INSIGHT16 November 2026

El Niño at site level: what changes between a neutral year and an El Niño year

El Niño is back on earnings calls, and boards ask which sites it touches. At site level, it can push one company's sites in opposite directions: in January, one Andean site gets 30% drier while a site in central Mexico gets 56% wetter.

R
Laura Alejandra Sánchez GuarnizoRepath
R
Leidy Gicela Vergara LópezRepath

Companies now mention El Niño in earnings calls and statements three times as often as at the 2023 peak, according to AllianceBernstein, as reported by Responsible Investor. Once El Niño is on the earnings call, the board asks which sites it touches and by how much. A global outlook can’t answer that for a single plant. So we compared neutral years with El Niño years at the sites of one industrial portfolio, using the same asset-level logic as any infrastructure climate risk assessment. At a site in the northern Andes, January in an El Niño year is about 30% drier and 0.7 °C warmer than in a neutral year. At a site in central Mexico, the same month gets 56% wetter.

Key takeaways

  1. El Niño can push one company’s sites in opposite directions. In January, a site in the northern Andes gets about 30% drier, while a site in central Mexico gets 56% wetter.
  2. Global data can get a single site badly wrong. At the Andean site, a global reanalysis puts rainfall at about 8 times the gauge-based value.
  3. A portfolio view shows where the exposure sits. El Niño shifts risk between sites rather than raising it everywhere.

One site, side by side

The site sits high in the northern Andes. It has the clearest El Niño signal in the portfolio, and most of it falls in December to February, when El Niño peaks and its effect on the northern Andes is strongest. The table compares January in the average neutral year with January in the average El Niño year.

Indicator (January) Neutral year El Niño year Change Signal strength
Mean temperature 13.9 °C 14.6 °C +0.7 °C strong
Daily maximum temperature 21.1 °C 22.1 °C +1.0 °C moderate
Total rain 120 mm 83 mm -30% moderate
Rain in the wettest 5 days 71 mm 41 mm -43% moderate
Heavy-rain days (≥ 20 mm) 2.0 1.0 -1 day moderate

Averages over past neutral and El Niño years. All five indicators pass our significance test. Source: Repath analysis.

In past El Niño years this happened reliably: less rain, fewer heavy-rain days and warmer days. In a strong El Niño year, January rain drops to 66 mm, 45% below a neutral year, and the same direction holds in December and February. Afternoons run about 1 °C warmer, which matters little at a daily maximum of around 22 °C. The story at this site is the dry side. Wind shows no detectable El Niño change here.

Same company, opposite directions

In the same month the Andean site gets drier, a site in central Mexico gets 56% wetter, and both changes pass our significance test. The other sites in the portfolio show no detectable January effect at all.

For the company, El Niño moves risk from one site to another. An average across the portfolio would hide both the dry site and the wet one. A global El Niño map shows the regional direction of change, but it can’t tell you which of your sites sits on which side, or how strongly.

Why global data can mislead at a single site

Global reanalysis data such as ERA5 works on a grid of roughly 30 km. In the Andes, a grid cell that size can’t resolve the mountains around the site, and at our Andean site ERA5 puts rainfall at about 8 times the gauge-based value. For rain at this site we switched to CHIRPS, a dataset that combines rain gauges with satellite data.

A site-level number is only as good as the data behind it. Choosing the right dataset for each site is part of the analysis, not a detail.

What operations and finance do with it

Operations. For December to February in an El Niño year, the Andean site should plan for less rain and fewer heavy-rain days. That means securing water storage and process-water supply, planning dust control, and expecting more outdoor work windows.

Finance. El Niño exposure isn’t spread evenly across the portfolio. Here it concentrates at the Andean site, which gets drier and warmer, and partly at the Mexican site, which gets wetter. The other sites show no detectable winter effect. Ranking sites by their El Niño signal shows where preparation and budget belong, and where nothing needs to change.

How we measured it

We classified each year by NOAA’s Relative Oceanic Niño Index for December to February. El Niño years are +0.5 °C or more, strong El Niño years +1.5 °C or more, and neutral years lie between -0.5 and +0.5 °C. La Niña years are left out. Over 1950-2025, that gives 27 El Niño years, 7 of them strong, and 26 neutral years.

Temperature comes from ERA5 reanalysis for 1950-2025, and rain from CHIRPS for 1981-2025. Each value is the average over past years of each kind. Signal strength shows how consistently an indicator has followed El Niño’s strength in past years, measured as a correlation, and every figure we show passes a strict significance test across all sites and indicators. It describes a historical relationship, not a forecast or a probability.

The comparison doesn’t cover hourly rain, hail, lightning, landslides or floods, and it doesn’t attribute any single event to El Niño. More on how we model each site on our science page.

Frequently asked questions

How does El Niño affect a single industrial site?

It depends on the site. At one site in the northern Andes, January in an El Niño year is about 30% drier and 0.7 °C warmer than in a neutral year, while a site in central Mexico gets 56% wetter in the same month. Only a site-level comparison of neutral and El Niño years shows which way each site moves.

What should operations prepare for in an El Niño year?

At sites that get drier, plan for less rain and fewer heavy-rain days in the peak months: secure water storage and process-water supply, plan dust control and use the extra outdoor work windows. Sites that get wetter need their own numbers.

Is global climate data accurate enough for site-level El Niño analysis?

Not always. In mountainous terrain, a global grid of roughly 30 km can’t resolve the local relief. At one Andean site, ERA5 put rainfall at about 8 times the gauge-based value, so we used CHIRPS for rain there.

Find out for your sites

Book an intro call with us - and within 48 hours of our conversation, you’ll see how a neutral year and an El Niño year compare at your sites.

The climate data your financial models are missing.

Get climate intelligence on your portfolio - in 48 hours.

Book a Call