← ENSO deep dives

El Niño 2026/27

Venezuela: El Niño 2026/27

What past El Niño and La Niña years brought, how this year has gone so far, and the forecast for the coming months: rain, floods, the Guri reservoir, vegetation and harvests.
Rain to September 2026 (CHIRPS) and September 2026 (ERA5) · vegetation to 11–20 October 2026 · SEAS5 issued October 2026

Versión en español Slides (PowerPoint)

For this season, plan on the seasonal forecast (section 3: ECMWF SEAS5, a new one each month), not on past El Niño years: those show what has happened before, as context, not what will happen this year.

1. What past El Niño years brought

By INAMEH rainy-season zone: El Niño cuts dry-season rain in the south (zones A and B, 21–33% less) and rainy-season rain in the north (zones C to E, 8–19% less)

Medians over the 15 El Niño years; in zones A and B, 14 or more of them had a below-normal dry season. The south's rainy season barely changes (within ±1%), and the north's dry season loses 8–15%. Zone A is never truly dry: Dec–Mar averages 128 mm a month.

As FEWS NET maps it: in El Niño years the season was in the driest third at least half the time across 73% of the mapped area in December–February, about 28% from June to November, and almost nowhere in March–May

Brown: how often the season fell in the driest third of 1991–2020 in the 15 El Niño years since 1981; green: the wettest third; white: neither reaches 40% (a third would be chance); hatched dry-season areas are left out. FEWS NET's maps, in the second figure, use the same scale for strong and very strong El Niño events since 1950. The small wetter-looking patch inland on the global survey map is Apure in February–April after El Niño (the green patch in the March–May map), a minor effect; from June to January it is normal or drier there too.

By zone: in El Niño years every zone is 21–36% drier from November to March; the northern coast (zone E) also loses 16–23% from June to November

Each cell is the median over the El Niño (or La Niña) years for that three-month window. The zone dry seasons above (December to the onset) lose less in the north because they run into April–May, when El Niño makes little difference.

El Niño years have had far fewer people affected by reported floods, landslides and storms than La Niña years: about 12,000 a year against 153,000

Neutral years: about 9,000. Without Vargas 1999 and August 2021, La Niña years still had 2.3 times the El Niño figure. El Niño years are not flood-free: June 2015 in Apure and February 2005 came in El Niño years.

By region: 30–64% fewer flood and landslide reports in El Niño years everywhere; more in La Niña years in 5 of 7 regions

In a typical El Niño year each region had 0.36–0.70 times the reports of an ordinary year. La Niña medians in the western Llanos and the centre-north: 0.75 and 0.96. Means are pulled up by single years (1999, 2011): 1.4 for La Niña nationally.

Vegetation: the three poorest dry seasons since 2000 (2002/03, 2009/10, 2015/16) all followed El Niño; 2023/24 did not

The stronger the El Niño, the poorer the dry season (correlation −0.57), with 2023/24 the exception. Averaged over the El Niño years, cropland in November–April was below normal in 21 of 23 states. CERF has made no drought allocation to Venezuela.

Harvests: no clear relation with El Niño, La Niña or rain

In all three panels the points scatter without a pattern: every correlation is between −0.3 and +0.3, and other ways of removing the trend and the crisis give the same answer (maize −0.26 to −0.12, rice −0.16 to +0.07). Maize fell 10% in 2002 and 14% in 2009, both El Niño years, but not in the strong El Niño years 1982, 1997 or 2023.

2. This year so far: rain, vegetation and the Guri reservoir

May–September 2026 rain was 65–89% of normal, among the five driest of all 46 years since 1981 (not just El Niño years) in every region

ERA5 puts May–September at 57–81% of normal. CHIRPS, which uses gauges, leads here; ERA5 is a reanalysis that does not use gauge rain. October data come out in November.

Vegetation in mid-October: near normal nationally, but cropland below ASAP's warning level in Sucre, Zulia and Falcón

Cropland below normal in 15 of 23 states (11–20 October 2026). FEWS NET's August briefing also reported localized soil-moisture deficits in rainfed areas.

The Guri reservoir starts this dry season near full (270.2 m); falls like those after past El Niño years would take it to 250–261 m by May 2027

After a typical El Niño year it has fallen 14 m through the dry season, against 5 m after La Niña. The low end of that range is 2 m above the 2010 low (249 m), the year of power rationing; the peak before the 2016 low was 262 m. How far it falls also depends on how much water is released for power, not only on rain.

3. The forecast for the coming months

SEAS5, October 2026: one of the driest October–February forecasts since 1981, everywhere it is the rainy season

All 45 state-seasons from October to February that fall in the rainy season are strongly below normal (the driest 10% of the 45 past forecasts) once the long-term trend is removed, as in the team's seas5-skill viewer; before that step, it is the driest November–January forecast in 12 of 24 states. The forecast has skill for these months (hindcast correlation 0.4–0.7). The October forecasts in 1997 and 2015 ranked about as dry; the rain that followed was among the driest on record in 1997 and below normal in 2015. February–April leans wet in the west and centre, but there the forecast has low skill or the months fall outside the rainy season, so the map masks it.

FEWS NET's seasonal calendar: the forecast months cover the summer planting of maize and rice and the coffee and sugar-cane harvest; the lean season is June–August

Redrawn from FEWS NET's Venezuela page. FEWS NET's El Niño briefing of August 2026: below-average rain and heat caused localized soil-moisture deficits in some rainfed areas; crops were mostly well established thanks to earlier rain and irrigation, with localized impacts on small producers but no major losses. For October 2026 to January 2027 it expects localized rainfed losses with limited impact, given irrigation and imports, and Stressed (IPC Phase 2) outcomes. It puts Venezuela's peak needs in August–October 2026, with the earthquake and the economy, not weather, as the key shocks.

About the data

• Rain years and phases. A rain year runs from May to April: nationally, the May–November rains, then the December–April dry season. It is classed by the Sep–Nov ONI of its first year (NOAA CPC): El Niño ≥ +0.5 °C, strong El Niño ≥ +1.5 °C, La Niña ≤ −0.5 °C. Since 1981: 15 El Niño years (1982, 1997, 2015, 2023 strong) and 15 La Niña years. Month axes run January to December: for El Niño and La Niña, months and windows starting January–April come from the year after the event develops, the rest from the year it develops. Sep–Nov is used because Venezuela's El Niño effect starts in the developing year; by its peak ONI (+1.5 °C in Nov–Jan) 2009/10, the year of the 2010 power rationing, would count as strong.
• Rain. CHIRPS v3 (satellite and gauges) at 0.5°, averaged over each state with area weights, to September 2026. ERA5 by state (team admin-1 statistics) to September 2026 is used as a check: it agrees with CHIRPS on the direction of the El Niño change in 94% of the 116 state-seasons where CHIRPS shows ±10% or more. Percentages are against the 1991–2020 normal for the same months; three-month windows with a normal under 60 mm are dry season and left out.
• Zones and seasons. The zones are INAMEH's map of rainy-season onset (five zones, from early April in the far south to late June on the northern coast), digitized onto the 0.5° grid, so their edges are approximate. Each zone's rainy season runs from its onset month to November, the end FEWS NET uses; its dry season from December until the onset.
• FEWS NET. The seasonal calendar is redrawn from FEWS NET's Venezuela page (rural areas), digitized so the Spanish page can carry it translated (deep_dives/data/fewsnet/ve_seasonal_calendar_rural.csv); the strong-El-Niño maps are FEWS NET's, reproduced with credit from pages 4–6 of its August 2026 El Niño briefing.
• Floods. FloodScan, the satellite flood-exposure record used on other pages, covers Africa only, so this page has no flood-exposure series for Venezuela. EM-DAT (CRED/UCLouvain, snapshot of 2025-08-27) lists the larger flood, wet-landslide and storm disasters since 1980; one Location field often names many states, so it is used nationally. DesInventar Venezuela (UNDRR) has 4,970 flood, rain, torrent, debris-flow, landslide and storm records by municipality; it stops in early 2016, and its reporting changes several-fold between decades, so each year is compared with the six rain years around it, and the figure shows the median over the years of each phase. CERF allocations: CERF allocation data (cerf.un.org; Venezuela only, not the regional allocations).
• Guri. NASA GSFC G-REALM satellite altimetry, every ten days since September 1992, in metres above the EGM2008 geoid. This is not the operator's gauge datum: compare falls and past lows within this record, not with published operating levels.
• Vegetation. JRC ASAP: z-score of MODIS FPAR (the share of sunlight the canopy absorbs) against the same ten days in 2001–present, averaged over cropland or rangeland in each state (GAUL admin-1), November–April.
• Harvests. FAOSTAT national yields, each year against the two years either side. From 2014 to 2019 the economic crisis drives the numbers and hides any weather signal; those years, and the years compared with them (2012–2013, 2020–2021), are left out of the correlations.
• Forecast. ECMWF SEAS5 ensemble mean by state (team admin-1 statistics; the database holds the ensemble mean only), October issuance. Three-month means are log-transformed, scaled to ERA5 and detrended, as in the team's seas5-skill app; the rank is among the 1981–2025 hindcasts for the same months, and skill is their correlation with ERA5. On the forecast maps, classes, skill levels and the rainy-season mask follow the seas5-skill viewer (outside the rainy season: the three months bring less than 15% of the year's ERA5 rain), not the 60 mm rule used for the rain composites.

OCHA Centre for Humanitarian Data · built 2026-11-02