 days and 50 km of a larger event) are counted together with their mainshock. Effective magnitude 6.2 or more gives Very High; 5.5 or more, or 12 or more independent M4+ sequences, gives High; 4.5 or more, or 3 or more sequences, gives Moderate; otherwise Low. A quiet 10-year record does not rule out hazard from faults that rupture rarely, so the report says to check the local hazard map. The national design code (where one is listed) is named only as the code to design to, never as the source of the level.
The earthquake log in the report lists the strongest events with date, magnitude, depth and distance.
Elevation and slope
Elevation comes from Open-Elevation, with Open-Topo-Data (SRTM 30 m) as backup. Slope is calculated from five points (the site and four neighbours about 280 m away), so it describes broad terrain rather than small local gradients. If no service returns valid terrain data, no report is produced and nothing is charged: we do not substitute a regional default.
Soil, moisture index and climate indications
These are the weakest parts of a desk report, so we say plainly how they are made.
- Soil class is assigned only for India, from broad latitude and longitude regions (for example the Brahmaputra plain, the Indo-Gangetic plain, the Deccan, the Himalaya). It is a regional indication, not a measurement at your plot. Outside India no soil map is used and the report says "Not assessed".
- Rainfall-moisture index (shown as a proxy, not as NDVI) is average annual rainfall divided by 3,000, limited to between 0.1 and 0.85. It is not measured from satellite imagery and is only a rough indicator of how wet the climate is. The monthly rainfall chart in Premium uses real ERA5 monthly averages.
- Foundation notes are qualitative prompts for the regional soil class (for example "expansive soil: confirm with a soil test"). The report does not quote bearing-capacity numbers, because they vary widely within one plot and only a geotechnical survey can give them.
When we do not produce a report
A wrong report is worse than none. All data is fetched before payment opens. If the terrain, the 10-year rainfall archive, the earthquake catalogue or the location lookup is missing or incomplete, or the site appears to be open water, you see an error and are not charged. We do not fill gaps with default values.
Heat and cold
Heat is rated high if the average daily maximum is above 36 °C or the hottest day above 44 °C, and moderate if the average daily maximum is above 32 °C or the hottest day is 38 °C or more. Cold is rated high if the average daily minimum is at or below −7 °C or the coldest day is at or below −20 °C, and moderate if the coldest day is below −5 °C or the average minimum below −3 °C. These are our own thresholds, based on the same 10-year record.
Legal flags (Standard and Premium)
Legal flags are rule-based reminders of things to verify, triggered by the data. For example: a seismic level of High or above, a medium or high flood level, elevation under 10 m, terrain of 15 degrees or steeper (or 8 to 15 degrees together with a heavy-rain history), a pin in North-East India (where river flooding is not in the rainfall data), or a country where a flood-map, hazard-map or planning check is customary (the US, Australia and New Zealand, Japan, Saudi Arabia). If none applies, the report says no critical flag was found and names the authority to check with.
This is not a check of land records, title, zoning or permits. Those need the local authority and a lawyer.
Premium extras
- Climate-based indications describe what the rainfall pattern normally supports (for example a wet climate with dense vegetation) and how often heavy-rain days occurred. They are inferences from rainfall only, not land records or imagery, and carry Medium or Low confidence.
- Air quality is shown as "not assessed" except for one documented regional flag (the Indo-Gangetic belt in India). No sensor data is used.
- Nearby mapped infrastructure comes from OpenStreetMap, as described above.
- Decadal comparison splits the 10-year record into an earlier half and a recent half and compares average rainfall (including years with no heavy rain), heavy-rain days per year and earthquake counts. Differences between two 5-year averages are often within normal variability, and the report says so.
- Monthly rainfall profile shows the real average rainfall for each calendar month from the same 10-year ERA5 archive, with the wettest three months highlighted.
- 5-year rainfall outlook is a straight-line trend fitted to the 10 annual rainfall totals and extended five years. Over a 10-year record this is a weak signal, and it is not a weather forecast. If the fit is weak (R2 below 0.3) the report says there is no clear trend. Otherwise a change of more than 1.5% of the average a year is labelled increasing or decreasing, and anything smaller is stable. The report shows the fit (R2) and rates reliability as Low, except Medium for the first two projected years when R2 is above 0.5.
Country standards
The country is taken from the pin's address. The report names the standard it refers to on the first page and never applies one country's code to another.
| Where | Seismic reference used in the report |
|---|---|
| India | IS 1893:2016, NBC 2016 |
| United States | ASCE 7-16, IBC 2021 |
| Canada | NBC Canada 2020 |
| United Kingdom | BS EN 1998 with UK National Annex |
| Many European countries, Ireland, Singapore, Malaysia | Eurocode 8 (EN 1998-1) |
| Japan | Building Standard Law and AIJ |
| Australia, New Zealand | AS 1170.4 / NZS 1170.5 |
| China | GB 50011-2010 |
| Saudi Arabia | SBC 301 |
| Everywhere else | No national code applied. The report says so and refers you to the local code. |
Data confidence
Each report states how much to rely on it. It is capped at 90, because every output is a model or proxy estimate, and it is 90 whenever all primary sources responded. It drops by 10 only if a Premium factory report could not fetch the nearby-infrastructure data. A missing weather, terrain or earthquake source does not lower it, because in that case no report is produced at all. The figure shows data availability, not accuracy.
What happens when you generate a report
- You place a pin, choose a project type and a plan, and accept the terms.
- Our server fetches the data before any payment is taken. If the weather record cannot be retrieved, the report is blocked and you are not charged. Other sources either have a backup or block the report in the same way.
- You pay through Razorpay.
- After payment is verified, the PDF is built in your browser and downloaded together with a GST invoice as a ZIP file.
See the privacy policy for what is stored and the delivery and refunds page for what happens if something fails.
Known limitations
- Rainfall grid: ERA5 is about 25 km wide, so local cloudbursts and flash floods in hills are under-counted. A LOW flood level in hill country is weaker evidence than in a plain.
- Heavy-rain threshold: one common threshold (80 mm in a day) is used everywhere; local services may use different values for their own climate.
- Seismic: outside India there is no national hazard map in the data, only the 10-year USGS record. In India the zone is an approximation of the IS 1893:2016 map, not a district lookup.
- Large areas: a report describes the pinned point. For a large district or county, other parts can differ.
- Soil and land use: indications only, with no soil map outside India and no land records or imagery anywhere.
Using a report well
Use it to shortlist plots and to prepare questions for your engineer. Before you buy or build, verify the findings on site, check land records with a lawyer, and commission a geotechnical survey. Our report versus survey guide explains the order.