#20 Linear Rainbands

Japan has a long history of natural disasters.

Now, the power of science is helping to overcome them.

BOSAI - Science that Can Save Your Life.

In recent years, the number of large-scale flood damage is increasing across Japan.

The 2018 Torrential Rain in Western Japan left more than 270 people dead or missing.

In 2020, the Kuma River in Kumamoto Prefecture overflowed, taking the lives of 65 people.

The cause of such flooding is a weather phenomenon known as a "linear rainband," in which rain clouds form in a linear pattern.

Linear rainbands cause heavy rainfall in the same location for an extended period of time.

In recent years, heavy rain caused by linear rainbands have been responsible for causing extensive damage in Japan year after year.

Now, what exactly is a linear rainband?

The term "senjo kosuitai," or linear rainband, has become more commonly used ever since torrential rain hit Hiroshima in 2014.

The Japan Meteorological Agency defines a linear rainband as follows: "A heavy rainfall area extending in a linear pattern 50-300 km long and 20-50 km wide."

The area with the most rainfall exceeds 150 mm in 3 hours, and has a high potential to cause disasters.

This weather phenomenon is produced by a series of developed cumulonimbus clouds that form successively, bringing heavy rainfall.

Cumulonimbus clouds are created by warm, moist air.

When it rains, a layer of cold air forms under the cumulonimbus cloud.

When warm, moist air collides with this cold air, cumulonimbus clouds are generated once again.

They emerge one after another and form a line, as if one cumulonimbus cloud is producing the next one.

This is what is known as a linear rainband.

Once a linear rainband occurs, intense rain continues to fall in the same area for many hours,

rapidly increasing the risk of deadly flooding.

Linear rainbands have been extremely difficult to predict, because researchers did not know when and where they would occur.

However, various studies are now underway in Japan.

How accurately can a linear rainband be predicted?

In this episode, we will take a look at the latest in the research on linear rainbands.

ITOU Wataru is a spokesperson for the Japan Meteorological Agency, that has been providing linear rainband forecasts since this year.

Why are linear rainbands hard to predict?

There’s still much we don’t know
about how a linear rainband forms.

The detailed mechanisms of linear rainbands, such as the timing of occurrence, and the build-up process are not yet understood.

The key to its prediction is water vapor, the source of cumulonimbus clouds.

Water vapor is invisible and it is extremely difficult to accurately measure the amount in the atmosphere, as it changes rapidly due to wind and temperature.

Now, the JMA is increasing its meteorological data by conducting water vapor observations over the ocean, which had not been done until now.

They are also analyzing airborne observation data taken with meteorological satellites in an effort to make predictions on linear rainbands.

How exactly do you make predictions?

The technique is called ensemble forecasting.

Ensemble forecasting involves generating multiple forecasts, instead of making a single forecast, for phenomena that are hard to predict such as the linear rainband.

If a number of similar results are obtained, it increases the probability of that particular forecast.

On the other hand, if the results are varied, the probability is low.

The prediction of linear rainbands is determined by its probability.

In July 2020, the JMA also predicted the linear rainband in southern Kyushu.

This ensemble forecast
predicted the linear rainband.

There are 21 patterns with varying results.

Results varied due to subtle uncertainties, such as wind and temperature changes.

Taking a closer look at one result, we can see linear rain clouds predicted to bring more than 100 mm of rain in 3 hours.

There were 8 patterns with similar forecasts and this lead the JMA to believe that linear rainbands were highly likely to occur.

The half-day forecast that began this year is based on the comprehensive judgement made using this ensemble forecast plus the JMA's knowledge from past experiences.

However, the hit rate of the forecast is about 1 in 4 times.

The forecast area is also broad, dividing Japan into 11 regions, including Kanto Koshin and northern Kyushu.

Currently, there are still issues with the broadness of the area and accuracy.

We want to provide linear rainband forecasts
half a day in advance.

Linear rainbands can occur at night
or before dawn.

Evacuation can be dangerous at night, and
nighttime announcements can be missed.

The current prediction system gives residents early warnings, but is not able to narrow down the location of where heavy rainfall would occur.

Now, research is underway to improve its accuracy.

Shingo SHIMIZU of the National Research Institute for Earth Science and Disaster Resilience is working to predict linear rainbands through the precise observation of water vapor.

We're setting up water vapor observation network
in Kyushu to predict linear rainbands.

Based on the idea that the altitude distribution of water vapor may deeply be involved in the development of linear rainbands, Shimizu and his colleagues have been using microwave radiometers that can estimate the altitude distribution of water vapor.

Water vapor monitoring had been insufficient.

So we observe low altitude water vapor
from the ground.

Water vapor monitors are positioned mainly in the western part of Kyushu.

They quickly capture water vapor coming from the ocean and allow researchers to monitor the direction and the amount of water vapor.

The vast amount of real-time data being observed in Kyushu is collected here at the National Research Institute for Earth Science and Disaster Resilience.

It needs to be updated within 10 minutes,
the shortest lifespan of a cumulonimbus cloud.

A linear rainband is a phenomenon in which cumulonimbus clouds form one after another, producing rainfall.

In other words, if the size of each cumulonimbus cloud is not accurately predicted, the rainfall amount would be underestimated, which would lead to overlooking the linear rainband.

In order to analyze the vast amount of data within 10 minutes, Shimizu has developed 2 computer systems.

The first is an observation system that accurately captures the conditions of the sky in real time.

The other is a numerical simulation system that uses observation results to make future predictions.

By analyzing the present and the future simultaneously, it allowed the system to be updated every 10 minutes,

and enabled researchers to make precise 2-hour predictions.

This is Shimizu's model created in 2020 that predicts a linear rainband.

On the right is the actual observation of the linear rainband.

The locations where the linear rainband occurred are almost identical.

The red and purple indicate areas where the 3-hour total rainfall exceeded 80 mm.

By closely observing water vapor and predicting each cumulonimbus cloud, Shimizu was able to predict the exact form of the linear rainband 2 hours in advance.

However...

The hit rate is about 30%.

Shimizu also needs to improve the hit rate on his forecast.

It's necessary to step up water vapor observations even more.

New experiments to improve forecasting accuracy have begun.

The latest observation instrument is the "water vapor lidar."

It was developed by the Japan Meteorological Agency for predicting linear rainbands.

Assistant professor Koichi Shiraishi of Fukuoka University conducts observations using water vapor lidars.

The water vapor lidar can accurately
measure the altitude distribution.

The water vapor lidar can accurately measure the amount of water vapor there is at what altitude.

It emits a powerful laser beam from the ground towards the sky, and measures the intensity and time of the light bouncing off water vapor molecules in the atmosphere.

Laser beams are emitted 10 times per second.

It operates unattended, 24 hours around the clock.

Studies show that the amount of moist air
and the altitude distribution are important.

This is data observed by the water vapor lidar.

The vertical axis is the altitude and the horizontal axis shows the time.

We can see there is a large amount of water vapor within the 1 kilometer altitude.

This is deeply connected to linear rainband formation.

In 2021, a linear rainband occurred in southern Kyushu.

Its precursor was captured by the water vapor lidar.

At that time, a linear rainband began to develop around midnight.

About 6 hours prior to that event, the amount of water vapor within the 1 kilometer altitude had increased rapidly.

What would happen if the observation data from the lidar was used on Shimizu's simulation system?

This is a simulation of the rain clouds at that time.

Cumulonimbus clouds are shown vertically.

The darker the color, the more rain is stored.

This image is assimilated with the water vapor lidar data.

The number of cumulonimbus clouds had increased, and the size of the rain clouds also grew.

They were able to accurately predict the amount of rain that fell 2 hours later.

The water vapor lidar had helped improve the accuracy of the linear rainband prediction.

If a linear rainband occurs at night,
it’s hard to evacuate.

I will continue my efforts to make
accurate half-day forecasts.

The accuracy for forecasting linear rainbands is improving steadily.

We may have a highly reliable prediction system in the near future.

However, once a linear rainband does occur, it is likely to cause record-breaking rainfall that leads to massive flooding.

Today, I was able to reconfirm the importance of always checking the latest weather news and being prepared to evacuate quickly in the event of an emergency.