ERA5 Precipitation Teleconnection Analysis

Niño3.4 (ENSO)

Correlation of Niño3.4 (ENSO) with total seasonal rainfall — total association (0–3 mo lag, country mean)

Niño3.4 (ENSO) total correlation l3 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Total association — pairwise Pearson r, p<0.05. Split diagonal = significant in both directions across non-overlapping seasons.

Correlation of Niño3.4 (ENSO) with total seasonal rainfall — unique signal (0–3 mo lag, country mean)

Niño3.4 (ENSO) partial correlation l3 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of Niño3.4 (ENSO) with total seasonal rainfall — total association (0–6 mo lag, country mean)

Niño3.4 (ENSO) total correlation l6 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Total association — pairwise Pearson r, p<0.05. Split diagonal = significant in both directions across non-overlapping seasons.

Correlation of Niño3.4 (ENSO) with total seasonal rainfall — unique signal (0–6 mo lag, country mean)

Niño3.4 (ENSO) partial correlation l6 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of Niño3.4 (ENSO) with total seasonal rainfall — total association (0–3 mo lag, 0.25° grid cell)

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Total association — pairwise Pearson r, p<0.05. Hatching = significant in both directions across non-overlapping seasons; the fill shows the stronger of the two.

Correlation of Niño3.4 (ENSO) with total seasonal rainfall — unique signal (0–3 mo lag, 0.25° grid cell)

Niño3.4 (ENSO) partial correlation l3 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of Niño3.4 (ENSO) with total seasonal rainfall — total association (0–6 mo lag, 0.25° grid cell)

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Total association — pairwise Pearson r, p<0.05. Hatching = significant in both directions across non-overlapping seasons; the fill shows the stronger of the two.

Correlation of Niño3.4 (ENSO) with total seasonal rainfall — unique signal (0–6 mo lag, 0.25° grid cell)

Niño3.4 (ENSO) partial correlation l6 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

IOD (Indian Ocean Dipole)

Correlation of IOD (Indian Ocean Dipole) with total seasonal rainfall — total association (0–3 mo lag, country mean)

IOD (Indian Ocean Dipole) total correlation l3 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Total association — pairwise Pearson r, p<0.05. Split diagonal = significant in both directions across non-overlapping seasons.

Correlation of IOD (Indian Ocean Dipole) with total seasonal rainfall — unique signal (0–3 mo lag, country mean)

IOD (Indian Ocean Dipole) partial correlation l3 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of IOD (Indian Ocean Dipole) with total seasonal rainfall — total association (0–6 mo lag, country mean)

IOD (Indian Ocean Dipole) total correlation l6 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Total association — pairwise Pearson r, p<0.05. Split diagonal = significant in both directions across non-overlapping seasons.

Correlation of IOD (Indian Ocean Dipole) with total seasonal rainfall — unique signal (0–6 mo lag, country mean)

IOD (Indian Ocean Dipole) partial correlation l6 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of IOD (Indian Ocean Dipole) with total seasonal rainfall — total association (0–3 mo lag, 0.25° grid cell)

IOD (Indian Ocean Dipole) total correlation l3 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Total association — pairwise Pearson r, p<0.05. Hatching = significant in both directions across non-overlapping seasons; the fill shows the stronger of the two.

Correlation of IOD (Indian Ocean Dipole) with total seasonal rainfall — unique signal (0–3 mo lag, 0.25° grid cell)

IOD (Indian Ocean Dipole) partial correlation l3 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of IOD (Indian Ocean Dipole) with total seasonal rainfall — total association (0–6 mo lag, 0.25° grid cell)

IOD (Indian Ocean Dipole) total correlation l6 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Total association — pairwise Pearson r, p<0.05. Hatching = significant in both directions across non-overlapping seasons; the fill shows the stronger of the two.

Correlation of IOD (Indian Ocean Dipole) with total seasonal rainfall — unique signal (0–6 mo lag, 0.25° grid cell)

IOD (Indian Ocean Dipole) partial correlation l6 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

TNA (Tropical N. Atlantic)

Correlation of TNA (Tropical N. Atlantic) with total seasonal rainfall — total association (0–3 mo lag, country mean)

TNA (Tropical N. Atlantic) total correlation l3 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Total association — pairwise Pearson r, p<0.05. Split diagonal = significant in both directions across non-overlapping seasons.

Correlation of TNA (Tropical N. Atlantic) with total seasonal rainfall — unique signal (0–3 mo lag, country mean)

TNA (Tropical N. Atlantic) partial correlation l3 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of TNA (Tropical N. Atlantic) with total seasonal rainfall — total association (0–6 mo lag, country mean)

TNA (Tropical N. Atlantic) total correlation l6 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Total association — pairwise Pearson r, p<0.05. Split diagonal = significant in both directions across non-overlapping seasons.

Correlation of TNA (Tropical N. Atlantic) with total seasonal rainfall — unique signal (0–6 mo lag, country mean)

TNA (Tropical N. Atlantic) partial correlation l6 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of TNA (Tropical N. Atlantic) with total seasonal rainfall — total association (0–3 mo lag, 0.25° grid cell)

TNA (Tropical N. Atlantic) total correlation l3 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Total association — pairwise Pearson r, p<0.05. Hatching = significant in both directions across non-overlapping seasons; the fill shows the stronger of the two.

Correlation of TNA (Tropical N. Atlantic) with total seasonal rainfall — unique signal (0–3 mo lag, 0.25° grid cell)

TNA (Tropical N. Atlantic) partial correlation l3 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of TNA (Tropical N. Atlantic) with total seasonal rainfall — total association (0–6 mo lag, 0.25° grid cell)

TNA (Tropical N. Atlantic) total correlation l6 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Total association — pairwise Pearson r, p<0.05. Hatching = significant in both directions across non-overlapping seasons; the fill shows the stronger of the two.

Correlation of TNA (Tropical N. Atlantic) with total seasonal rainfall — unique signal (0–6 mo lag, 0.25° grid cell)

TNA (Tropical N. Atlantic) partial correlation l6 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

TSA (Tropical S. Atlantic)

Correlation of TSA (Tropical S. Atlantic) with total seasonal rainfall — total association (0–3 mo lag, country mean)

TSA (Tropical S. Atlantic) total correlation l3 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Total association — pairwise Pearson r, p<0.05. Split diagonal = significant in both directions across non-overlapping seasons.

Correlation of TSA (Tropical S. Atlantic) with total seasonal rainfall — unique signal (0–3 mo lag, country mean)

TSA (Tropical S. Atlantic) partial correlation l3 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of TSA (Tropical S. Atlantic) with total seasonal rainfall — total association (0–6 mo lag, country mean)

TSA (Tropical S. Atlantic) total correlation l6 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Total association — pairwise Pearson r, p<0.05. Split diagonal = significant in both directions across non-overlapping seasons.

Correlation of TSA (Tropical S. Atlantic) with total seasonal rainfall — unique signal (0–6 mo lag, country mean)

TSA (Tropical S. Atlantic) partial correlation l6 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of TSA (Tropical S. Atlantic) with total seasonal rainfall — total association (0–3 mo lag, 0.25° grid cell)

TSA (Tropical S. Atlantic) total correlation l3 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Total association — pairwise Pearson r, p<0.05. Hatching = significant in both directions across non-overlapping seasons; the fill shows the stronger of the two.

Correlation of TSA (Tropical S. Atlantic) with total seasonal rainfall — unique signal (0–3 mo lag, 0.25° grid cell)

TSA (Tropical S. Atlantic) partial correlation l3 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of TSA (Tropical S. Atlantic) with total seasonal rainfall — total association (0–6 mo lag, 0.25° grid cell)

TSA (Tropical S. Atlantic) total correlation l6 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Total association — pairwise Pearson r, p<0.05. Hatching = significant in both directions across non-overlapping seasons; the fill shows the stronger of the two.

Correlation of TSA (Tropical S. Atlantic) with total seasonal rainfall — unique signal (0–6 mo lag, 0.25° grid cell)

TSA (Tropical S. Atlantic) partial correlation l6 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

AMM (Atlantic Meridional Mode)

Correlation of AMM (Atlantic Meridional Mode) with total seasonal rainfall — total association (0–3 mo lag, country mean)

AMM (Atlantic Meridional Mode) total correlation l3 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Total association — pairwise Pearson r, p<0.05. Split diagonal = significant in both directions across non-overlapping seasons.

Correlation of AMM (Atlantic Meridional Mode) with total seasonal rainfall — unique signal (0–3 mo lag, country mean)

AMM (Atlantic Meridional Mode) partial correlation l3 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of AMM (Atlantic Meridional Mode) with total seasonal rainfall — total association (0–6 mo lag, country mean)

AMM (Atlantic Meridional Mode) total correlation l6 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Total association — pairwise Pearson r, p<0.05. Split diagonal = significant in both directions across non-overlapping seasons.

Correlation of AMM (Atlantic Meridional Mode) with total seasonal rainfall — unique signal (0–6 mo lag, country mean)

AMM (Atlantic Meridional Mode) partial correlation l6 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of AMM (Atlantic Meridional Mode) with total seasonal rainfall — total association (0–3 mo lag, 0.25° grid cell)

AMM (Atlantic Meridional Mode) total correlation l3 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Total association — pairwise Pearson r, p<0.05. Hatching = significant in both directions across non-overlapping seasons; the fill shows the stronger of the two.

Correlation of AMM (Atlantic Meridional Mode) with total seasonal rainfall — unique signal (0–3 mo lag, 0.25° grid cell)

AMM (Atlantic Meridional Mode) partial correlation l3 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of AMM (Atlantic Meridional Mode) with total seasonal rainfall — total association (0–6 mo lag, 0.25° grid cell)

AMM (Atlantic Meridional Mode) total correlation l6 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Total association — pairwise Pearson r, p<0.05. Hatching = significant in both directions across non-overlapping seasons; the fill shows the stronger of the two.

Correlation of AMM (Atlantic Meridional Mode) with total seasonal rainfall — unique signal (0–6 mo lag, 0.25° grid cell)

AMM (Atlantic Meridional Mode) partial correlation l6 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

PDO (Pacific Decadal Oscillation)

Correlation of PDO (Pacific Decadal Oscillation) with total seasonal rainfall — total association (0–3 mo lag, country mean)

PDO (Pacific Decadal Oscillation) total correlation l3 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Total association — pairwise Pearson r, p<0.05. Split diagonal = significant in both directions across non-overlapping seasons.

Correlation of PDO (Pacific Decadal Oscillation) with total seasonal rainfall — unique signal (0–3 mo lag, country mean)

PDO (Pacific Decadal Oscillation) partial correlation l3 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of PDO (Pacific Decadal Oscillation) with total seasonal rainfall — total association (0–6 mo lag, country mean)

PDO (Pacific Decadal Oscillation) total correlation l6 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Total association — pairwise Pearson r, p<0.05. Split diagonal = significant in both directions across non-overlapping seasons.

Correlation of PDO (Pacific Decadal Oscillation) with total seasonal rainfall — unique signal (0–6 mo lag, country mean)

PDO (Pacific Decadal Oscillation) partial correlation l6 adm0

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) No signal Not calculated

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of PDO (Pacific Decadal Oscillation) with total seasonal rainfall — total association (0–3 mo lag, 0.25° grid cell)

PDO (Pacific Decadal Oscillation) total correlation l3 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Total association — pairwise Pearson r, p<0.05. Hatching = significant in both directions across non-overlapping seasons; the fill shows the stronger of the two.

Correlation of PDO (Pacific Decadal Oscillation) with total seasonal rainfall — unique signal (0–3 mo lag, 0.25° grid cell)

PDO (Pacific Decadal Oscillation) partial correlation l3 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Correlation of PDO (Pacific Decadal Oscillation) with total seasonal rainfall — total association (0–6 mo lag, 0.25° grid cell)

PDO (Pacific Decadal Oscillation) total correlation l6 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Total association — pairwise Pearson r, p<0.05. Hatching = significant in both directions across non-overlapping seasons; the fill shows the stronger of the two.

Correlation of PDO (Pacific Decadal Oscillation) with total seasonal rainfall — unique signal (0–6 mo lag, 0.25° grid cell)

PDO (Pacific Decadal Oscillation) partial correlation l6 px

Positive strong (r≥0.5) Positive moderate (0.30–0.5) Negative moderate Negative strong (r≤−0.5) Both signs across seasons No signal Arid — no wet season Ocean / outside grid

Unique signal — partial r, other climate modes held constant. Shrinkage vs Total = shared variance, not absent signal.

Index Collinearity

Pearson r between climate indices over the full analysis period (1981–2025). High collinearity (e.g. AMM–TNA r≈0.81) means total vs unique signal may diverge substantially for those modes — shrinkage in the unique-signal map reflects shared variance, not absent signal.

Dominant Climate Mode

Each country colored by the index with the strongest significant total correlation (|r|≥0.3). Split diagonal = second-strongest index (different mode), shown only when it acts on a non-overlapping trimester. Respects the Max lag toggle above.

Each 0.25° land cell colored by the index with the strongest significant total correlation (|r|≥0.3), where that mode leads the runner-up by at least 0.10. Respects the Max lag toggle above.

Niño3.4 (ENSO)IOD (Indian Ocean Dipole)TNA (Tropical N. Atlantic)TSA (Tropical S. Atlantic)AMM (Atlantic Meridional Mode)PDO (Pacific Decadal Oscillation) No signal Not calculated

Niño3.4 (ENSO)IOD (Indian Ocean Dipole)TNA (Tropical N. Atlantic)TSA (Tropical S. Atlantic)AMM (Atlantic Meridional Mode)PDO (Pacific Decadal Oscillation) No signal Not calculated

Niño3.4 (ENSO)IOD (Indian Ocean Dipole)TNA (Tropical N. Atlantic)TSA (Tropical S. Atlantic)AMM (Atlantic Meridional Mode)PDO (Pacific Decadal Oscillation) No single dominant mode (top two within 0.10 of each other) No signal Arid — no wet season Ocean / outside grid

Top mode only — the runner-up mode shown as a split diagonal on the country map is not legible at 0.25°. A cell is coloured only where the leading mode's |r| exceeds the runner-up's by at least 0.10; a single grid cell does not average enough area to make the arg-max over six collinear modes stable, so cells where the top two are within noise of each other are left grey rather than assigned to an arbitrary winner.

ENSO Composites — El Niño & La Niña

Mean rainfall anomaly (standard deviations from climatology) in each unit's headline trimester when Niño3.4 ≥ ±0.5 concurrent with that trimester. The two maps are independent — ENSO impacts are asymmetric. Respects the Resolution toggle above.

El Niño composite — mean anomaly when Niño3.4 ≥ +0.5 (brown = drier than normal, blue = wetter).

La Niña composite — mean anomaly when Niño3.4 ≤ −0.5. Roughly opposite to El Niño in ENSO-sensitive regions, but magnitude and pattern differ.

El Niño composite rainfall anomaly, pixel

El Niño composite (0.25° grid) — mean anomaly when Niño3.4 ≥ +0.5 (brown = drier than normal, blue = wetter), scored per cell in its own headline season.

La Niña composite rainfall anomaly, pixel

La Niña composite (0.25° grid) — mean anomaly when Niño3.4 ≤ −0.5.

Reference — Typical Global ENSO Impacts

Generalized seasonal impact patterns associated with El Niño and La Niña, for context. These are schematic climatological composites, not derived from this analysis. Each map shows Northern Hemisphere winter (top) and summer (bottom). Source: NOAA PMEL, via climate.gov.

Typical global El Niño impacts for winter and summer

El Niño — typical global impacts for winter (top) and summer (bottom). Credit: climate.gov.

Typical global La Niña impacts for winter and summer

La Niña — typical global impacts for winter (top) and summer (bottom). Credit: climate.gov.

Documented ENSO Impacts (Literature)

The generalized NOAA and IRI impact maps above are built from a few canonical global studies and miss several real teleconnections — most notably summer rainfall over highland Yemen and the southwestern Arabian Peninsula. The maps and table below are a curated, citation-backed catalogue of documented ENSO–seasonal-rainfall links covering all 153 monitored countries (those in the ERA5 analysis), researched region by region from peer-reviewed literature plus authoritative operational sources (NOAA, IRI, ICPAC/IGAD). Rows tagged gap vs NOAA/IRI are well-supported in the literature yet absent or under-represented on the standard maps. The final column reports what this study's own ERA5 analysis found, as an independent check.

Documented El Niño rainfall impacts across humanitarian-priority regions

Showing El Niño — dominant documented rainfall response in each country's main humanitarian rainy season.

El Niño wetter:single studymoderaterobustdrier:single studymoderaterobust(shade = confidence)Opposing signals (split)Monitored — no documented linkNot monitored (not researched)

Countries whose seasons respond in opposite directions (e.g. Ethiopia, Yemen, Brazil, Peru, China) are split; 3-month season codes (e.g. OND, JJAS, DJF) label each signal. Grey = monitored but no robust documented ENSO link in the literature; near-white = not monitored (not researched — e.g. USA, most of Europe). The curated highlights are in the table; every monitored country is in the collapsible catalogue below.

Region / area	Season	El Niño	La Niña	Evidence & source	This study (ERA5)
Horn of Africa Kenya, S. Somalia, SE Ethiopia	OND “short rains”	wetter	drier	Robust, widely replicated Park et al. 2020 (JGR-A)	Agrees — SOM OND r=+0.65, KEN +0.58, ETH +0.59
Horn of Africa Ethiopia / Sudan / S. Sudan highlands gap vs NOAA/IRI	JJAS (kiremt / Blue Nile)	drier	wetter	Robust Dai & Wigley 2000 (GRL)	Agrees — ETH JAS r=−0.63, SSD −0.59, SDN −0.40
Horn of Africa East Africa	MAM “long rains”	weak / none	weak / none	Long rains only weakly tied to ENSO Wainwright et al. 2018 (Clim Dyn)	Agrees — little MAM ENSO signal in the Horn
Sahel & W. Africa Senegal, Mali, Niger, Chad, N. Nigeria gap vs NOAA/IRI	JJAS monsoon	drier	wetter	Direction robust; mechanism contested Dai & Wigley 2000 (GRL)	Agrees — SEN JJA −0.49, MLI JAS −0.41, TCD −0.43
Middle East / Arabia SW Arabia / highland Yemen gap vs NOAA/IRI	JJA summer rains	drier (drought)	wetter	Single robust modelling+obs study; direct Pacific teleconnection Atif et al. 2017 (npj Clim Atmos Sci)	Agrees — YEM JAS r=−0.59 (summer); winter NDJ +0.53
Middle East / Arabia Iraq, Iran, Levant, N. Arabia gap vs NOAA/IRI	SON–DJF autumn–winter	wetter	drier	Multiple studies; autumn phase-transition evidence Middle East autumn–ENSO study 2019	Agrees — IRQ SON +0.63, IRN OND +0.59, SAU MAM +0.51
South Asia India (all-India monsoon)	JJAS	drier	wetter	Robust historically but non-stationary (weakened since ~1980s) Indian-monsoon non-stationarity reviews	India is not in the ERA5 set (no admin-0 extract); the ENSO–monsoon link is documented but weakening
South Asia Pakistan, Afghanistan	OND & MAM (winter–spring)	wetter	drier	Established Dai & Wigley 2000 (GRL)	Agrees — PAK OND +0.55, MAM +0.54; AFG AMJ +0.59
Central America Guatemala, Honduras, El Salvador, Nicaragua (Dry Corridor)	JJA–SON (incl. canícula)	drier (drought)	wetter	Canonical Dai & Wigley 2000 (GRL)	Agrees — GTM ASO −0.61, HND −0.60, NIC −0.61
Southern Africa Zimbabwe, Zambia, Malawi, Mozambique, S. Africa, Botswana	DJF austral summer	drier (drought)	wetter	Robust, widely replicated Mason & Goddard 2001 (BAMS)	Agrees — ZWE DJF −0.68, ZAF NDJ −0.62, MOZ JFM −0.50
SE Asia / Maritime Continent Indonesia, Philippines, Vietnam, Thailand	JJA–SON dry season	drier (drought)	wetter	Very robust (drought & fire risk) Dai & Wigley 2000 (GRL)	Agrees strongly — IDN ASO −0.87, PHL FMA −0.86, THA −0.81

Notes: directions are the dominant documented response — ENSO teleconnections are asymmetric, so the La Niña column is the typical (not guaranteed) opposite of El Niño. “Evidence” flags how well replicated each link is. The Yemen/SW-Arabia summer link is a direct Pacific teleconnection (Atif et al. 2017), not an Indian-Ocean-Dipole artefact. The ENSO–Indian-monsoon correlation is non-stationary and has weakened since the 1980s, consistent with the weak all-India signal in our ERA5 analysis. Framing on map incompleteness follows Lenssen, Goddard & Mason (2020), who detect many additional teleconnections when the standard maps are updated.

Complete per-country catalogue (128 documented of 154 monitored) — click to expand

Every monitored country (153 ERA5-analysis countries), researched region by region from the literature. Cell shade encodes confidence (hatched/pale = single study, solid/dark = robust); the “La Niña” column shows the documented opposite where known (italic = inferred inverse). “—” = no robust documented ENSO rainfall link (grey on the map). Final column is this study's strongest significant ERA5 Niño3.4 correlation (season and sign), or “no sig”. Rows tagged gap are documented links absent/under-represented on the standard NOAA/IRI maps.

Country	El Niño	La Niña	Season	Evidence	Source	ERA5 Niño3.4
Afghanistan AFG	wetter	drier	OND	moderate	Dai & Wigley 2000 (GRL)	MAM +0.55
Albania ALB	—	—	—	—	Mariotti et al. 2002 (GRL, Euro-Mediterranean rainfall & ENSO); Zhang et al. 2014 (J. Climate)	no sig
Algeria DZA gap	wetter	drier	JFM	moderate	Euro-Mediterranean late-winter ENSO study 2021 (Clim. Dyn.); Mariotti et al. 2002 (GRL)	SON +0.30
Angola AGO	drier	wetter	DJF	robust	Mason & Goddard 2001 (BAMS)	DJF +0.34
Antigua and Barbuda ATG gap	drier	wetter	ASO	moderate	Giannini, Kushnir & Cane 2000 (J. Climate); Jury, Malmgren & Winter 2007 (JGR-Atmos)	JAS −0.43
Argentina ARG	wetter	drier	NDJ	robust	Grimm et al. 2000 (J. Climate); Vera et al. 2004 (GRL); Cai et al. 2020 (Nat. Rev. Earth Environ.)	OND +0.54
Armenia ARM	—	—	—	—	Bronnimann 2007 (Reviews of Geophysics) - review finds no coherent Caucasus signal	no sig
Azerbaijan AZE	—	—	—	—	Bronnimann 2007 (Reviews of Geophysics)	no sig
Bahamas BHS gap	wetter	drier	DJF	moderate	Jury, Malmgren & Winter 2007 (JGR-Atmos); Giannini, Kushnir & Cane 2000 (J. Climate)	no sig
Barbados BRB gap	drier	wetter	ASO	moderate	Jury, Malmgren & Winter 2007 (JGR-Atmos); Giannini, Kushnir & Cane 2000 (J. Climate)	ASO −0.46
Belarus BLR	—	—	—	—	Bronnimann 2007 (Reviews of Geophysics)	JJA −0.32
Belize BLZ	drier	wetter	JJA	robust	Giannini, Kushnir & Cane 2000 (J. Climate); FEWS NET Central America/Belize analyses	ASO −0.51
Benin BEN gap	wetter	drier	ASO	single-study	Camberlin, Janicot & Poccard 2001 (Int. J. Climatol.)	no sig
Bermuda BMU	—	—	—	—	No robust peer-reviewed ENSO-Bermuda rainfall teleconnection	JFM +0.42
BGD BGD	—	—	—	—	—	JAS +0.32
Bhutan BTN gap	wetter	drier	JJA	moderate	Gyeltshen et al. 2024 (Int. J. Applied Earth Observation and Geoinformation); Kuenzang et al. 2021 (Atmosphere - MDPI, Bhutan ENSO/IOD case study)	no sig
Bolivia BOL gap	split	split	DJF wet / DJF dry	robust	Vuille 1999 (Int. J. Climatol.); Garreaud & Aceituno 2001 (J. Climate); Garreaud et al. 2003 (Palaeo3)	NDJ −0.31
Bonaire (Caribbean Netherlands) BES gap	drier	wetter	OND	moderate	Giannini, Kushnir & Cane 2000 (J. Climate); regional ABC-islands climatology (Curacao/Bonaire)	ASO −0.66
Botswana BWA	drier	wetter	DJF	robust	Mason & Goddard 2001 (BAMS)	NDJ −0.52
Brazil BRA	split	split	OND wet / FMA dry	robust	Grimm 2003/2004 (J. Climate); Grimm & Tedeschi 2009 (J. Climate); Coelho et al. 2002 (Int. J. Climatol.); Cai et al. 2020 (Nat. Rev. Earth Environ.)	NDJ −0.62
British Virgin Islands VGB gap	drier	wetter	ASO	single-study	Mendez-Lazaro et al. 2018 (Int. J. Biometeorol., PR/USVI region); Jury, Malmgren & Winter 2007 (JGR-Atmos)	JAS −0.45
Bulgaria BGR	—	—	—	—	Mariotti et al. 2002 (GRL); Zhang et al. 2014 (J. Climate)	MJJ +0.35
Burkina Faso BFA gap	drier	wetter	JJAS	moderate	Dai & Wigley 2000 (GRL)	JAS −0.38
Burundi BDI gap	wetter	drier	OND	moderate	Indeje et al. 2000 (Int. J. Climatol.); Nicholson 2017 (Reviews of Geophysics)	OND +0.46
Cambodia KHM	drier	wetter	JAS	robust	Dai & Wigley 2000 (GRL)	AMJ −0.72
Cameroon CMR gap	drier	wetter	JJA	moderate	Diatta & Fink 2014 (Int. J. Climatol.); Giannini, Saravanan & Chang 2003/2008 on Sahel ENSO; Nicholson 2018 (Glob. Planet. Change) review; Cameroon agroecological ENSO study 2025 (Earth Syst. Environ.)	no sig
Cape Verde CPV gap	drier	wetter	ASO	moderate	Janicot, Trzaska & Poccard 2001 (Clim. Dyn.); Camberlin et al. 2001 (Int. J. Climatol.)	JJA −0.36
Cayman Islands CYM	drier	wetter	JAS	single-study	Giannini, Kushnir & Cane 2000 (J. Climate); Jury, Malmgren & Winter 2007 (JGR-Atmos)	JAS −0.45
Central African Republic CAF	—	—	—	—	Nicholson 2018 (Glob. Planet. Change) review of African rainfall; Balas et al. 2007 (Int. J. Climatol.)	JAS −0.38
Chad TCD gap	drier	wetter	JJAS	moderate	Dai & Wigley 2000 (GRL)	JAS −0.43
Chile CHL	wetter	drier	JJA	robust	Montecinos & Aceituno 2003 (J. Climate); Garreaud et al. 2017 (Int. J. Climatol., megadrought)	JAS+ / AMJ−
China CHN	split	split	JJA wet / JJA dry	robust	Zhang et al. 1999; Wu et al. 2003 (J. Climate)	MAM +0.57
Colombia COL	drier	wetter	DJF	robust	Poveda et al. 2011 (Climate Dynamics); Poveda & Mesa 1997 (GRL); Cordoba-Machado et al. 2015 (Climate Dynamics)	JJA −0.50
Comoros COM	—	—	—	—	No robust peer-reviewed rainfall teleconnection (limited regional studies, e.g. Salim et al. conference work on northern Madagascar/Comoros)	NDJ +0.47
Costa Rica CRI	drier	wetter	JAS	robust	Dai & Wigley 2000 (GRL)	ASO −0.62
Cote d'Ivoire CIV gap	wetter	drier	ASO	moderate	Camberlin, Janicot & Poccard 2001 (Int. J. Climatol.)	MJJ −0.46
Cuba CUB gap	split	split	DJF wet / JAS dry	robust	Giannini, Kushnir & Cane 2000 (J. Climate); Jury, Malmgren & Winter 2007 (JGR-Atmos)	JAS −0.47
Democratic Republic of the Congo COD gap	split	split	OND wet / MAM dry	moderate	Balas, Nicholson & Klotter 2007 (Int. J. Climatol.); Preethi, Ratnam et al. 2015 (Sci. Rep.); Hua et al. 2016 (Clim. Dyn.); Nicholson 2018 (Glob. Planet. Change) review	no sig
Djibouti DJI gap	drier	wetter	JJAS	robust	Dai & Wigley 2000 (GRL)	JAS −0.62
Dominica DMA gap	drier	wetter	ASO	single-study	Giannini, Kushnir & Cane 2000 (J. Climate); Jury, Malmgren & Winter 2007 (JGR-Atmos)	ASO −0.47
Dominican Republic DOM	drier	wetter	MJJ	robust	Jury, Malmgren & Winter 2007 (JGR-Atmos); Giannini, Kushnir & Cane 2000 (J. Climate)	MAM+ / JAS−
Ecuador ECU	split	split	JFM wet / MAM dry	robust	Rossel & Cadier 2009 (J. Hydrology); Vicente-Serrano et al. 2017 (Int. J. Climatol.); Bendix & Bendix 2006	MAM +0.50
Egypt EGY gap	wetter	drier	JFM	moderate	Rainfall variability over Sinai Peninsula and its teleconnection to El Nino SST 2021 (J. Arid Environ.); Euro-Med late-winter ENSO study 2021 (Clim. Dyn.)	no sig
El Salvador SLV	drier	wetter	JAS	robust	Dai & Wigley 2000 (GRL)	ASO −0.62
Equatorial Guinea GNQ gap	drier	wetter	MAM	single-study	Balas, Nicholson & Klotter 2007 (Int. J. Climatol.); Dezfuli & Nicholson 2013 (J. Climate)	AMJ −0.41
Eritrea ERI gap	drier	wetter	JJAS	robust	Dai & Wigley 2000 (GRL)	JAS −0.41
Eswatini SWZ	drier	wetter	DJF	robust	Pomposi et al. 2018 (Environmental Research Letters); Steinkopf & Engelbrecht 2025 (Environmental Research Letters)	DJF −0.59
Ethiopia ETH gap	split	split	OND wet / JJAS dry	robust	Dai & Wigley 2000; Korecha & Barnston 2007	JAS −0.63
Fiji FJI	drier	wetter	DJF	robust	Vincent et al. 2011 (Clim. Dyn.); Cai et al. 2012 (Nature); Brown et al. 2020 (Nat. Rev. Earth Environ.); PACCSAP Fiji report 2014; McGree et al. 2014 (Int. J. Climatol.)	NDJ −0.62
French Guiana GUF gap	drier	wetter	MAM	single-study	Bovolo et al. 2012 (Int. J. Climatol.); Yoon & Zeng 2010 (Amazon ENSO, Climate Dynamics); regional Meteo-France analyses	DJF −0.69
Gabon GAB gap	drier	wetter	MAM	moderate	Marechal et al. 2021 (Proc. IAHS 384); Dezfuli & Nicholson 2013 Part I (J. Climate); Balas et al. 2007 (Int. J. Climatol.)	no sig
Gambia GMB gap	drier	wetter	JJAS	moderate	Dai & Wigley 2000 (GRL)	JJA −0.38
Georgia GEO	—	—	—	—	Bronnimann 2007 (Reviews of Geophysics)	no sig
Ghana GHA gap	wetter	drier	ASO	moderate	Camberlin, Janicot & Poccard 2001 (Int. J. Climatol.)	JJA −0.41
Grenada GRD gap	drier	wetter	DJF	single-study	Jury, Malmgren & Winter 2007 (JGR-Atmos); Sookram et al. (ENSO dry-season Trinidad, regional analogue)	ASO −0.54
Guadeloupe GLP gap	drier	wetter	ASO	single-study	Giannini, Kushnir & Cane 2000 (J. Climate); Jury, Malmgren & Winter 2007 (JGR-Atmos)	JAS −0.43
Guatemala GTM	drier	wetter	JAS	robust	Dai & Wigley 2000 (GRL)	ASO −0.61
Guinea GIN gap	drier	wetter	JJAS	moderate	Dai & Wigley 2000 (GRL)	JJA −0.32
Guinea-Bissau GNB gap	drier	wetter	JJAS	moderate	Dai & Wigley 2000 (GRL)	no sig
Guyana GUY gap	drier	wetter	JJA	moderate	Bovolo et al. 2012 (Int. J. Climatol.); IRI/CariCOF & WMO regional analyses; Giannini et al. 2000 (Caribbean ENSO)	JAS −0.55
Haiti HTI	drier	wetter	MJJ	robust	Jury, Malmgren & Winter 2007 (JGR-Atmos); FEWS NET Haiti climate analyses	JAS −0.57
Honduras HND	drier	wetter	JAS	robust	Dai & Wigley 2000 (GRL)	ASO −0.60
Hungary HUN	—	—	—	—	Bronnimann 2007 (Reviews of Geophysics); Zhang et al. 2014 (J. Climate)	no sig
India IND (not monitored)	drier	wetter	JJAS	moderate	Indian-monsoon non-stationarity reviews	no sig
Indonesia IDN	drier	wetter	JAS	robust	Dai & Wigley 2000 (GRL)	OND −0.70
Iran IRN gap	wetter	drier	SON	moderate	Middle East autumn–ENSO study 2019	MAM +0.50
Iraq IRQ gap	wetter	drier	SON	moderate	Middle East autumn–ENSO study 2019	OND +0.62
Jamaica JAM	drier	wetter	ASO	robust	Chen & Taylor 2002 (Int. J. Climatol.); Taylor, Enfield & Chen 2002 (JGR-Oceans); Giannini, Kushnir & Cane 2000 (J. Climate)	JAS −0.48
Kazakhstan KAZ gap	wetter	drier	MAM	robust	Yao et al. 2024 (npj Climate and Atmospheric Science)	OND +0.41
Kenya KEN	wetter	drier	OND	robust	Park et al. 2020 (JGR-A)	NDJ +0.61
Kiribati KIR	wetter	drier	DJF	robust	PCCSP/PACCSAP Kiribati country report 2011/2014; Australian BoM-CSIRO Pacific Climate Change Science Program 2011	DJF +0.91
Kuwait KWT gap	wetter	drier	DJF	moderate	Sandeep & Ajayamohan 2018 (J. Geophys. Res. Atmos., Arabian Gulf); Marcella & Eltahir / Kuwait hydroclimatology studies	OND +0.40
Kyrgyzstan KGZ gap	wetter	drier	MAM	robust	Yao et al. 2024 (npj Climate and Atmospheric Science)	AMJ +0.52
Laos LAO	drier	wetter	JAS	robust	Dai & Wigley 2000 (GRL)	JAS+ / AMJ−
Lebanon LBN gap	wetter	drier	DJF	moderate	Mariotti, Zeng & Lau 2002 (Geophys. Res. Lett.); Hoell et al. 2024 (J. Climate, Fertile Crescent)	OND +0.47
Lesotho LSO	drier	wetter	DJF	robust	Mason & Goddard 2001 (BAMS)	NDJ −0.48
Liberia LBR	—	—	—	—	Camberlin, Janicot & Poccard 2001 (Int. J. Climatol.)	no sig
Libya LBY gap	wetter	drier	DJF	single-study	Euro-Mediterranean late-winter ENSO study 2021 (Clim. Dyn.); Mariotti et al. 2002 (GRL)	DJF −0.30
LKA LKA	—	—	—	—	—	SON +0.49
Madagascar MDG gap	split	split	NDJ wet / DJF dry	moderate	Steinkopf & Engelbrecht 2025 (Environmental Research Letters); FEWS NET El Nino impact assessment 2024	no sig
Malawi MWI	drier	wetter	DJF	robust	Mason & Goddard 2001 (BAMS)	FMA −0.41
Malaysia MYS	drier	wetter	JAS	robust	Dai & Wigley 2000 (GRL)	DJF −0.57
Maldives MDV gap	drier	wetter	MJJ	moderate	Foley & Kelman 2020 (Int. J. Disaster Risk Reduction 50:101726); follow-up Kelman et al. 2025 (Theoretical and Applied Climatology)	OND +0.64
Mali MLI gap	drier	wetter	JJAS	moderate	Dai & Wigley 2000 (GRL)	JAS −0.41
Marshall Islands MHL gap	split	drier	OND wet / FMA dry	robust	Pacific ENSO Applications Climate Center (PEAC) / NWS; Pacific RISA 'El Nino and its Impacts on the RMI' 2015; PACCSAP/PCCSP Marshall Islands country report 2011/2014	JJA +0.41
Martinique MTQ gap	drier	wetter	ASO	single-study	Giannini, Kushnir & Cane 2000 (J. Climate); Jury, Malmgren & Winter 2007 (JGR-Atmos)	MJJ+ / ASO−
Mauritania MRT gap	drier	wetter	JJAS	moderate	Dai & Wigley 2000 (GRL)	JJA −0.54
Mauritius MUS	—	—	—	—	No robust peer-reviewed rainfall teleconnection; cf. Mascarene-High variability studies (Xulu et al. 2020, Sci. Rep.)	no sig
Mexico MEX	split	split	DJF wet / JJA dry	robust	Magana, Vazquez, Perez & Perez 2003 (Geofisica Internacional); Pavia, Graef & Reyes 2006 (J. Climate)	JAS −0.58
Micronesia (Federated States of) FSM gap	split	drier	OND wet / MAM dry	robust	Pacific ENSO Applications Climate Center (PEAC) / NWS; PACCSAP FSM country report 2014; Lander & Khosrowpanah; PEAC El Nino impacts on Eastern FSM fact sheet	JJA+ / OND−
Moldova MDA	—	—	—	—	Mariotti et al. 2002 (GRL); Bronnimann 2007 (Rev. Geophys.)	no sig
Mongolia MNG	—	—	—	—	Hessl et al. 2018 (Science Advances); Wu et al. 2019 (Climate Dynamics, EASM-ENSO)	no sig
Morocco MAR gap	split	split	NDJ wet / NDJ dry	moderate	Lopez-Parages & Rodriguez-Fonseca 2012 / Mariotti et al. 2002 (Mediterranean ENSO); Euro-Med late-winter ENSO study 2021 (Clim. Dyn.)	no sig
Mozambique MOZ	drier	wetter	DJF	robust	Mason & Goddard 2001 (BAMS)	JFM −0.50
Myanmar MMR	drier	wetter	JAS	robust	Dai & Wigley 2000 (GRL)	JAS+ / AMJ−
Namibia NAM	drier	wetter	DJF	robust	Mason & Goddard 2001 (BAMS)	JFM −0.57
Nicaragua NIC	drier	wetter	JAS	robust	Dai & Wigley 2000 (GRL)	ASO −0.61
Niger NER gap	drier	wetter	JJAS	moderate	Dai & Wigley 2000 (GRL)	JJA −0.31
Nigeria NGA gap	drier	wetter	JJAS	moderate	Dai & Wigley 2000 (GRL)	JAS −0.40
North Korea PRK	—	—	—	—	Choi et al. 2014 (Climate Dynamics); Ho et al. 2016 (Asia-Pacific J. Atmos. Sci.)	JAS −0.35
NPL NPL	—	—	JJAS	—	—	MJJ −0.57
Oman OMN gap	wetter	drier	MAM	moderate	Middle East autumn–ENSO study 2019	FMA +0.42
Pakistan PAK	wetter	drier	OND	moderate	Dai & Wigley 2000 (GRL)	FMA+ / JAS−
Palestine PSE gap	wetter	drier	DJF	single-study	Mariotti, Zeng & Lau 2002 (Geophys. Res. Lett.); Price, Stone & Rind 1998 (J. Climate, E. Med. rainfall-ENSO)	OND +0.56
Panama PAN	drier	wetter	JAS	robust	Dai & Wigley 2000 (GRL)	JJA −0.72
Papua New Guinea PNG	drier	wetter	DJF	robust	Murphy, Power & McGree 2014 (J. Climate); Cai et al. 2011 (J. Climate); PACCSAP/PCCSP PNG country report 2011/2014	JFM +0.39
Paraguay PRY	wetter	drier	OND	robust	Grimm et al. 2000 (J. Climate); Barros et al. 2000/2008; Cai et al. 2020 (Nat. Rev. Earth Environ.)	OND +0.44
Peru PER	split	split	DJF wet / DJF dry	robust	Lagos et al. 2008 (Adv. Geosci.); Garreaud & Aceituno 2001 (J. Climate); Sulca et al. 2018 (Int. J. Climatol.)	OND +0.32
Philippines PHL	drier	wetter	JAS	robust	Dai & Wigley 2000 (GRL)	OND −0.74
Poland POL	—	—	—	—	Bronnimann 2007 (Reviews of Geophysics)	JJA −0.49
Puerto Rico PRI	drier	wetter	AMJ	robust	Mendez-Lazaro et al. 2018 (Int. J. Biometeorology, Teleconnections between ENSO and rainfall/drought in Puerto Rico); Jury, Malmgren & Winter 2007 (JGR-Atmos)	MAM+ / JAS−
Qatar QAT gap	wetter	drier	DJF	single-study	Sandeep & Ajayamohan 2018 (J. Geophys. Res. Atmos., Arabian Gulf); Hoell et al. 2024 (J. Climate, Middle East/SW Asia)	MAM +0.41
Republic of the Congo COG gap	drier	wetter	MAM	moderate	Balas, Nicholson & Klotter 2007 (Int. J. Climatol.); Dezfuli & Nicholson 2013 Part I Boreal Spring (J. Climate)	no sig
Romania ROU	—	—	—	—	Mariotti et al. 2002 (GRL, Euro-Mediterranean rainfall & ENSO); Bronnimann 2007 (Rev. Geophys.)	no sig
Russia RUS	—	—	—	—	Bronnimann 2007 (Reviews of Geophysics)	JAS −0.37
Rwanda RWA gap	wetter	drier	OND	moderate	Indeje et al. 2000 (Int. J. Climatol.); Nicholson 2017 (Reviews of Geophysics)	NDJ +0.49
Saint Kitts and Nevis KNA gap	drier	wetter	ASO	single-study	Giannini, Kushnir & Cane 2000 (J. Climate); Jury, Malmgren & Winter 2007 (JGR-Atmos)	JAS −0.45
Saint Lucia LCA gap	drier	wetter	ASO	single-study	Giannini, Kushnir & Cane 2000 (J. Climate); Jury, Malmgren & Winter 2007 (JGR-Atmos); CariCOF outlooks	MJJ+ / ASO−
Saint Vincent and the Grenadines VCT gap	drier	wetter	ASO	single-study	Giannini, Kushnir & Cane 2000 (J. Climate); Jury, Malmgren & Winter 2007 (JGR-Atmos)	ASO −0.57
Sao Tome and Principe STP	—	—	—	—	Balas et al. 2007 (Int. J. Climatol.); Nicholson 2018 (Glob. Planet. Change) review (Atlantic-dominated western-equatorial regime)	no sig
Saudi Arabia SAU gap	wetter	drier	MAM	moderate	Middle East autumn–ENSO study 2019	MAM +0.51
Senegal SEN gap	drier	wetter	JJAS	moderate	Dai & Wigley 2000 (GRL)	JJA −0.49
Seychelles SYC gap	drier	wetter	NDJ	single-study	Seychelles Meteorological Authority / regional operational guidance (ICPAC GHACOF)	OND +0.50
Sierra Leone SLE	—	—	—	—	Camberlin, Janicot & Poccard 2001 (Int. J. Climatol.)	no sig
Slovakia SVK	—	—	—	—	Bronnimann 2007 (Reviews of Geophysics)	JAS −0.32
Solomon Islands SLB	drier	wetter	DJF	robust	Vincent et al. 2011 (Clim. Dyn.); Cai et al. 2011/2012; PACCSAP Solomon Islands country report 2014; Brown et al. 2020 (Nat. Rev. Earth Environ.)	NDJ −0.33
Somalia SOM	wetter	drier	OND	robust	Park et al. 2020 (JGR-A)	OND +0.65
South Africa ZAF	drier	wetter	DJF	robust	Mason & Goddard 2001 (BAMS)	NDJ −0.62
South Sudan SSD gap	drier	wetter	JJAS	robust	Dai & Wigley 2000 (GRL)	JAS −0.59
Sudan SDN gap	drier	wetter	JJAS	robust	Dai & Wigley 2000 (GRL)	JAS −0.40
Suriname SUR gap	drier	wetter	MAM	moderate	Bovolo et al. 2012 (Int. J. Climatol.); IRI/CariCOF regional outlooks; Nurmohamed et al. (Suriname hydrology)	JJA −0.64
Syria SYR gap	wetter	drier	SON	moderate	Middle East autumn–ENSO study 2019	OND +0.47
Tanzania TZA gap	split	split	OND wet / DJF dry	robust	Nicholson 2017 (Reviews of Geophysics); Palmer et al. 2023 (Nature Reviews Earth & Environment); Indeje et al. 2000 (Int. J. Climatol.)	NDJ +0.50
Thailand THA	drier	wetter	JAS	robust	Dai & Wigley 2000 (GRL)	AMJ −0.70
Timor-Leste TLS	drier	wetter	DJF	robust	Hendon 2003 (J. Climate 16:1775); Aldrian & Susanto 2003 (Int. J. Climatology 23:1435); Moron et al. 2009 (Int. J. Climatology, Indonesian monsoon onset predictability)	NDJ −0.72
Togo TGO gap	wetter	drier	ASO	single-study	Camberlin, Janicot & Poccard 2001 (Int. J. Climatol.)	no sig
Tonga TON	drier	wetter	DJF	robust	Vincent et al. 2011 (Clim. Dyn.); Cai et al. 2012 (Nature); PACCSAP Tonga country report 2014; Brown et al. 2020 (Nat. Rev. Earth Environ.)	DJF −0.63
Trinidad and Tobago TTO	drier	wetter	DJF	moderate	Sookram & ... (Impact of ENSO on dry-season rainfall in Trinidad); Jury, Malmgren & Winter 2007 (JGR-Atmos)	ASO −0.57
Tunisia TUN gap	wetter	drier	JFM	moderate	Euro-Mediterranean late-winter ENSO study 2021 (Clim. Dyn.); Mariotti, Zeng & Lau 2002 (GRL)	no sig
Turkey TUR gap	wetter	drier	DJF	robust	Karabork & Kahya 2009 (Hydrological Processes); Karabork, Kahya & Karaca 2005 (Hydrological Processes)	AMJ +0.39
Turks and Caicos Islands TCA gap	drier	wetter	ASO	single-study	Jury, Malmgren & Winter 2007 (JGR-Atmos); Giannini, Kushnir & Cane 2000 (J. Climate)	JAS −0.30
UAE ARE gap	wetter	drier	MAM	moderate	Middle East autumn–ENSO study 2019	FMA +0.44
Uganda UGA	wetter	drier	OND	robust	Indeje et al. 2000 (Int. J. Climatol.); Nicholson 2017 (Reviews of Geophysics); Palmer et al. 2023 (Nature Reviews Earth & Environment)	OND+ / JAS−
Ukraine UKR	—	—	—	—	Bronnimann 2007 (Reviews of Geophysics); Zhang et al. 2014 (J. Climate, seasonal ENSO-Europe precip)	JAS −0.34
United States Virgin Islands VIR gap	drier	wetter	ASO	moderate	Jury, Malmgren & Winter 2007 (JGR-Atmos); Mendez-Lazaro et al. 2018 (Int. J. Biometeorol.)	JAS −0.50
Uruguay URY	wetter	drier	OND	robust	Grimm et al. 2000 (J. Climate); Cazes-Boezio et al. 2003 (J. Climate); Mernild et al. (Uruguay studies)	OND +0.65
Uzbekistan UZB gap	wetter	drier	MAM	moderate	Yao et al. 2024 (npj Climate and Atmospheric Science)	MAM +0.50
Vanuatu VUT	drier	wetter	DJF	robust	Vincent et al. 2011 (Clim. Dyn.); Cai et al. 2012 (Nature); PACCSAP Vanuatu country report 2014; Brown et al. 2020 (Nat. Rev. Earth Environ.)	NDJ −0.70
Venezuela VEN	drier	wetter	JJA	moderate	Pulwarty et al. 1992/1998; Cardenas et al. 2002 (Bull. IFEA); Poveda et al. 2006 review	ASO −0.67
Vietnam VNM	drier	wetter	JAS	robust	Dai & Wigley 2000 (GRL)	OND −0.69
Western Sahara ESH	—	—	—	—	No country-specific peer-reviewed ENSO rainfall study	no sig
Yemen YEM gap	split	split	DJF wet / JJA dry	single-study	Atif et al. 2017 (npj Clim Atmos Sci)	FMA+ / JAS−
Zambia ZMB	drier	wetter	DJF	robust	Mason & Goddard 2001 (BAMS)	DJF −0.40
Zimbabwe ZWE	drier	wetter	DJF	robust	Mason & Goddard 2001 (BAMS)	DJF −0.68

Methodology

Data

Rainfall is drawn from ERA5 reanalysis (ECMWF), extracted as country-level area-weighted mean precipitation for 153 countries at admin-0 level, covering 1981–2025. Units are mm/day; a trimester value is the mean of the three constituent monthly means (not a sum). Climate mode indices are monthly anomaly series downloaded from NOAA PSL: Niño3.4 (ERSSTv5 SST anomaly in 5°N–5°S, 120–170°W), IOD (Dipole Mode Index, HadISST), TNA (Tropical North Atlantic SST anomaly), TSA (Tropical South Atlantic SST anomaly), AMM (Atlantic Meridional Mode), and PDO (Pacific Decadal Oscillation, NOAA).

Spatial resolution (Country vs Pixel)

The Resolution toggle selects the unit of analysis; the method below is identical in both cases. Country (ADM0) uses the pre-computed ERA5 admin-0 raster statistics held in the team database — one area-weighted mean series per country, 153 countries. Pixel (0.25°) reads the source ERA5 monthly precipitation COGs directly and runs the whole pipeline independently on every land grid cell in the map viewport (~163,453 cells at 0.25°, ≈28 km at the equator), with no spatial smoothing or pooling between cells.

The pixel view exists because a national mean can hide as much as it shows: countries spanning more than one rainfall regime (Kenya, Ethiopia, Indonesia, Brazil) average opposing signals toward zero, and a signal confined to one basin or one side of a mountain range disappears entirely. It carries two costs. First, each cell is a single ~45-year series, so at p < 0.05 a few percent of cells will pass by chance; because no field-significance or false-discovery correction is applied, isolated coloured cells should be read as noise and only spatially coherent regions as signal. Second, an extra filter is needed that the country pass does not require: a cell–season is analysed only if its climatological mean exceeds 0.25 mm/day, which removes hyper-arid cells (Sahara, Rub' al Khali, Taklamakan interiors) where correlations against near-zero rainfall are numerically large but meaningless. Those cells are drawn in the “arid — no wet season” shade. Ocean cells are not analysed.

On the pixel maps, cells significant in both directions across non-overlapping seasons are hatched rather than split diagonally, and the coloured fill shows the stronger of the two. The pixel dominant-mode map shows the top mode only, and only where it leads the runner-up by at least 0.10 in |r| — one cell does not average enough area to make the arg-max over six collinear modes stable, so near-ties are left grey rather than assigned a winner.

Seasonal aggregation

All 12 rolling 3-month windows are assessed for every country: NDJ, DJF, JFM, FMA, MAM, AMJ, MJJ, JJA, JAS, ASO, SON, OND. Using rolling windows rather than four fixed seasons avoids misaligning the analysis window with the actual rainy season (e.g. a country whose rains peak Oct–Dec is better captured by OND than by JAS or DJF). Year labels use the first month of the window: NDJ and DJF are labeled by November and December respectively (e.g. DJF 2024 = Dec 2024–Feb 2025); all others by the year of their first month.

A country–trimester pair is included only if that trimester's climatological mean rainfall is at least 25% of the country's annual mean, suppressing correlations in dry seasons. Annual mean is computed from the four non-overlapping canonical trimesters (DJF + MAM + JAS + OND), which together cover each calendar month exactly once. Multiple rolling windows can qualify for the same country.

Correlation sweep (total association)

For each country × trimester × index combination, Pearson r is computed across all lags up to the selected maximum (index leading rainfall). The lag with the highest |r| is retained as the "best lag" for that combination. The Max lag toggle controls this cap: 3 months (default) restricts the index to at most one preceding non-overlapping season, which keeps the relationship within the same evolving event and is the more forecast-relevant view; 6 months additionally admits prior-season relationships (e.g. a previous winter's ENSO state predicting the following monsoon), which can have the opposite sign to the concurrent signal. Significance is assessed at p < 0.05 (two-tailed). Results below |r| = 0.3 are treated as no reliable signal and shown in gray; |r| ≥ 0.5 is shown in a darker shade. The best-lag values are frozen after the total pass and reused in the partial pass below.

On each per-index map, a country is shown in a single color if all its significant correlations for that index have the same sign. It is shown as a split diagonal if it has significant correlations in both directions across non-overlapping trimesters (e.g. ENSO drives wet conditions in one season and dry in another). With 12 rolling windows, the non-overlapping check is required to avoid pairing near-identical windows (e.g. JFM and FMA) as spuriously "bidirectional".

Partial correlation (unique signal)

The unique-signal maps show partial correlations: the correlation between an index and rainfall after removing the linear influence of all other indices. This is implemented via the residuals method — for a given country, trimester, and target index, both the rainfall series and the target index series are separately regressed on all other indices (at their own frozen best lags), and the Pearson r of the residuals is taken as the partial correlation. PDO still has its own unique signal computed, but it is excluded from the control set — i.e. it is never regressed out of the other modes — because as a low-frequency mode strongly collinear with ENSO, controlling for it would absorb genuine ENSO signal. Best lags are identical to those from the total pass.

When an index is significant in the unique-signal view but not in the total view, this is a suppressor variable effect — the index has real predictive value that is masked in the raw correlation because it partially cancels shared variance from another index. These are shown in the unique-signal maps. The reverse — total significant, unique not — simply means the signal is largely shared with other modes.

ENSO composites

El Niño and La Niña composites are computed per-country using the concurrent trimester's Niño3.4 value (not a single annual classification). For each country, the trimester with the strongest correlation to Niño3.4 is used as the headline season. Years in which the concurrent Niño3.4 mean ≥ +0.5 are composited as El Niño; years ≤ −0.5 as La Niña. Anomalies are expressed in standard deviations from the full-period climatological mean for that country–trimester. The two maps are independent — ENSO teleconnections are asymmetric and the La Niña composite is not simply the mirror image of El Niño.

Documented-impacts literature review

The “Documented ENSO Impacts (Literature)” section complements the data-driven maps with teleconnections established in the scientific literature, which together are more complete than the generalized NOAA and IRI impact maps. It was assembled with a multi-agent deep-research pass: the question was decomposed into per-region search angles, parallel web searches retrieved candidate sources, and roughly two dozen peer-reviewed articles plus authoritative operational sources (NOAA, IRI, ICPAC/IGAD) were fetched and read. From these, falsifiable directional claims (region × season × phase × sign) were extracted and each was adversarially verified by an independent panel — a claim was retained only if it survived attempts to refute it, and dropped otherwise. This filtering removed plausible-but-unsupported statements; for example, the framing of the Yemen signal as Indian-Ocean-Dipole-mediated was refuted, so it is reported as a direct Pacific teleconnection (Atif et al. 2017). Coverage was then extended in a second region-by-region sweep over the rest of the monitored world (South America, the Caribbean, Central/East Africa, East Asia, the Pacific and the mid-latitudes), so that every one of the 153 monitored countries has been researched.

Verified regional findings were combined with canonical global ENSO-rainfall climatologies (Dai & Wigley 2000; Mason & Goddard 2001; Lenssen, Goddard & Mason 2020) into the per-link table. Each row carries a citation and an evidence-strength flag (robust and widely replicated vs. single-study or contested), and a “gap vs NOAA/IRI” tag where a well-supported link is absent or under-represented on the standard maps. Every documented link is cross-referenced against this study's own ERA5 correlation result (agree / partial / weak) as an independent check.

The map colors each country by its dominant documented El Niño response, labelled with the relevant 3-month season; countries that respond in opposite directions across seasons are shown split. Grey means a monitored country was researched but has no robust documented ENSO rainfall link; near-white means the country is not monitored and was not researched. The La Niña map is the mechanical inverse of El Niño — a first-order picture only, since ENSO is asymmetric. Evidence strength is uneven (robust in well-studied regions; single-study or contested elsewhere — see the per-country catalogue), and the “gap vs NOAA/IRI” tags are editorial judgements.

Caveats

Pearson r assumes a linear, stationary relationship. Teleconnections can be non-linear and non-stationary; correlations over 1981–2025 may not represent future conditions.
ERA5 reanalysis is model-dependent. In data-sparse regions the precipitation field is more influenced by the model background than by observations.
Multiple-testing is not corrected for (no FDR/Bonferroni). With ~600 country–trimester combinations per index, some false positives are expected at p < 0.05.
Partial correlations do not partition the total explained variance cleanly when predictors are collinear (e.g. AMM and TNA share r ≈ 0.81). Unique-signal maps should be read as indicating whether an index contributes beyond the others, not as a clean attribution.
Coverage is limited to countries where ERA5 admin-0 extracts are available (153 countries). Small island states are shown as dots rather than choropleth polygons.