Counting raindrops using mobile-phone towers

NO ONE knows exactly how many people died in a series of mudslides that happened in and near Freetown, the capital of Sierra Leone, on August 14th. The upper estimates are more than a thousand. The areas swept away had not been evacuated partly because no one knew how much rain had actually fallen beforehand, laments Modeste Kacou, a rainfall expert at Félix Houphouët-Boigny University in Abidjan, in nearby Ivory Coast. Rain gauges are sparse in Sierra Leone. Satellites detect rainfall in the tropics, but estimates for small areas are often inaccurate. Worse, these numbers are calculated hours after the fact. Many countries therefore use cloud-scanning ground radar to measure precipitation as it is happening, but Sierra Leone has no such radar.

Nor do many other poor countries. Ivory Coast has double the GDP per person of Sierra Leone, but like most of west Africa, it also lacks precipitation radar. Indeed, maintenance costs mean that the number of weather stations around the entire world is shrinking, making it harder to forecast flash floods and landslides even in some rich countries. It would be useful, therefore, if some other way of measuring rainfall—preferably a cheap one that employs existing, widespread equipment—could be devised. Fortunately, there is just such a method, and it involves mobile-phone networks.

The basic insight is straightforward enough: rain weakens electromagnetic signals. Many mobile-phone towers, especially in remote areas, use microwaves to communicate with other towers on the network. A dip in the strength of those microwaves could therefore reveal the presence of rain. The technique is not as accurate as rooftop rain gauges. But, as Dr Kacou points out, transmission towers are far more numerous, they report their data automatically and they cost meteorologists nothing. He runs the Ivory Coast operations of Rain Cell Africa, an effort paid for by the World Bank, the UN Foundation, a charity, and the Institute for Development Research, which is based in France, to map rainfall in parts of Africa using data donated by Orange, a big telecoms firm, and Telecel Faso of Burkina Faso, a small one. Had the system been running in Sierra Leone, he reckons evacuations could have been carried out in time.

Rich countries are interested, too. A pilot project in the Netherlands a few years ago produced promising results, but it has not yet been followed up. This month officials in Gothenburg, Sweden, began to study rainfall maps derived from data collected every ten seconds from 418 mobile-phone towers owned by a firm called Hi3G. The hope is this will provide more accurate estimates of rainwater about to slosh into the municipal waterworks, helping managers to limit flooding and sewage overflows. Until now, the city has relied on 13 rain gauges, backed up by radar sweeps of the sky that are neither sufficiently frequent nor sufficiently precise, says Jafet Andersson, a hydrologist behind the scheme at the Swedish Meteorological and Hydrological Institute. Satellite data on rain in northern latitudes are so poor the agency does not bother using them at all.

Though it is useful to know how much rain is falling right now, forecasting is even better. Telecoms data promise to make this easier as well. Some newer networks are sufficiently sensitive that they can detect humidity and fog, both of which are predictors of imminent rain. Newer generations of mobile-phone masts use shorter wavelengths in their transmissions, because these can carry more data. Serendipitously, that also permits tinier amounts of water to be detected, for moisture weakens short wavelengths more than long ones. Using data from about 5,000 towers operated by three telecoms firms in Israel, Pinhas Alpert of Tel Aviv University creates moisture maps that, he says, are far more precise than those drawn with data from the Israel Meteorological Service’s humidity gauges, of which there are fewer than 70.

The right wavelength

Moreover, because transmission towers are so common, predicting where rainclouds are being pushed by winds is easy and accurate, notes Dr Andersson. Several governments in Africa and South-East Asia have asked his team to set up rainfall-measurement networks for them. No deals have yet been signed, though, for there is a stumbling block: money.

For the time being, telecoms companies are happy to let forecasters use their data free of charge. As the value of such data becomes clearer, says Frédéric Cazenave of the Institute for Development Research, that is likely to change. Consider ClimaCell, a firm based in Boston, Massachusetts. In April ClimaCell began selling forecasts based on phone-tower data which are so precise, it claims, that its customers will be able to tell “if a plane, crane or game” will get soaked.

The firm’s clients include three airlines, several sports leagues, a construction company, a drone operator and a hedge fund (which uses weather forecasts to make trades). It plans to offer its forecasts in India by the end of this year, and to expand into ten more countries in 2018. If ClimaCell pulls that off profitably, telecoms firms in both the rich and poor world are likely to start demanding a slice of the action.