When the two oceans align: compound climate risk in coffee and the monsoon

We have argued that ENSO and the Indian Ocean Dipole are two separate switches, correlated but distinct. The interesting question is what happens when they line up. Aligned, their rainfall effects can reinforce — a positive dipole stacked on El Niño turned East Africa's 1997 short rains to roughly twice normal. But the same alignment can cancel elsewhere: that very pairing is why the 1997 Indian monsoon survived a record El Niño. Same two oceans, opposite outcome — because the answer depends entirely on the region. Here is the compound signal, and an honest account of how rare and how noisy it really is.

Reinforcement: when both push the same way Amplify

Start with East Africa, because it is the cleanest case and the one where keeping a separate dipole index pays off most. Over the Ethiopian, Kenyan and Tanzanian highlands, the October–December "short rains" are governed primarily by the dipole — more than by ENSO, whose influence there arrives largely second-hand through the dipole. And the dipole's leverage on that season is enormous.

Now stack ENSO on top. Because El Niño also wets East Africa, a positive dipole co-occurring with El Niño is two drivers pushing the same direction — and the result is not additive but nonlinearly amplified. The 1997 pairing of a record El Niño with a strong positive dipole drove rainfall to around 200% of the climatological mean and devastating floods. The mirror image is just as real: La Niña stacked on a negative dipole is double-dry, the compound setup behind the multi-season East African droughts that recur in the record. Same-sign alignment is the compound-risk case, and it cuts both ways.

The same alignment, the opposite meaning Cancel

Here is the subtlety that makes a single "compound score" impossible, and it is worth slowing down for. Take the Indian summer monsoon. ENSO and the dipole push it in opposite directions: El Niño tends to suppress the monsoon, while a positive dipole tends to support it. So the exact same ocean state — El Niño plus a positive dipole — that reinforces a wet anomaly over East Africa instead offsets over India.

1997 is the textbook case. A record El Niño should have produced a major monsoon failure. It did not — India's monsoon came in near normal, widely attributed to the strong positive dipole running at the same time and buffering the Pacific's drag. A monsoon watcher reading only ENSO would have braced for a drought that the dipole quietly cancelled. This is the counterweight you intuited — but notice it is not "the two oceans cancel." It is that their rainfall fingerprints over a given region can have opposite signs, so whether an alignment reinforces or offsets is a property of the region, not of the oceans. That is precisely why the compound read has to be built region by region, with each region's own ENSO and dipole signs — and why you cannot collapse it to one number.

The matrix — and its honest limits The Grid

Lay it out as a grid and the structure becomes legible. Below is the compound picture for the East African short rains: the three ENSO states down the side, the three dipole states across the top, each cell the typical rainfall outcome — with the number of years that combination has actually occurred since 1982, because a matrix is only as trustworthy as its sample.

The grid is honest about its own weak spots, and that honesty is the point. The diagonal is robust: the reinforced wet and dry corners are each backed by five-to-eight observed years and a clear physical mechanism. But look at the off-diagonal corners — El Niño with a negative dipole, and La Niña with a positive dipole, have simply not happened in forty-four years. The positive correlation between the two oceans means the strongly-opposing combinations are vanishingly rare, so any "what if they fought each other" cell is a mechanism guess, not a measured outcome. We label those cells rather than invent a number for them. A compound product that pretends every cell is equally trustworthy is selling false precision; the defensible version is a strong diagonal, a hedged middle, and an explicit "not observed" where the record is silent.

How often does this actually happen? Frequency

This is the question that decides whether the compound signal deserves your attention every month or only some of the time — and the data gives a clear, deflating, useful answer. Of the last forty-four autumns, a genuine same-sign alignment — both oceans active and pushing the same way — occurred in just eleven. The rest were ENSO acting largely alone (19), the dipole acting alone (6, including 2019), or a quiet ocean (8).

That is exactly why we do not publish a standalone compound report on a monthly calendar — most months, the honest reading is "no alignment, nothing compound to say," and a report obligated to fill that space would manufacture signal. Instead the compound read lives as a conditional layer inside the ENSO and dipole reports: silent when the oceans are uncoordinated, and prominent — with the matrix and the regional translation — precisely when an alignment loads up over a region that matters. The rarity is not a weakness to paper over. It is the reason the signal is worth flagging loudly on the handful of occasions it is real.

A closing caution we hold ourselves to: the East African case is the strongest and best-sampled; the monsoon case is canonical but its offset statistics deserve their own dedicated treatment before we lean on them hard; and the Southeast Asian robusta belt — Vietnam especially — rides on the general maritime-continent mechanism rather than a region-specific study, so we treat it as inferred, not measured. The framework is real. Its confidence is not uniform, and we would rather mark the soft spots than hide them.