A Dry Year You Cannot Pin on Anyone

The Mekong rises on the Tibetan plateau, runs through Yunnan as the Lancang, and then crosses or borders Myanmar, Laos, Thailand, Cambodia and Vietnam before reaching the sea. Tens of millions of people eat from it. The Tonle Sap in Cambodia is one of the most productive inland fisheries on earth, and it works because the river reverses into the lake each wet season and floods it.

The institution is the Mekong River Commission, created by a 1995 agreement among Cambodia, Laos, Thailand and Vietnam. It does real work: monitoring, modelling, notification procedures, basin planning. It has one structural problem, which is not fixable from inside.

China is not a member. Neither is Myanmar. Both were invited to join and both remained dialogue partners instead, which means the commission's procedures bind the four countries downstream of the dams and not the country holding them. China has built a cascade of large hydropower dams on the Lancang, several of them with major storage, and those reservoirs can shift the river's timing by holding water in the wet season and releasing it in the dry.

Then came 2019, when the lower basin fell to levels variously described as the lowest in decades or in living memory. The Tonle Sap's reverse flow was weak and late. Fisheries collapsed for a season.

The argument about why has never been resolved, and following it carefully teaches more than picking a side.

In April 2020 a study using satellite data, funded from the United States, reported that Yunnan had received above-average water during the period when the lower basin was parched, and that upstream reservoirs appeared to be holding it. China's embassy rejected the method, arguing that the work inferred flows rather than measuring them and ignored rainfall patterns, and called for independent review. The commission's own analysis emphasised exceptionally low rainfall and El Nino conditions, while also noting it lacked full data on upstream operations.

Notice what is being contested. Not whether the river was low, which everybody agreed on. Whether the dams caused it.

That question requires the inflow records, the reservoir levels and the operating decisions of the upstream dams, and those are held by the operator. Downstream states had satellites, models and inference. In a dispute where the harm is a dry season rather than a diverted canal, the party with the structures also holds the evidence about the structures, and in the absence of evidence the loss stays with whoever suffered it.

So the achievable demand turned out not to be water.

In August 2020, at a summit of the Lancang-Mekong Cooperation, the Chinese-led forum established in 2016 alongside the commission, China announced it would share year-round hydrological data rather than wet-season figures alone. Transmission from two stations on the Lancang began that November, with a data platform following. This is a genuine change: flood warnings and dry-season planning both improve when the figures arrive before the water does.

It is also a long way short of a limit on operations. Nobody agreed to release water at any particular time, and the commission's flow rules still do not reach upstream of the border.

The transferable lesson is about evidence rather than rivers.

Where a harm is diffuse, delayed and mediated by weather, the first fight worth having is over disclosure, because every later fight depends on it. That is as true of groundwater depletion, fishery collapse and emissions as it is of the Mekong. The party causing a diffuse harm benefits from uncertainty by default, which means data is not a consolation prize for failing to win the real argument. It is the precondition for ever being able to make it.