Methodology & evidence limitations

What the map and regional summaries mean

ENSO is a climate driver. Seasonal rainfall outlooks, observed anomalies, reported impacts and suggested actions are different evidence types. The country panel separates climate forecasts, monitoring and reported impacts, each with its own dates. Legacy mixed local reports remain explicitly labelled until their claims are classified. Changing the map layer does not switch or relabel country evidence. Country marker colours summarize the selected information; they are not a local disaster-risk score. A regional summary draws together available country and regional evidence, retains important differences and does not assign one hazard level to every selected place.

Observed anomalies and time

The map background and the observed anomaly timeline share one selection. Rainfall uses NOAA CPC CAMS_OPI monthly anomalies in mm/day against 1979–1995, on a 2.5° grid. Sea-surface temperature uses the NOAA Coral Reef Watch 1991–2020 anomaly product, in °C.

The SST catalog currently contains 10 daily observations from 2026-07-18 to 2026-09-12, mostly seven days apart. These are individual daily products, not weekly means or a continuous daily series. Playback advances through downloaded observations only. Missing dates are not interpolated. Country reports, forecasts and historical episodes keep their own dates when the background date changes.

SST native 0.05° water pixels are averaged into 0.25° display cells only where at least 13 of 25 pixels are valid. Land, ice and missing pixels are excluded; display values are rounded to 0.01°C. This is display precision, not an accuracy estimate. Ocean temperature is not air temperature, an ENSO intensity index, a coral-bleaching alert or an impact measure. A coarse cell containing an island and surrounding ocean is not a local station measurement.

Download SST dates, source links and processing metadata. Every frame retains the official validity interval, issue time, source URL and checksum. A date available in one layer does not establish data coverage in another.

Trace findings and assess confidence

Select a numbered citation to inspect its source, publisher, date and available page or section. Indexed text may be a portal summary; it is not a verbatim quotation unless stated. An exact passage or screenshot is shown only when available. Source reliability describes the source; analytical confidence describes how well the evidence supports the interpretation. An authoritative source can still leave local impacts uncertain.

Subnational boundaries show geography, not automatically a mapped impact. Local evidence is mapped only where its location can be supported. The Fiji pilot uses published 3-, 6- and 12-month rainfall categories through August 2026. Island reference points locate source accounts; they are not weather stations or affected-area boundaries. Unmapped evidence and missing observations are coverage gaps, not evidence of no impact.

Structural context and experimental models

Published INFORM vulnerability and coping-capacity values remain structural context. They do not measure current water shortage or forecast losses. Experimental water scenarios, INFORM component adjustments and the interactive cyclone comparison have moved out of the public portal for development and review.

Historical losses, people and experience

Accounts of the same event may use different indicators, geographic scopes and estimates. They are not automatically summed. Temporal overlap with an El Niño episode does not demonstrate causation. Source editions, original record identifiers and transformation notes are retained in the versioned shared-data export. EM-DAT, DesInventar and other coverage depends on access, matching and redistribution permissions; listing a source does not mean its whole database is included.

People & inclusion selections match evidence for each group separately. Several selections do not prove impacts at their intersection. Suggested actions distinguish documented responses, reported outcomes and analytical suggestions. A response being recorded does not demonstrate that it was effective; experience transferred between countries requires local assessment.

Institutions and reported readiness

Institutions cites the Pacific 2026/27 El Niño Readiness Stocktake by response, questionnaire field and version. Organizational affiliation and scope are respondent-reported and do not establish official representation. Assessments are labelled as respondent assessments; reported activities remain self-reports. Expertise and supporting documents are not automatically verified. Publication review permits display and is separate from source reliability and confidence in an analytical conclusion. Aggregate counts describe received responses in the selected scope, not representative country or regional readiness. Missing responses are gaps in coverage.

Talanoa and contributed material

Tools → Talanoa is the shared entry for feedback, corrections, evidence and experience. Feedback and correction requests are private. Documents, datasets, photographs, URLs and experience accounts enter review. Automatic extraction suggests editable fields and available source passages; it does not verify the document.

Approval for repository display is separate from approval for analytical use. Only contributions approved for analytical use, with confirmed rights and public-use permission, enter the contributed analytical evidence corpus. Existing source text and contributor changes remain distinguishable. Closing and reopening a form preserves its draft during the current page session; device-saved notes and retry uploads require their explicit save options.

Ask and report generation

Ask uses eligible repository evidence and gives citations and uncertainty. Reuse is allowed only when the question and relevant context match the cache requirements, including the evidence version. It does not assume that any similar-looking question has the same answer. Reuse is an internal process; it does not make unsupported evidence acceptable.

Ask privacy: questions and answers are stored privately for service review and retrieval for 30 days. Expired conversation content is purged when Ask or its history is next accessed; accounting records retain no question or answer text after that purge. Email addresses and sensitive links are masked, but names or other identifying details may remain. Do not enter personal or confidential information. Administrators can inspect recent service history. Saved report files follow their separate private storage workflow.

Tools → Create report assembles a shared, cited content model. Choose a full report, executive brief, sector brief or meeting presentation; set scope and period, select chapters, actions and illustrations, then review the preview. Exporting the template does not require a new AI answer for every file. User-added recommendations are identified as such. Saved map snapshots retain their captured date and do not silently follow subsequent map changes.

PDF supports English, French and Spanish. Word and PowerPoint support headings and document settings for all seven interface languages, including right-to-left Arabic and Persian. Source narratives remain in their original language unless a reviewed translation is available. Interface translation does not make the underlying analysis a reviewed translation.

Updates and dates

“Scan monitored source pages” detects changes to monitored web pages. A changed page is not a verified factual update, and that button does not itself extract, verify or publish new analysis. Published revisions, the tracker evidence cutoff and each source or map layer’s own date describe different things. Check them together; the cutoff does not mean every country has an observation from that day.

Disclaimer & data limitations

This is a public beta under active development. Information, source matching, translations, AI-assisted answers and analyses may contain errors or omissions. Information is provided as is, without a guarantee of accuracy, completeness, timeliness or fitness for a particular purpose. Forecasts are uncertain. The portal does not warrant the correctness of original third-party reports or endorse their conclusions. To the extent permitted by applicable law, liability is excluded for errors, omissions and loss arising from use of the portal, its analyses or linked and reproduced third-party material. This limitation does not exclude responsibilities that cannot lawfully be excluded.

Use the portal to locate evidence, compare conditions, explore stated scenarios and prepare briefings. Follow the issuing national authority for warnings and operational instructions. Coverage is uneven by country, sector, period and population group; absence from the portal does not mean that no relevant information exists. Public beta means that the interface, coverage and methods may change as they are reviewed.

AI-generated interpretations can be incomplete or incorrect and are not automatically validated by a human reviewer. Do not use them as the sole basis for emergency, safety-critical or funding decisions. Check the original source and responsible authority.

Country names and boundaries support navigation; they do not imply a position on legal status or endorsement by the organizations cited. Historical losses are not current caseloads. Forecast probabilities and climate anomalies are not measures of affected people.

Reuse and responsibility

Portal-authored content may be reused free with attribution and without sale or paid access. Original sources retain their own rights. Read the full reuse terms and limitations of responsibility.