Cemaden places the Negro and Xingu in moderate drought on its longer-duration indicators and points to subsurface reserves that never recharged. The Brazilian Geological Service, on 16 September, has most Amazon gauges in ordinary seasonal recession, with the Rio Branco the exception. Both readings can be true, because river stage and water storage are different things, and the next rainy season may still begin with parts of the basin carrying years of accumulated deficit.
Brazil's disaster monitoring centre, Cemaden, briefed on 15 September that two major Amazon rivers, the Negro and the Xingu, are already in moderate drought on its longer-duration indicators. El Niño has not reached its expected peak.
The reason given is the part worth reading twice. Adriana Cuartas, a Cemaden researcher, put it this way: "não estamos abastecendo as reservas subterrâneas, que são as que mantêm os rios durante as secas." We are not recharging the groundwater reserves, which are what sustain the rivers through the dry season. And: "É muita escassez que está se acumulando ao longo do tempo." A great deal of scarcity accumulating over time.
That is a different claim from "it has not rained enough this year." It says parts of the basin are carrying a debt.
The warnings are not uniform across monitoring systems. Cemaden's longer-duration indicators place the Negro and Xingu in moderate drought and point to accumulated subsurface deficits. But on 16 September, the Brazilian Geological Service reported that most Amazon gauges remained within their expected ranges for the season, with the Rio Branco basin the main exception. River stage can be seasonally normal while longer-term water storage remains depleted. The evidence therefore describes uneven vulnerability across parts of the basin, not an already abnormal basin-wide low-water event.
The Xingu is the sharper signal. Cemaden reports its August 2026 discharge close to August 2023, which is the month before the last El Niño produced a historic Amazon drought. Cuartas expects the trend in the region to worsen over the following three months.
The Negro is the counterweight, and it is why the headline says "parts." The Port of Manaus gauge had the river at 24.09 m on 1 September and 21.62 m on 16 September, falling 10 to 19 cm a day. SGB's 16 September bulletin reads that recession as ordinary for the date, listing the Solímões, Madeira, Amazonas main stem and Tapajós in the same category, and noting the contrast with 2023 and 2024 when many rivers fell together.
The exception SGB does flag is the Rio Branco in Roraima, where it says levels are "abaixo da média e das mínimas históricas registradas para esta época do ano," below both the average and the historical minimums for this time of year, with Boa Vista at 0.93 m and Caracaraí at 1.47 m.
For scale on the Negro, the record low is 12.11 m on 9 October 2024, the lowest since the Manaus series began in 1902. That broke 12.70 m from 2023, which had broken 13.63 m from 2010. SGB's early-September scenarios put the Negro somewhere between 12.31 m and 16.50 m at the end of October 2026. That is a wide scenario range rather than a probabilistic forecast, and its lower bound sits within 20 cm of the all-time low.
Rain raises a river quickly. It refills soil moisture and groundwater slowly. Between rainy seasons, what keeps a river flowing is baseflow, the slow drainage of water stored underground. If successive wet seasons fail to rebuild that storage, each dry season starts from a lower base, and the same rainfall deficit produces a lower river.
There is now a measured version of this, though it comes from a different part of the basin. De la Cruz and colleagues (2025, Journal of Hydrology: Regional Studies) analysed western Amazonia at the Tamshiyacu station, which drains roughly 16 percent of Amazon discharge. Across 1981 to 2024 they find the high-runoff season starting 12 days later per decade and low-flow season discharge falling 116.3 m³/s per year. During the 2022 to 2024 event, terrestrial water storage anomalies reached their lowest level on record, below −15 cm, and the drought ran a record 24 months against 16 months for the 2010 event.
That station is on the Marañón and Ucayali system in Peru, not the Negro or the Xingu, and the numbers should not be transplanted onto rivers they did not measure. What the paper establishes is that the mechanism Cuartas describes is real and measurable across a major western Amazon sub-basin: storage depletion that outlasts the rainfall deficit that caused it.
On 26 August, IBAMA authorised the Belo Monte complex to reduce the minimum flow in its intermediate reservoir from 300 m³/s to 100 m³/s through 30 November 2026. The stated reasoning was to let the plant preserve both electricity generation and flows through the Volta Grande do Xingu, the river bend where Indigenous and riverine communities depend on the water for fishing, transport and daily life. Norte Energia says the required flow through the Volta Grande itself is unchanged and that it will monitor water quality and fish movements.
The objection raised is not about generation. A reservoir moving less water is a reservoir whose water quality degrades faster, with consequences downstream for aquatic life and for people. One assessment quoted in coverage of the decision: between a worse option and a bad one, they chose the bad one.
The point for this site is narrower. A regulator agreeing to a two-thirds cut in a minimum flow requirement, three months before the wet season, is operational evidence that regulators and the operator are planning around anticipated water scarcity.
The instinct with a strengthening El Niño is to ask how strong it gets. That is a real question, and it is not the whole of one.
CPC's 10 September discussion puts the chance of a very strong event in October to December, a three-month RONI of +2.0°C or more, at 98 percent, and a historic event at +2.5°C or more at 75 percent. Weekly Niño 3.4 on CPC's relative weekly table, rel_wksst9120.txt, reached +2.0°C in the week centred 9 September. That is a single week of the relative Niño 3.4 index and is not the same quantity as the three-month RONI the probabilities above are stated on. Those numbers describe the Pacific.
El Niño strength shifts the probability of rainfall anomalies over the basin. What converts a rainfall anomaly into a hydrological outcome is the timing, placement and volume of the rain that actually falls, and whether it is enough to recharge storage rather than merely lifting river stage for a few weeks. Both matter, and the Pacific number alone will not tell you which way this goes.
A very strong El Niño does not guarantee a record Amazon drought, and the most important caveat runs against the headline of this piece rather than for it.
World Weather Attribution examined the 2023 Amazon drought and separated the two questions that usually get merged. On rainfall, they found El Niño and climate change contributed comparably to the deficit. On severity, they found human-caused warming, and the increased evapotranspiration that came with it, was the dominant reason the agricultural drought reached exceptional rather than merely severe intensity. They put the meteorological drought at roughly 10 times more likely and the agricultural drought at about 30 times more likely because of warming, and concluded that without climate change the event would have been classed severe rather than exceptional.
Read that against the title. El Niño was a genuine contributor to the rainfall shortfall in 2023, so this is not a case of the Pacific being irrelevant. What it was not was the reason that shortfall turned into an exceptional drought; the heat did that. The Amazon is therefore entering 2026-27 depleted by a combination in which warming did the heavier work, and an El Niño arriving on top of that is an additional stress rather than a repeat of a single cause.
What is still unsettled: Atlantic sea surface temperatures, where the rain actually falls within the basin, land surface condition, and when the atmospheric response couples. Reuters reported scientists expecting the severity to become clearer only by the end of the year, and depending partly on the Atlantic.
The other half of the Brazilian picture is the opposite problem. In the south, operators are preparing for too much water. July rainfall in southern hydroelectric basins ran at 256 percent of the long-term average, lifting reservoirs to 88 percent of capacity from 30 percent in April. China Three Gorges moved its flood risk protocols forward from September to June, running spillway tests, buying flood barriers and powerhouse equipment, and setting emergency protocols for communities downstream of its 14 plants. Engie Brasil has built contingency plans around enhanced weather monitoring.
This is the textbook South American El Niño signature: wet in the subtropics, dry in the north. It is also why national-level statements about Brazil and El Niño tend to be useless. The two halves need opposite preparations.
The pricing has arrived ahead of the drought. Container lines including Hapag-Lloyd, Maersk, Aliança, Log-In, Ocean Network Express, CMA CGM and Mercosul Line announced low-water surcharges on Manaus traffic, citing 2026 drought logistics. The quoted amounts are not on a common unit and should not be collapsed into a single range: CMA CGM's lowest figure is US$753 per TEU, while Maersk's upper figure is US$4,646 per container. These are proposed charges, and as of 16 September they are suspended, so they are a measure of what carriers expect rather than a cost already borne.
ANTAQ, the waterway transport regulator, has pushed back. Deliberation-DG nº 70 of 13 September now requires 30 days' advance notice and technical, operational and economic justification before such a surcharge can be charged. On 16 September the agency rejected the carriers' appeals, with director Caio Farias's position endorsed unanimously, keeping the surcharge suspended for container operations at Manaus.
Buried in that regulation is the most useful number in this article, and it needs stating precisely. ANTAQ's ordinary route allows a surcharge to take effect by tacit homologation once its 30-day decision window expires without a ruling. The 17.7 m official Rio Negro level at the Port of Manaus is the 2026 reference that can permit early tacit homologation, before that window closes, provided the notice period, the technical, operational and economic documentation and the regulation's other requirements are satisfied. It is not a threshold every approval must clear, not a navigation cutoff, and not a drought classification. It is a gauge reading that can accelerate one regulatory route.