Skip to content

GDACS

GDACS is a cooperation framework between the United Nations, the European Commission and disaster managers worldwide to improve alerts, information exchange and coordination in the first phase after major sudden-onset disasters.

Collection: gdacs-events

A STAC collection hold all the GDACS events. An example of the GDACS collection is here.

Data

Accessible data is a set of GDACS entries. Each entry is a disaster event. The event data list is available in the form of a geojson collections via the API endpoint https://www.gdacs.org/gdacsapi/api/events/geteventlist/SEARCH?. Individual events can be accessed via the API endpoint https://www.gdacs.org/gdacsapi/api/events/geteventdata?eventtype=FL&eventid=1102983.

Important

It is important to note that GDACS has its own specific models according to the type of event. This must be taken into account when mapping the data to the STAC model. When necessary, the present document will provide the specific mapping for each type of event.

Event Item

A GDACS event and episode will ALWAYS produce an event STAC item as in the example for the flood in Spain from 27 Oct 2024 04 Nov 2024.

The event URL of the geteventdata API endpoint is stored in the links field of the STAC item with the via relation.

Here is a table with the fields that are mapped from the GDACS event to the STAC event:

STAC field GDACS field Description
id properties.eventid + properties.episodeid Unique identifier for the event per episode
bbox bbox Bounding box of the event
geometry geometry Geometry of the event
collection gdcas-events The collection for GDACS events
title properties.name Name of the event
description properties.description
properties.htmldescription		 Description of the event. HTML description should be privileged over plain text description and translated to markdown
datetime properties.fromdate Date and time of the event converted in UTC ISO 8601 format
start_datetime properties.fromdate Start date of the event converted in UTC ISO 8601 format
end_datetime properties.todate End date of the event converted in UTC ISO 8601 format
monty:country_codes[0] properties.iso3 ISO3 code of the country where the event occurred. Keywords shall also contain the human readable country name
monty:country_codes[1..*] properties.affectedcountries.iso3 List of ISO3 codes of the other countries affected by the event
monty:hazard_codes properties.eventtype List of hazard codes converted following the GDACS event type to Hazard profile mapping
assets.icon properties.icon Asset with the icon of the event
asset.report properties.url.report Asset with the link to the GDACS report
via link in [links] properties.url.details Link to the GDACS event details page
related link in [links] properties.source and<br>properties.sourceid If the source is present, create a related link to the item in the corresponding collection (e.g. GLOFAS-> glofas-events) with roles: ["event"]
related link in [links] properties.glide If the glide number is present, create a related link to the item in glide-events collection with roles: ["event"]
monty:src_event_id Source event ID Unique identifier of the event
monty:corr_id Generated Generated following the event correlation convention

Hazard Item

A GDACS event and episode will ALWAYS produce one hazard STAC item as in the example for the flood in Spain from 27 Oct 2024 04 Nov 2024.

Here is a table with the STAC fields that are mapped from the GDACS event to the STAC hazard:

STAC field GDACS field Description
id properties.eventid + properties.episodeid Unique identifier for the hazard per episode
bbox bbox Bounding box of the hazard
geometry geometry Geometry of the hazard
collection gdcas-hazards The collection for GDACS hazards
title properties.name Name of the hazard
description properties.description
properties.htmldescription		 Description of the hazard. HTML description should be privileged over plain text description and translated to markdown
datetime properties.fromdate Date and time of the hazard converted in UTC ISO 8601 format
start_datetime properties.fromdate Start date of the hazard converted in UTC ISO 8601 format
end_datetime properties.todate End date of the hazard converted in UTC ISO 8601 format
monty:country_codes[0] properties.iso3 ISO3 code of the country where the hazard occurred. Keywords shall also contain the human readable country name
monty:country_codes[1..*] properties.affectedcountries.iso3 List of ISO3 codes of the other countries affected by the hazard
monty:hazard_codes mappings from properties.eventtype List of hazard codes converted following the GDACS hazard type to Hazard profile mapping
assets.icon properties.icon Asset with the icon of the hazard
asset.report properties.url.report Asset with the link to the GDACS report
via link in [links] properties.url.details Link to the GDACS hazard details page
monty:hazard_detail properties.hazard_detail Detailed description of the hazard (more details in next section)
monty:src_event_id Source event ID Unique identifier of the event

Hazard Detail

The hazard_detail field is a JSON object that contains the detailed information about the hazard. The object is a mapping of the hazard codes to the detailed information. The detailed information is a JSON object with the following fields:

STAC field GDACS field Description
severity_value properties.episodealertlevel GDACS alert score
severity_unit gdacs GDACS alert level according to GDCAS event type and model
Mapping from GDACS event type to Hazard profile

GDACS event types map to multiple classification systems for cross-system interoperability. The 2025 UNDRR-ISC code is the reference classification for the Monty extension:

GDACS Type GLIDE EM-DAT UNDRR-ISC 2025 (Reference) Cluster Description
FL FL nat-hyd-flo-flo MH0600 MH-WATER Flooding (chapeau)
EQ EQ nat-geo-ear-gro GH0101 GEO-SEIS Earthquake
TC TC nat-met-sto-tro MH0306 MH-WIND Cyclone or Depression
TS TS nat-geo-ear-tsu MH0705 MH-MARINE Tsunami
VO VO nat-geo-vol-vol GH0201 GEO-VOLC Lava Flows
DR DR nat-cli-dro-dro MH0401 MH-PRECIP Drought

Note

All three classification codes (GLIDE, EM-DAT, UNDRR-ISC 2025) should be included in the monty:hazard_codes array for maximum interoperability. More specific hazard codes can be added following the characteristics of the event.

Hazard Magnitude and Units

In GDACS, the alert level is a score that is calculated based on the event type. Each event ype uses a specific model to calculate the alert level. The alert level is a score that is used to determine the magnitude of the event. The following table shows the magnitude scale and unit to be used for each event type:

GDACS event type Magnitude scale Magnitude unit
Flood [FL] event type uses a severity score based on the Global Flood Detection System 1-3 GDACS Flood Severity Score

Impact Item

According to the event type and the fields available in the GDACS event, one or more impact STAC items can be created. The following sections describe the mapping of specific GDACS event information to the STAC impact item.

GDACS carries impact data in two different places. The mapping to use depends on the event type:

GDACS event type Source field Section
FL properties.sendai Sendai indicators
TC properties.impacts[].resource.timeline Tropical cyclone advisory timeline
WF properties.impacts[].resource.impact Wildfire population exposure
EQ, DR, TS, VO none No impact item is produced. See Event types without an impact mapping

Important

The impacts[] figures are exposure estimates, not observed losses. GDACS computes them with a model. It counts the population inside a hazard footprint. It does not count the persons that the hazard harmed. All impact items built from impacts[] therefore use type: potentially_affected and estimate_type: modelled. See Decision: exposure is not impact.

Sendai indicators

When the sendai field is present in the GDACS event, it contains an array of Sendai indicators. Each Sendai indicator is a JSON object that shall produce an impact item. The impact item shall have the following fields from both the GDACS event and the Sendai indicator:

STAC field GDACS field Description
id properties.eventid + properties.episodeid + properties.sendai.sendaitype + properties.sendai.sendainame + properties.sendai.country + properties.sendai.region Unique identifier for the impact
bbox bbox of the geocoding of the properties.sendai.country + properties.sendai.region Bounding box of the impact
geometry geometry of the geocoding of the properties.sendai.country + properties.sendai.region Geometry of the impact
collection gdcas-impacts The collection for GDACS impacts
title properties.name + properties.sendai.sendaitype + properties.sendai.sendainame + properties.sendai.country + properties.sendai.region Name of the impact
description properties.sendai.description Description of the impact. HTML description should be privileged over plain text description and translated to markdown
datetime properties.sendai.onset_date Date and time of the event converted in UTC ISO 8601 format
start_datetime properties.sendai.onset_date Start date of the event converted in UTC ISO 8601 format
end_datetime properties.sendai.expires_date End date of the event converted in UTC ISO 8601 format
monty:country_codes[0] properties.iso3 ISO3 code of the country where the event occurred. Keywords shall also contain the human readable country name
monty:country_codes[1..*] properties.affectedcountries.iso3 List of ISO3 codes of the other countries affected by the event
monty:hazard_codes mappings from properties.eventtype List of hazard codes converted following the GDACS event type to Hazard profile mapping
monty:src_event_id Source event ID Unique identifier of the event
assets.icon properties.icon Asset with the icon of the event
asset.report properties.url.report Asset with the link to the GDACS report
via link in [links] properties.url.details Link to the GDACS event details page
related link in [links] properties.source and<br>properties.sourceid If the source is present, create a related link to the item in the corresponding collection (e.g. GLOFAS-> glofas-events) with roles: ["event"]
monty:corr_id Generated Generated following the event correlation convention

The impacts field

For tropical cyclones and wildfires, GDACS does not put the impact figures in the episode payload. It puts links to them in properties.impacts[]. Each entry has a source field that names the advisory agency, and a resource object of URLs:

"impacts": [
  {
    "source": "JTWC",
    "resource": {
      "buffer39": "https://www.gdacs.org/gdacsapi/api/export/getimpact?id=769814",
      "buffer74": "https://www.gdacs.org/gdacsapi/api/export/getimpact?id=769813",
      "timeline": "https://www.gdacs.org/gdacsapi/api/export/gettimeline?id=769812",
      "locations": "https://www.gdacs.org/gdacsapi/api/export/getlocations?id=769738"
    }
  }
]

The set of resource keys depends on the event type:

Resource key Endpoint Event type Content Mapped
timeline gettimeline TC One entry for each advisory point. Track position, wind speed and population exposure. Yes
buffer39 getimpact TC Total exposure in the 39 kt wind buffer, with exposed countries and infrastructure. No
buffer74 getimpact TC The same for the 74 kt wind buffer. No
locations getlocations TC Named places along the track. No
impact getimpact WF Population exposure for the burnt area. Yes

Note

Read the impacts[] field from getepisodedata, not from geteventdata. geteventdata ignores the episodeid parameter and always returns the current episode. GDACS publishes the correct per-episode URL in properties.episodes[].details.

Tropical cyclone advisory timeline

The timeline resource returns channel.item[]. Each entry describes one point on the storm track, as issued by the advisory agency (for example JTWC).

GDACS field Type Meaning
id string Identifier of the timeline entry. It is unique, and stable across episodes.
advisory_number string Advisory sequence number from the advisory agency.
actual string "True" for an observed position. "False" for a forecast position.
current string "true" on the entry that the current advisory observed.
advisory_datetime string Validity time of the entry, in UTC, format %d %b %Y %H:%M.
coordinates string "<longitude> , <latitude>" of the track point.
wind_speed string Maximum sustained wind, in m/s.
pop39 string Population in the wind field of 39 kt or more.
pop74 string Population in the wind field of 74 kt or more.
popstormsurge string Population in the storm surge zone.
pop string Meaning not confirmed. Do not map this field. See rule 3 below.
alertscore string GDACS alert score at this track point.

Reference files, trimmed to the mapped fields:

Three properties of the timeline that control the mapping

1. The timeline is cumulative, and it repeats. Each episode returns every earlier advisory entry again, unchanged. The 9 observed entries of episode 9 are identical to the same 9 entries of episode 13, and they keep the same id. A transformer that keys the impact item on the episode therefore creates one duplicate for each later episode.

2. advisory_number is not unique. The current advisory appears once as an observed position, then once more for each forecast lead time. In episode 13 below, advisory 13 appears 4 times. Advisory 12 is absent. advisory_number is therefore neither unique nor dense. Only id is a safe key.

3. pop is not a total. The table below is the timeline of episode 13. In advisory 9, pop is 0 while pop39 counts 75 million people. In the last forecast entry, pop counts 19 million while pop39 is 0. pop is therefore not the sum of the wind bands, and it is not a headline figure. Its definition is not published. Until JRC confirms it, use pop39 and pop74, which the GDACS tropical cyclone model defines.

advisory actual current advisory_datetime wind_speed pop pop39 pop74
8 True false 25 Jul 2026 00:00 36.008 0 18,897,670 0
9 True false 25 Jul 2026 06:00 38.58 0 75,490,475 614,941
10 True false 25 Jul 2026 12:00 41.152 0 94,850,067 6,094,547
11 True false 25 Jul 2026 18:00 43.724 12,226,693 103,014,755 11,167,854
13 True true 26 Jul 2026 00:00 38.58 38,543,219 80,042,232 10,640,921
13 False false 26 Jul 2026 12:00 23.148 7,488,955 14,518,892 0
13 False false 27 Jul 2026 00:00 15.432 8,596,681 0 0
13 False false 27 Jul 2026 12:00 10.288 19,811,513 0 0
Mapping

Each timeline entry shall produce one impact item for each non-zero exposure band. An entry with pop39 and pop74 both above zero therefore produces 2 items. An entry with every band at zero produces none.

STAC field GDACS field Description
id gdacs-impact- + properties.eventid + - + item.id + - + band Band is pop39, pop74 or surge. The episode is not part of the id. See rule 1
geometry Point built from item.coordinates Position of the track point
collection gdacs-impacts The collection for GDACS impacts
datetime item.advisory_datetime Validity time of the advisory point, converted to UTC ISO 8601
start_datetime item.advisory_datetime The same value. The entry describes an instant, not a period
end_datetime item.advisory_datetime The same value
title item.name + advisory number + band For example NOUL-26 advisory 13, population in 39 kt wind field
monty:country_codes properties.iso3 and properties.affectedcountries.iso3 Taken from the episode. The timeline entry has no reliable country field
monty:hazard_codes mappings from properties.eventtype TC maps to MH0306
monty:episode_number properties.episodeid The episode being processed. See rule 5
monty:src_event_id properties.eventid Unique identifier of the event
monty:corr_id Generated Generated following the event correlation convention
via link in [links] the timeline URL of the episode The URL changes with each episode. Consumers shall not use it as identity

The monty:impact_detail object:

Field Value
category people
type potentially_affected
value pop39, pop74 or popstormsurge, as an integer, for the band of the item
unit count
estimate_type modelled
description The band, in words. For example Population in the wind field of 39 kt or more

Rules:

  1. Key the item on item.id, not on the episode. The identifier is stable across episodes. Re-ingestion of a later episode then updates the same item, instead of creating a duplicate.
  2. Ingest every entry, of every episode. Forecast entries of an earlier episode carry their own id, so they stay as a record of what GDACS forecast at the time. Observed entries collapse onto the item that an earlier episode already created.
  3. estimate_type is always modelled. The actual field describes the track position, not the exposure figure. GDACS models the exposure for observed positions in the same way as for forecast positions.
  4. type is always potentially_affected. See Decision: exposure is not impact.
  5. monty:episode_number is the episode being processed. Because observed entries repeat, this field can change on re-ingestion. It carries the last episode that published the entry, not the first.

Warning

Monty has no field that marks an estimate as a forecast. estimate_type describes where a figure comes from. It does not say whether the figure is about the future. A cyclone forecast and a post-event model run both map to modelled. A consumer can therefore tell a forecast entry from an observed entry only by its datetime. See Open questions.

Wildfire population exposure

For wildfires, properties.impacts[].resource.impact returns a getimpact document. The source is GWIS. The document has this shape:

GDACS field Meaning
modelname WF
modelrun Timestamp of the model run
modelstatus Status of the model run, for example info: processing ok
datums[].datum[] with datasource: "POP" The population scalars
datums[].datum[] with datasource: "country" Exposed countries, with ISO_3DIGIT
datums[].datum[] with datasource: "INPUT PARAMETERS" The burnt area, as a WKT MULTIPOLYGON in Shape

The POP datum carries a series of scalars:

Scalar Meaning
POPAFFECTED Population in the burnt area. Equal to SUMPOP0.0
SUMPOP1.0 Population within 1 km of the burnt area
SUMPOP2.0 Population within 2 km
SUMPOP5.0 Population within 5 km
SUMPOP10.0 Population within 10 km

Each getimpact document shall produce one impact item from POPAFFECTED.

STAC field GDACS field Description
id gdacs-impact- + properties.eventid + - + properties.episodeid + -popaffected The impact document has no identifier of its own
geometry the Shape WKT of the INPUT PARAMETERS datum The burnt area. Fall back to the event geometry if Shape is absent
datetime properties.fromdate of the episode The document has no observation time, only a model run time
collection gdacs-impacts The collection for GDACS impacts

The monty:impact_detail object:

Field Value
category people
type potentially_affected
value POPAFFECTED, as an integer
unit count
estimate_type modelled
description Population in the burnt area

The modelname, modelrun and modelstatus fields show that GDACS derived this figure from a model. estimate_type is therefore modelled, and never primary.

Event types without an impact mapping

GDACS publishes no usable impact figures for EQ, DR, TS and VO. These event types produce an event item and a hazard item only. For EQ, the alert score and the population exposure appear in the event description text, not in a structured field. A mapping from free text is not reliable enough to specify here.

Decision: exposure is not impact

pop39, pop74 and POPAFFECTED count the persons inside a hazard footprint. They do not count the persons that the hazard harmed. UNDRR keeps these two concepts apart: exposure is the presence of persons or assets in hazard-prone areas, and impact is the effect of the hazard on them.

Monty has no exposure class. The nearest impact type is potentially_affected, and this document uses it for every figure that comes from impacts[]. The choice is deliberate, and it is a compromise. A consumer that sums affected_total across sources does not pick up these figures, which is the intended behaviour. A consumer that wants the exposure figures shall filter on type = potentially_affected.

Open questions raised by this mapping

Two questions in this mapping belong to the Monty model, not to GDACS. They are open for the Montandon Technical Working Group:

  1. Should Monty mark an estimate as a forecast? estimate_type describes provenance. It cannot say whether an estimate is about the future.
  2. Should exposure be a class of its own? potentially_affected carries the exposure figures today, and it also carries other meanings for other sources.

Until the group settles these questions, the rules above hold.