# EVE Data Product Documentation
Last revision date: 22 October 2026
## Contents
- [1. Document Information](#document-information)
- [1.1 Purpose and Scope](#purpose-and-scope)
- [1.2 Intended Users](#intended-users)
- [1.3 Version, Ownership, and Citation](#version-ownership-and-citation)
- [2. Product Overview](#product-overview)
- [2.1 What EVE Final Data Provides](#eve-final-data)
- [2.2 Coverage and Update Frequency](#coverage-and-update-frequency)
- [2.3 Official Access Points](#official-access-points)
- [3. Methodological Summary](#methodological-summary)
- [3.1 Source Data](#source-data)
- [3.2 Flood and Exposure Concepts](#flood-and-exposure-concepts)
- [3.3 Aggregation to Administrative Levels](#aggregation-to-administrative-levels)
- [3.4 Percentages and Denominators](#percentages-and-denominators)
- [4. Data Structure](#data-structure)
- [4.1 Product Inventory](#product-inventory)
- [4.2 Field Dictionary](#core-fields)
- [4.3 Reading the Values](#units-values-and-naming)
- [5. How to Use the Data](#how-to-use-the-data)
- [5.1 Basic User Workflow](#basic-user-workflow)
- [5.2 Filtering, Joining, and Downloading](#filtering-joining-and-downloading)
- [5.3 Interpretation Guidance](#interpretation-guidance)
- [6. Quality, Limitations, and Responsible Use](#quality-limitations-and-responsible-use)
- [6.1 Known Limitations and Uncertainty](#known-limitations-and-uncertainty)
- [6.2 Responsible Use and Disclaimers](#responsible-use-and-disclaimers)
- [7. Support Materials](#support-materials)
- [7.1 Glossary and Acronyms](#glossary-and-acronyms)
- [7.2 Contact and Further Information](#contact-and-further-information)
## 1. Document Information
### 1.1 Purpose and Scope
Events Visualization in Emergencies (EVE) is a public operational data product produced by the Office of Emergencies and Resilience (OER) of the Food and Agriculture Organization of the United Nations (FAO). It provides aggregated flood exposure information designed to support emergency monitoring, early action, food security analysis, geospatial analysis, programme planning, and public information products.
This document describes the final EVE data products available to users. It explains what the data contains, how it can be accessed, how indicators should be interpreted, and what limitations apply.
EVE uses remote sensing because field access during flood events can be delayed, unsafe, or impossible. Satellite-derived data provides a rapid and consistent way to screen flood exposure across large areas, especially during emergencies when timely field information is limited.
### 1.2 Intended Users
This documentation is intended for all users of the final EVE data products, including FAO technical teams, FAO country offices, government counterparts, external partners, analysts, emergency and early action teams, food security specialists, geospatial staff, programme planners, and public users.
Users are expected to have a basic understanding of administrative boundaries, tabular data, map layers, percentages, and time periods.
### 1.3 Version, Ownership, and Citation
| Item | Value |
| --- | --- |
| Document title | EVE Data Product Documentation |
| Product name | Events Visualization in Emergencies (EVE) |
| Producer | Food and Agriculture Organization of the United Nations (FAO) |
| Responsible unit | FAO Data in Emergencies (DIEM) |
| FAO Division/Office | Office of Emergencies and Resilience (OER) |
| Contact point | FAO DIEM contact page: https://data-in-emergencies.fao.org/pages/contactus |
| Publication status | Public operational documentation |
| Licence | Creative Commons Attribution 4.0 International (CC BY 4.0), unless otherwise specified |
| Standards alignment | Prepared in line with the FAO Statistical Standard on Data Visualizations and Dashboards, Version 1.0 |
When using EVE data, please cite as:
> FAO. (2026). *Events Visualization in Emergencies (EVE)*. Food and Agriculture Organization of the United Nations, Data in Emergencies (DIEM), Office of Emergencies and Resilience (OER). Available at: https://diem-eve.apps.fao.org/. Accessed [day month year]. Reporting period: [dekad and year].
## 2. Product Overview
### 2.1 What EVE Final Data Provides
EVE provides information on flood impact estimates, based on Earth Observations and exposure layers, aggregated by administrative unit and dekadal reporting period. The final public data products summarize flood-related indicators at ADM0, ADM1, and ADM2 levels. These products are designed for use in dashboards, downloads, analytical workflows, and API-based access.
The final aggregated data includes indicators related to flooded area and exposed assets, such as cropland, population, and livestock by species. These indicators support comparison across places and reporting periods, while keeping the data at an administrative summary level suitable for broad operational use.
Large raster and imagery products are not part of the standard public download package because of their size. They may be shared upon request, subject to the applicable FAO process and any relevant data-use conditions.
### 2.2 Coverage and Update Frequency
EVE flood extent, severity, and exceptionality analysis is performed globally. Aggregation to administrative levels is produced for a subset of approximately 40-50 countries. The country list is operational and may change by reporting period; users should consult the dashboard for the current list of available countries.
EVE final data is organized by administrative level:
| Level | Description | Public access |
| --- | --- | --- |
| ADM0 | Country-level summaries | Dashboard, API, CSV |
| ADM1 | First-level administrative summaries | Dashboard, API, CSV |
| ADM2 | Second-level administrative summaries | Dashboard, API, CSV |
The temporal unit is the dekad. A dekad is a standard ten-day reporting period used in agricultural and food security monitoring. Each month contains three dekads: days 1-10, days 11-20, and day 21 through the end of the month.
EVE is updated on a dekadal operational cycle. On the first day of each dekad, EVE analyses and publishes data for the previous dekad. Publication is usually completed before 12:00 CET, subject to source data availability, processing conditions, and service availability.
Historical records are treated as immutable once published. If an exceptional correction is ever required, it should be communicated as a formal product correction rather than silently replacing the published historical record.
### 2.3 Official Access Points
The official public access point for EVE final data is the EVE dashboard: https://diem-eve.apps.fao.org/. Dashboard pages provide direct user access and additional information on available API endpoints and CSV download options.
Users should treat the dashboard, linked API endpoints, and CSV exports as the authoritative access routes for public aggregated EVE data. Any local copies or third-party reproductions should be checked against the official dashboard source and cited with the access date.
EVE is designed to support interoperable use in analytical and humanitarian workflows through public downloads and API access. Administrative identifiers are based on FAO GAUL codes and names.
EVE data is accessible through the following direct links:
| Product | Access link |
| --- | --- |
| ADM0 data | https://data-in-emergencies.fao.org/datasets/f0e0a74518e54394a7e5d5a90bfd3e25_6/explore |
| ADM1 data | https://data-in-emergencies.fao.org/datasets/3a5bb86bbc2b4a539138faf84a68af07_7/explore |
| ADM2 data | https://data-in-emergencies.fao.org/datasets/b9325ce882f44c9aab178e29afed3abf_0/explore |
To download filtered CSV data from these pages, users should filter the dataset by country and reporting period, select **Download data**, choose **Toggle filters**, and then download the filtered result as CSV.
Requests for large raster or imagery products should be submitted through the FAO DIEM contact page.
## 3. Methodological Summary
### 3.1 Source Data
EVE final indicators are produced from a combination of satellite flood information, a historical flood reference, administrative boundaries, land-cover information, population data, livestock data, and crop-calendar information used for interpretation. The core sources are summarized below and then described individually.
| Theme | Source |
| --- | --- |
| Flood detection | VIIRS 5-day composite flood product from NOAA operations, accessed through the SSEC flood product service |
| Global flood reference | VISTA (VIIRS Surface Water Temporal Atlas) |
| Land cover and cropland | ESA WorldCover 2021 global land-cover product, used in the standard EVE workflow for all countries and modes |
| Local land cover and cropland | FAO Land and Water Division (NSL) land-cover data for Mozambique: Maputo and Gaza provinces (2022 - 10 m - 14 classes), used only in localization mode |
| Population | WorldPop population datasets |
| Livestock | FAO Gridded Livestock of the World 4 (GLW4), 1 km resolution |
| Administrative boundaries | FAO Global Administrative Unit Layers (GAUL) |
| Crop calendar context | FAO Crop Calendar |
#### Flood Detection - NOAA / SSEC VIIRS Flood Products
**Provider:** NOAA operations, accessed through the SSEC flood product service
**Dataset:** VIIRS 5-day composite flood product
**Processed by:** FAO Data in Emergencies (DIEM-Flood)
**Links:** NOAA / SSEC VIIRS flood products; NASA VIIRS overview
**Role in EVE:** Primary flood-detection input
The VIIRS 5-day composite flood product is based on VIIRS flood observations from NOAA operational products and represents a composite of the previous five days. VIIRS, the Visible Infrared Imaging Radiometer Suite, is carried aboard Suomi NPP and NOAA-20 satellites. EVE uses the VIIRS flood information to identify flooded pixels, classify flood intensity, and create the dekadal flood layer used in the exposure analysis.
#### Global Flood Reference - VISTA
**Initiative:** Multi-stakeholder initiative coordinated by FAO and the European Commission's Joint Research Centre (JRC)
**Dataset:** VISTA (VIIRS Surface Water Temporal Atlas)
**Link:** VISTA repository
**Role in EVE:** Historical flood reference used to separate seasonally expected flooding from exceptional flooding
VISTA builds a global historical atlas from NOAA/GMU VIIRS flood observations, summarizing flood and surface-water conditions by dekad. Its products include time-specific dekadal and rolling-monthly archives and multi-year count, frequency, and dominant-class summaries for each recurring period. VISTA provides a historical reference for interpreting current observations.
EVE compares each dekadal flood layer with the VISTA monthly seasonal flood-frequency reference to identify where detected flooding is seasonally expected and where it is exceptional for the same period of the year. The monthly reference is not a calendar month: it is one of 36 rolling windows, each centred on one dekad of the year and covering the preceding, central, and following dekad. This gives the historical comparison one dekad of tolerance on either side, because weather-driven flooding is not precise to a fixed ten-day window.
#### Land Cover and Cropland - ESA WorldCover
**Provider:** European Space Agency (ESA)
**Dataset:** ESA WorldCover 2021 global land-cover product
**Link:** ESA WorldCover
**Role in EVE:** Standard cropland exposure and land-cover context
ESA WorldCover provides global land-cover information for 2020 and 2021 at 10 m resolution. EVE uses ESA WorldCover cropland classes in the standard workflow to estimate flooded cropland for all countries and modes. Cropland is extracted from the ESA WorldCover land-cover product and then intersected with the flood extent to compute flooded-cropland exposure indicators.
#### Local Land Cover and Cropland - FAO NSL Mozambique
**Provider:** FAO Land and Water Division (NSL)
**Dataset:** Mozambique: Maputo and Gaza provinces (2022 - 10 m - 14 classes)
**Link:** NSL Mozambique land cover
**Role in EVE:** Localized cropland exposure for Mozambique localization mode
The FAO NSL Mozambique land-cover dataset is a separate local land-cover source for Maputo and Gaza provinces. EVE uses this dataset only in localization mode for Mozambique-specific outputs. In that localized pipeline, cropland is extracted from the FAO NSL land-cover classes and then intersected with the flood extent to compute flooded-cropland exposure indicators for the covered provinces.
#### Population - WorldPop
**Provider:** WorldPop
**Dataset:** WorldPop population datasets
**Link:** WorldPop
**Role in EVE:** Population exposure
WorldPop provides open spatial demographic datasets. EVE uses WorldPop population data to estimate the number of people located within flooded areas after spatial overlay with the dekadal flood layer.
#### Livestock - FAO Gridded Livestock of the World
**Provider:** Food and Agriculture Organization of the United Nations (FAO)
**Dataset:** Gridded Livestock of the World 4 (GLW4), 1 km resolution
**Link:** FAO Gridded Livestock of the World
**Role in EVE:** Livestock exposure by species
GLW4 provides spatial livestock distribution information produced by FAO. EVE uses GLW4 to estimate livestock exposure within flooded areas for buffalo, cattle, chicken, goats, pigs, and sheep.
#### Administrative Boundaries - FAO GAUL
**Provider:** Food and Agriculture Organization of the United Nations (FAO)
**Dataset:** Global Administrative Unit Layers (GAUL) 2024, levels 0, 1, and 2
**Link:** FAO Global Administrative Unit Layers 2024
**Role in EVE:** Administrative aggregation, dashboard filtering, and map display
GAUL provides standardized administrative boundary layers maintained by FAO. EVE uses GAUL codes and names to aggregate pixel-level flood and exposure statistics to ADM2, ADM1, and ADM0 units and to support consistent joins with external datasets.
#### Crop Calendar Context - FAO Crop Calendar
**Provider:** Food and Agriculture Organization of the United Nations (FAO)
**Dataset:** FAO Crop Calendar
**Link:** FAO Crop Calendar
**Role in EVE:** Interpretation support for flooded cropland indicators
The FAO Crop Calendar provides crop-stage and seasonal timing context for multiple countries. EVE uses crop-calendar information in the dashboard where available to help users interpret whether flooded cropland overlaps with relevant planting, growing, or harvest periods.
### 3.2 Flood and Exposure Concepts
EVE summarizes flood exposure by administrative unit. The source flood product is used to identify flooded areas, and those areas are intersected with reference datasets to estimate exposed land, population, livestock by species, and other relevant indicators.
#### Analytical Framework and Processing Flow
The diagram below summarizes the analytical workflow from the NOAA VIIRS flood input to the final EVE public data products. It is intended as a conceptual overview; detailed processing rules are described in the steps that follow.
```mermaid
flowchart TD
A[NOAA VIIRS 5-day composite flood product] --> B[Reclassification to EVE working classes]
B --> C[Dekadal temporal aggregation into flood raster]
C --> E[Comparison with VISTA monthly seasonal frequency reference]
E --> F[Exceptional flood classes]
E --> G[Seasonal flood classes]
F --> H[Exposure overlay]
G --> H
H --> I[Area, cropland, population, and livestock exposure]
I --> J[ADM2 flood and exposure statistics]
J --> K[Aggregation to ADM1]
K --> L[Aggregation to ADM0]
J --> M[Final EVE ADM2 table]
K --> N[Final EVE ADM1 table]
L --> O[Final EVE ADM0 table]
M --> P[Dashboard, API, and CSV]
N --> P
O --> P
```
EVE flood exposure indicators are produced from the NOAA VIIRS 5-day composites through four processing steps:
1. **Reclassification** — each individual 5-day composite is reclassified from the NOAA class scheme into the EVE working classes.
2. **Temporal aggregation** — the reclassified 5-day composites that fall within a dekad are combined in time into a single 10-day dekadal raster.
3. **Comparison with the VISTA monthly seasonal frequency reference** — the dekadal raster is compared against the VISTA monthly flood-frequency reference, a centred three-dekad window, to separate flooding that is seasonally expected from exceptional flooding.
4. **Impact assessment** — the flood is intersected with area, cropland, population, and livestock datasets to compute total and exceptional exposure statistics.
Each step is described separately below.
#### Step 1 — Reclassification of the NOAA VIIRS 5-day composites
This step is applied independently to each NOAA VIIRS 5-day composite, before any temporal aggregation takes place.
The NOAA VIIRS flood product provides information related to the percentage of each pixel detected as water. The NOAA input classes relevant to the 5-day composites are:
| NOAA input class | Meaning |
| --- | --- |
| 1 | Fill value: bad data, solar eclipse, or data with solar zenith angles outside the processing range |
| 15 | Open water without water fraction retrieval |
| 16 | Clear-sky bare land |
| 17 | Clear-sky vegetation |
| 20 | Snow cover |
| 27 | River/lake ice cover |
| 30 | Cloud cover |
| 38 | Supra-snow/ice water, mixed ice-water, or ice in melting status |
| 50 | Shadow: cloud shadow or terrain shadow |
| 100 | Open normal water: river, lake, reservoir, or ocean |
| 101-200 | Water fractions of supra-vegetation bare land floodwater |
EVE reclassifies the NOAA source raster values into the following working classes:
| Source raster value | EVE raster value | Interpretation |
| --- | --- | --- |
| 1 | 0 | Unclassified / nodata |
| 15 | 11 | Permanent water |
| 16-17 | 10 | Land |
| 20 | 4 | Snow cover |
| 27 | 5 | River/lake ice cover |
| 38 | 6 | Supra-snow/ice water, mixed ice-water, or melting ice |
| 30 | 9 | Cloud |
| 50 | 0 | Unclassified / nodata |
| 99 | 11 | Permanent water |
| 100 | 11 | Permanent water |
| 101-149 | 1 | Flood low |
| 150-179 | 2 | Flood medium |
| 180-200 | 3 | Flood high |
In label terms, fill values and shadow are treated as unclassified or nodata; open water classes are treated as permanent water; clear-sky bare land and vegetation are treated as land; snow, ice, cloud, and mixed ice-water classes are retained as interpretation classes; and water-fraction flood classes are mapped to low, medium, or high flood intensity depending on the NOAA source range.
The public aggregated data focuses on the flood exposure indicators derived from these classes.
#### Step 2 — Temporal aggregation to the 10-day dekadal raster
Once each 5-day composite has been reclassified in Step 1, the reclassified composites belonging to a dekad are aggregated in time into a single dekadal raster. Because a dekad is a ten-day reporting period, each dekadal raster combines two or three reclassified 5-day composites.
For each dekad, EVE combines the relevant classified VIIRS 5-day composites to produce one final dekadal raster:
| Dekad | Classified 5-day composite dates used |
| --- | --- |
| Dekad 1 | Calendar composite dates 5 and 10 |
| Dekad 2 | Calendar composite dates 15 and 20 |
| Dekad 3, 28-day month | Calendar composite dates 25 and 28 |
| Dekad 3, 29-day month | Calendar composite dates 25 and 29 |
| Dekad 3, 30-day month | Calendar composite dates 25 and 30 |
| Dekad 3, 31-day month | Calendar composite dates 25, 30, and 31 |
Aggregation is pixel-wise across the two or three classified 5-day rasters for the dekad. Flood classes have the highest priority, and among flood classes the highest flood severity is retained. The priority order used for dekadal aggregation is:
`3 > 2 > 1 > 6 > 5 > 4 > 10 > 11 > 9 > 0`
This means that high flood overrides medium and low flood; medium flood overrides low flood; and, when no flood is present, interpretation classes are retained according to the same priority order. For example, if one raster has medium flood and another has low flood, the dekadal output keeps medium flood. If no flood is present and one raster has river/lake ice while another has snow, the output keeps river/lake ice. If only cloud and nodata are present, the output keeps cloud.
The final dekadal raster is written as a single-band `uint8` GeoTIFF with the embedded class colormap. At this point, the raster describes the highest flood extent and intensity recorded by the VIIRS sensors during the approximately ten observations available in the given dekad period. It does not yet indicate how anomalous or exceptional that flooding is for a specific area and time of year.
#### Step 3 — Comparison with the VISTA monthly seasonal frequency reference
The dekadal raster identifies where water is present, but not whether that water is normal for the location and time of year. In this step, EVE compares the dekadal observation against the VISTA monthly seasonal frequency reference to separate expected seasonal flooding from exceptional flooding.
EVE overlays the dekadal flood layer with the VISTA monthly flood-frequency reference, a historical frequency record that accounts for the number of valid historical observations available for each pixel. The reference is selected by the dekad being processed, but it summarises a centred three-dekad window rather than that dekad alone: for each historical year, a pixel counts as flooded if it was flooded in the preceding, central, or following dekad. This deliberately allows one dekad of tolerance on either side, because weather-driven flooding is not precise to a fixed ten-day window and the same seasonal flood can arrive slightly early or late from year to year.
A pixel is treated as seasonally expected flooding when the VISTA `freq_flood_any` frequency is at least 0.25, meaning the pixel was flooded in at least 25 percent of the valid historical years for that window. Flooding outside these areas is considered exceptional. Pixels with missing or no-data historical frequency values are treated as expected dry. This helps mitigate false positives in places with regular flood presence while accounting for spatial differences in historical observability, including cloud-related observation gaps. In seasonal areas, flood intensity values 1, 2, and 3 are encoded in the raster as 21, 22, and 23 so they can be separated from exceptional flood values during aggregation.
Because the reference window spans three dekads rather than one, the area treated as seasonally expected is larger than it would be under a dekad-exact reference at the same threshold. Total detected flood is unaffected by this step; only the division between exceptional and seasonal flood changes, with more flood classified as seasonal and correspondingly less as exceptional.
Internally, EVE groups the selected flood classes into two analytical categories:
| Internal category | Source raster classes | User-facing meaning |
| --- | --- | --- |
| 200 | 1, 2, and 3 | Exceptional flood (low, medium, high) |
| 220 | 21, 22, and 23 | Seasonal flood (low, medium, high) |
The final tables report three user-facing values for each metric — "exceptional", "seasonal", and "total" — where the total is the sum of the exceptional and seasonal categories.
#### Step 4 — Impact assessment: exposed area, cropland, population, and livestock
In this final step, the flooded areas identified in the previous steps are intersected with exposure reference datasets to estimate the assets potentially affected by the flood. For each impact category — flooded area, cropland, population, and livestock by species — EVE produces statistics for both total water detected and exceptional flood only:
- **Total water detected** — the assets falling within all detected flood, combining exceptional and seasonal flood.
- **Exceptional only** — the assets falling within exceptional flood, excluding water that is seasonally expected.
The seasonal share is the difference between the two, so each metric has exceptional, seasonal, and total values. The exceptional statistic isolates the flood signal most relevant to impact, while the total statistic captures the full extent of detected flood water.
Cropland exposure is based on ESA WorldCover cropland classes in the standard workflow for all countries and modes. Mozambique localization-mode outputs use a separate FAO NSL land-cover pipeline for Maputo and Gaza provinces. In both pipelines, cropland is extracted from the relevant land-cover source and intersected with the flood extent, but the source datasets, processing pipelines, and output modes are separate.
Flooded cropland should not be interpreted directly as crop loss. The standard cropland indicator identifies cropland area intersecting with flood classes, but it does not by itself determine whether the land was actively cultivated, already harvested, fallow, or at a crop stage sensitive to flooding. To support interpretation, the EVE dashboard also displays EVE information together with crop calendar information where available. Crop calendar information is available for multiple countries and helps users understand crop stage by period and agro-ecological zone. Reference crop calendar information is available from FAO at https://cropcalendar.apps.fao.org/#/home.
The final data should be interpreted as aggregated flood exposure information based on these selected categories. It is not a field-validated damage assessment, a direct measurement of economic loss, or a legal determination of affected territory.
### 3.3 Aggregation to Administrative Levels
The flood and exposure statistics produced by Step 4 are computed at the level of the second administrative division (ADM2) and are then aggregated upward to first-level units (ADM1) and countries (ADM0).
For every reporting dekad, EVE produces one record per administrative unit at each of the three levels:
- **ADM2** — the base unit, where flood and exposure statistics are computed directly.
- **ADM1** — first-level subnational units, obtained by summing the values of their constituent ADM2 units.
- **ADM0** — countries, obtained by summing their constituent ADM1 units.
Administrative units are linked through FAO GAUL codes: each ADM2 belongs to exactly one ADM1, and each ADM1 to exactly one ADM0. Absolute indicators — flooded area, cropland, population, and livestock, in their exceptional, seasonal, and total versions — are summed, so the figures remain consistent across the three levels. Percentage indicators are not summed; they are recomputed at each level from the aggregated values, as described in Section 3.4.
The result is the three final EVE tables — ADM0, ADM1, and ADM2 — published for each reporting dekad.
### 3.4 Percentages and Denominators
Percentage indicators express the relationship between a flooded or exposed value and the relevant denominator for the same administrative unit and theme. For example, a cropland percentage should be interpreted in relation to the total cropland denominator for that administrative unit, not the total area of the country unless explicitly stated.
EVE uses static ADM2 denominators for area, cropland, population, and livestock. In the standard workflow, these denominators include total ADM2 area, total cropland area from ESA WorldCover class 40, total population from WorldPop 2020, and total livestock counts by species from GLW4 2020. ESA WorldCover is the cropland source for all published indicators and denominators.
ADM1 and ADM0 percentages are computed from aggregated numerator and denominator values. They should not be interpreted as simple averages of ADM2 percentages.
Area-based metrics are computed using pixel area in a deterministic working coordinate reference system. Cropland is handled fractionally, so partial cropland coverage within an analysis pixel can contribute proportionally. Population and livestock counts are computed on their source datasets' native grids. Livestock exposure is also fractional, based on the overlap between flooded areas and GLW4 grid cells.
## 4. Data Structure
### 4.1 Product Inventory
The essential public EVE final products are:
| Product | Administrative level | Format | Access | Intended use |
| --- | --- | --- | --- | --- |
| EVE ADM0 final table | ADM0 | Table / CSV / API | Dashboard, CSV, and ADM0 API/dataset access | Country-level monitoring and comparison |
| EVE ADM1 final table | ADM1 | Table / CSV / API | Dashboard, CSV, and ADM1 API/dataset access | Subnational monitoring and prioritization |
| EVE ADM2 final table | ADM2 | Table / CSV / API | Dashboard, CSV, and ADM2 API/dataset access | Detailed administrative analysis |
| EVE raster and imagery products | Raster grid | Raster / imagery | Visualized on the EVE dashboard; available for download only upon request through the FAO DIEM contact page | Visual inspection, technical review, and advanced geospatial analysis |
Aggregated information up to ADM2 level is publicly downloadable through the dashboard, API, and CSV exports. EVE raster products are available for visualization on the EVE dashboard. Because of their size, they are not available as standard downloads, but users may request access by contacting the FAO DIEM team through the FAO DIEM contact page.
### 4.2 Field Dictionary
The three public tables share the same fields. They differ only in the administrative identifiers each level carries. Each row is one administrative unit for one dekad: it is identified by the table's lowest-level GAUL code (`gaul0_code`, `gaul1_code` or `gaul2_code`) together with `dekad_year`, `dekad_month` and `dekad_number`.
**Identifiers and dates**
| Field | Levels | Type | Description |
| --- | --- | --- | --- |
| `gaul0_code` | ADM0, ADM1, ADM2 | Integer | FAO GAUL country code |
| `gaul0_name` | ADM0, ADM1, ADM2 | Text | Country name (GAUL) |
| `iso3_code` | ADM0, ADM1, ADM2 | Text | ISO 3166-1 alpha-3 country code |
| `gaul1_code` | ADM1, ADM2 | Integer | FAO GAUL first-level unit code |
| `gaul1_name` | ADM1, ADM2 | Text | First-level unit name (GAUL) |
| `gaul2_code` | ADM2 | Integer | FAO GAUL second-level unit code |
| `gaul2_name` | ADM2 | Text | Second-level unit name (GAUL) |
| `dekad_year` | All | Integer | Year of the dekad |
| `dekad_month` | All | Integer | Month of the dekad (1–12) |
| `dekad_number` | All | Integer | Dekad within the month: 1 = days 1–10, 2 = days 11–20, 3 = day 21 to month end |
| `dekad_start` | All | Date | First day of the dekad (YYYY-MM-DD) |
| `dekad_end` | All | Date | Last day of the dekad (YYYY-MM-DD) |
**Flood and exposure indicators**
Each indicator is published in three versions, which differ only in the suffix at the end of the field name:
| Suffix | Meaning |
| --- | --- |
| `exceptional` | Flood where VISTA shows flooding is not normally expected at this time of year (historical flood frequency below 25%) |
| `seasonal` | Flood where VISTA shows flooding is normally expected at this time of year (historical flood frequency of 25% or more) |
| `total` | Exceptional + seasonal |
| Field (add a suffix) | Unit | Description | Percentage denominator (per administrative unit) |
| --- | --- | --- | --- |
| `flooded_area_sqkm_` | km² | Flooded area | — |
| `flooded_area_perc_` | % | Flooded area as a share of the unit | Total area |
| `flooded_cropland_sqkm_` | km² | Flooded cropland | — |
| `flooded_cropland_perc_` | % | Share of the unit's cropland that is flooded | Cropland area (ESA WorldCover 2021) |
| `flooded_pop_n_` | people | Estimated population in flooded areas | — |
| `flooded_pop_perc_` | % | Share of the population in flooded areas | Population (WorldPop 2020) |
| `flooded__n_` | head | Estimated livestock in flooded areas | — |
| `flooded__perc_` | % | Share of the livestock in flooded areas | Livestock by species (GLW4 2020) |
`` is one of `buffalo`, `cattle`, `chicken`, `goats`, `pigs`, `sheep`. Example: `flooded_goats_perc_seasonal`.
**Observation fields**
| Field | Unit | Description |
| --- | --- | --- |
| `cloud_area_sqkm` | km² | Area obscured by cloud in the dekad, where flooding could not be observed. Not split by flood category. |
| `cloud_area_perc` | % | Cloud area as a share of the unit's total area |
`OBJECTID` (ADM0, ADM1) and `ObjectId` (ADM2) are row identifiers added by the hosting platform. They are not indicators and are not stable across republications.
### 4.3 Reading the Values
- **Rounding.** Indicator and cloud values are decimal numbers, rounded to two decimals by default. Totals are calculated before rounding, so the displayed exceptional and seasonal values may differ from the displayed total by 0.01.
- **Zero values.** In flood and exposure indicators, a value of 0 usually means no flooding was detected. It is also used when an input or denominator is unavailable, so a 0 is not proof that no flooding occurred. Check the cloud fields: a unit mostly under cloud may show 0 flood.
- **Decimal counts.** Population and livestock are spatial estimates, so counts can be decimals.
- **ADM1 and ADM0 percentages.** These are calculated from the summed ADM2 values and summed ADM2 denominators of the units included. They are not averages of ADM2 percentages.
- **Denominators.** The denominators are not published as fields.
## 5. How to Use the Data
### 5.1 Basic User Workflow
A typical user workflow is:
1. Open the official EVE dashboard.
2. Select the reporting period of interest.
3. Review the list of countries available for that reporting period.
4. Select the country or administrative unit.
5. Review the relevant flood and exposure indicators.
6. Compare absolute values and percentages together.
7. Check observation or cloud-related fields where available.
8. Download the required ADM0, ADM1, or ADM2 table as CSV, or access the same information through the API.
9. Cite EVE, FAO, the access date, and the reporting period in any derived analysis or product.
### 5.2 Filtering, Joining, and Downloading
Users should filter by reporting period, country, administrative level, and indicator type. When joining EVE data to external datasets, users should use stable FAO GAUL administrative codes rather than names wherever possible. Names may vary across languages, spellings, or boundary versions.
Downloaded extracts should retain the original field names, reporting period, access date, and source URL. Because historical records are treated as immutable once published, users should also retain the extraction date used for analysis and reporting. Users should avoid mixing different administrative boundary versions unless they have documented and validated the join logic.
### 5.3 Interpretation Guidance
Each row in an EVE final table represents one administrative unit for one reporting period. Administrative fields identify the place, temporal fields identify the dekad, and indicator fields describe flood and exposure values for that unit and period. The final public tables are simplified wide tables, with exceptional, seasonal, and total versions of the main metrics.
When comparing places, users should consider both absolute values and percentages. A large administrative unit may have a high flooded area but a low percentage of affected cropland, while a smaller unit may show the opposite pattern. When comparing periods, users should account for differences in cloud cover, observation availability, input data updates, and boundary or denominator changes.
Users should remember that standard aggregated exposure indicators include low, medium, and high flood. Cloud area and cloud percentage should be used to understand observation constraints and should not be interpreted as flooded area. Cloud percentage is calculated against the total administrative unit area.
Users should monitor change over time rather than relying only on a single static value. Trends across dekads can help distinguish meaningful increases or decreases in flood exposure from persistent detections in dynamic water environments.
Users should not interpret EVE indicators as direct damage, confirmed field impact, household-level exposure, or real-time warning information. EVE is best used as an operational screening and monitoring product that can guide further analysis, prioritization, and contextual review.
## 6. Quality, Limitations, and Responsible Use
### 6.1 Known Limitations and Uncertainty
Users should consider the following limitations:
- Satellite flood detection may miss or misclassify some flooded areas because of clouds, vegetation, mixed pixels, surface conditions, sensor characteristics, or temporal coverage.
- VIIRS is based on optical daytime observations and cannot observe through cloud cover. Persistent cloud, especially in tropical, monsoon-prone, or storm-affected areas, can lead to underestimation of flood extent.
- Short-duration floods, including flash floods, may be missed if they occur between observations or recede before a usable composite captures them.
- Urban floods may be difficult to detect because buildings, infrastructure, roads, drainage systems, and mixed urban surfaces can obscure or confuse the satellite signal.
- Some environments are prone to false positives, including coastal zones, marshlands, riverbanks, lakeshores, regularly flooded rice fields, tidal areas, dynamic water bodies, and wet or muddy soil. EVE mitigates part of this issue by using the VISTA monthly flood-frequency reference to distinguish expected seasonal flooding from exceptional flooding.
- Aggregated administrative summaries may hide important variation within administrative units.
- Exposure indicators depend on the quality, resolution, date, and assumptions of the reference datasets.
- Livestock indicators are based on modelled livestock distribution data and do not capture seasonal livestock movements or temporary herd relocation.
- Flooded cropland indicators do not indicate crop loss by themselves. They must be interpreted with crop calendars, crop stage, agro-ecological zone, planting and harvest timing, and local agricultural information where available.
- Percentages depend on denominators and should not be compared without understanding the underlying population, cropland, livestock, or area base.
- Administrative boundaries are used for statistical aggregation and may not represent legal or political positions.
- Raster and imagery products are large and are not part of the default public download package.
- Public dashboard, API, or CSV availability may depend on operational publication schedules and service availability.
### 6.2 Responsible Use and Disclaimers
EVE final data should be used as flood exposure and situational awareness information. It should be combined with local knowledge, field reports, national data, and other authoritative sources when used for decisions with operational, financial, legal, or public communication consequences.
Users should not use EVE data to make unsupported claims about direct damage, confirmed losses, individual people or households, legal boundary status, or final impact attribution. Any public use should include appropriate citation and acknowledge the main limitations of satellite-derived, aggregated exposure data.
EVE final data is released under the Creative Commons Attribution 4.0 International (CC BY 4.0) license, unless otherwise specified. Under this license, users may share, copy, redistribute, adapt, remix, transform, and build upon the material for any purpose, including commercial use.
Users must provide appropriate credit to FAO, provide a link to the license, and indicate if changes were made. Attribution must not suggest that FAO endorses the user or the user's use of the data.
© FAO. All rights reserved. FAO encourages the use, reproduction, and dissemination of material in this information product. For permissions beyond the scope of the CC BY 4.0 license, contact copyright@fao.org.
**Mandatory boundary disclaimer:**
The boundaries and names shown and the designations used on these map(s) do not imply the expression of any opinion whatsoever on the part of FAO concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers and boundaries. Dashed lines on maps represent approximate border lines for which there may not yet be full agreement.
## 7. Support Materials
### 7.1 Glossary and Acronyms
Essential terms to define:
| Term | Definition |
| --- | --- |
| ADM0 | Country-level administrative unit |
| ADM1 | First-level subnational administrative unit |
| ADM2 | Second-level subnational administrative unit |
| API | Application Programming Interface |
| CSV | Comma-separated values file |
| Dekad | Reporting period normally covering ten days; each month has three dekads, and the last dekad of the month may cover 8, 9, 10, or 11 days depending on month length and leap years |
| DIEM | FAO Data in Emergencies programme |
| EVE | Events Visualization in Emergencies |
| GAUL | Global Administrative Unit Layers |
| GLW4 | Gridded Livestock of the World 4 |
| NSL | FAO Land and Water Division |
| OER | FAO Office of Emergencies and Resilience |
| VIIRS | Visible Infrared Imaging Radiometer Suite |
### 7.2 Contact and Further Information
For questions about data sources, methodology, or issue reporting, contact FAO Data in Emergencies (DIEM) through the FAO DIEM contact page.