Every drop counts

Have you ever paid attention to the pitter patter of raindrops against your umbrella and noticed just how different it can sound from one moment the next? During summer rainstorms, the drops strike heavily, each one making a separate splash. During a spring drizzle, the droplets can hardly be distinguished and all you hear is a constant hum. And if you look through your umbrella, you’ll notice that at any given time, the drops fall in a wide range of sizes. There are more small ones than large ones, and they don’t all seem to fall at the same speed.

Rain isn’t simply rain. Many complicated microphysical processes are involved in determining the final size of each raindrop, from its birth in a cloud to its final demise upon the ground. Professor Alexis Berne and PhD students in his lab have spent four years measuring the size of raindrops. “When we tell people that we count raindrops, it often makes them laugh. They probably wonder, ‘Don’t you have anything better to do?’” says Berne.

But Berne is quick to point out that information on raindrop size distributions is actually quite useful, particularly for weather services. Weather prediction models rely on all kinds of data to make weather forecasts as accurate as possible. Weather radars, for example, are used to monitor huge swaths of the atmosphere for precipitation. “What meteorologists and hydrologists are interested in,” says Berne, “is the intensity of the precipitation. But rather than giving us a direct measurement of intensity, radar measurements tell us how strongly the emitted radio waves are reflected by raindrops in the atmosphere.” Using an empirically determined power law, this reflectivity can be used to estimate the rainfall intensity.

But to today’s weather radars, rain is simply rain. Although different types of precipitation reflect more or less of the radar signal, the same empirical relationships are used to convert reflectivity to rainfall intensity without distinguishing between different types of rain events. And because these empirical relationships were determined using a single point measurement to infer the amount of rain falling onto a large area, they fail to account for the rainfall’s spatial variability. These two issues lead to errors that end up creeping into weather forecasts and natural disaster predictions. The upcoming generation of weather radars will be able to distinguish between different types of rain events and provide better data on raindrop size distributions, thus significantly reducing these two sources of error.

In a four-year research project, PhD student Joël Jaffrain studied the spatial variability of raindrop size distributions. He set up a network of disdrometers – sensors that can determine the number, size and speed of raindrops. Using the data, he produced a detailed picture of the rain intensity and drop sizes. “This is the very first experimental dataset that captures the variability within one-pixel of a radar image (1 square km),” says Berne.

The experimental campaign took place on the EPFL campus. Jaffrain showed that the uncertainty in rain amount derived from radar measurements can reach 15%. It might not sound like much, but when radar measurements are used to “nowcast” the potential of extreme weather events, such as floods and land-slides, improved accuracy both in the measured value and the associated uncertainty can save lives.

Although the research project has come to an end, the disdrometer network is being used in a number of other field experiments. The data obtained during the campaign will be made public in the hope that other research groups around the world will use them to further improve the performance of weather radars. In the meantime, next time it rains and you’re lucky enough to be carrying an umbrella, why not tune into the sound of the drops and imagine how different they all are, how unique, each one ending its trip with a signature flourish on the fabric above your head.