On the frontline of heavy rain prediction

Japan often experiences periods of extreme rainfall, and damage can be extensive.

But accurately predicting the formation and path of these bands of heavy rain clouds is difficult. They can bring continuous torrential rain in the same area and cause severe flooding and landslides.

Of the more than 20 bouts of heavy rainfall that occurred last year caused by such bands, Japan's Meteorological Agency failed to accurately predict 14 of them.

With the rainy season approaching in Japan, NHK is on the frontline with the agency in their efforts to warn people of likely deluges.

Ships monitor vapor amounts that could form bands of heavy rain clouds

Rising warm air with accumulated sea water vapor can generate a series of cumulonimbus clouds.

Sea water vapor is key to predicting the formation of bands, but continuous monitoring is difficult.

The agency uses an observation ship to measure vapor amounts to predict the occurrence of clouds. Based on weather forecasts, the ship heads to the area where a high influx of vapor is expected.

This observation ship left Tokyo in May and headed to the southwestern coast of Kyushu, where the amount of rain is high at this time of year.

One billionth of a second influences forecast accuracy

The ship forecasts the amount of vapor using antennas. A lot of vapor delays the radio waves that reach these antennas from an artificial satellite by about a billionth of a second.

To monitor vapor flows, a 1.5-meter balloon equipped with a sensor is launched from the ship. The sensor can continue to observe temperatures, humidity, wind direction and altitude up to about 30 kilometers above sea level.

Data sent from a sensor at a certain interval enables the Agency to gauge wind flow. This ascertains the time and destination of large amounts of vapor.

Supercomputer enables more detailed analysis

In the past, researchers calculated airflow and water vapor every 5 kilometers. But analysis by supercomputer makes it possible to determine the situation every 2 kilometers.

This supercomputer analyzes collected data.

In July last year, nine people died during a heavy rain storm in northern Kyushu, but the agency failed to accurately predict the occurrence and warn people in advance.

Subsequent analysis by the supercomputer, however, showed very similar results to the bands of heavy rain clouds that actually developed.

Two-km interval analysis over Kyushu in July last year accurately shows developed cumulonimbus rain cloud bands.

Take appropriate safety measures when heavy rain is predicted

The Meteorological Agency issues alerts in each prefecture about half a day before linear rainbands are expected to occur.

An agency official says that the new forecasts using ships and supercomputers will help them improve the accuracy and timeliness of their calls.

Northern Kyushu suffered flooding in July last year due to torrential rain.

If an alert is issued, people should check hazard maps to assess landslide and flooding risks in their areas, collect emergency items, and check evacuation sites.